Commit 0df730fb authored by Mahé Perrette's avatar Mahé Perrette
Browse files

update all drought country reports

parent 64ddef1b
This source diff could not be displayed because it is stored in LFS. You can view the blob instead.
This (link: glossary/climate-impact-assessment text: climate impact) assessment presents (link: glossary/projections text: projections) of the land area affected by droughts and the number of people exposed to droughts in Afghanistan, based on 1 (link: glossary/global-hydrological-models text: global hydrological model) (GHMs) and 3 different global (link: glossary/climate-models text: climate models)(GCMs) participating in [ISIMIP](https://www.isimip.org/).
### Areas affected by - people exposed to drought
Here, (link: glossary/droughts text: droughts) refer to a situation where the monthly (link: glossary/soil-moisture text: soil moisture) conditions fall short of the 2.5th percentile (glossary/link: variable-monthly-threshold text: variable monthly threshold) for at least 7 consecutive months. To calculate the total land area affected by drought per country the land area exposed to drought is summed up across all grid cells belonging to Afghanistan. Multiplying, at each grid cell, with the number of people living in the grid cell yields the number of people exposed to drought.
Authors: Stefan Lange<sup>1</sup>, Ted Veldkamp<sup>2</sup>, Matthias Mengel<sup>1</sup>, Hannes Müller Schmied<sup>3</sup>, Katja Frieler<sup>1</sup>
We analyze the land area affected by droughts and the number of people exposed to droughts at different (link: glossary/levels-of-global-warming text: levels of global warming)(0, 1, 2, and 3 degrees Celsius) and during different time periods. We cover the time before (link: glossary/anthropogenic-climate-change text: anthropogenic climate change) started (conventionally set at 1850), present-day (2001-2020), mid-century (2041-2060) and end-of-century (2081-2100). We compare the later periods against the climate conditions before anthropogenic climate change started (1850). A ranking of countries for each indicator is provided to allow comparison between countries on the basis of their (link: glossary/relative-change text: relative change) in the land area affected or the number of people exposed under different levels of global warming and for various time horizons, compared to conditions before anthropogenic climate change started. Here, a ranking of 1 implies that the country experiences a change in the (link: glossary/indicator text: indicator) that is more than all other ranked countries, whereas a lower ranking entails comparatively less effects due to climate change.
It is important to note that our (link: glossary/future-projections text: future projections) assume that the number of people living in an area, as well as the area’s (link: glossary/land-use text: land use) and (link: glossary/land-cover text: land cover), remain constant to that of the year 2005. This is not necessarily meant to be realistic, but to isolate the influence of climate change from the influence of other changes.
Affilitations:
1. Potsdam Institute for Climate Impact Research, Germany
2. Vrije Universiteit Amsterdam, The Netherlands
3. Goethe University Frankfurt, Germany
Published: [to be added]
Doi: [to be added]
In the key messages we report the results for the (link: glossary/median text: median) of the (link: glossary/model-ensemble text: model ensemble) of global hydrological models providing data to ISIMIP. The median represents the middle of the ensemble, meaning that 50% of the ensemble members provide higher numbers and 50% provide lower numbers. After the key message, which only references the median, you can find the model-specific results in the figures, a discussion of the methodology and the restrictions of the analysis at the end of the article.
### Key messages
* Without human-made greenhouse gas emissions, 1.398% of Afghanistan’s land area and 0.967% of Afghanistan’s population would be affected by droughts each year, on average.
* At today’s levels of 1 degrees Celsius of global warming the land area affected is 3.097 times as much: 209.721% of the total land area, while the number of people affected is 8.32 times as much: 2.932% of the total population.
* At 2 degrees Celsius of global warming, Afghanistan’s land area affected by droughts is projected to increase by a factor of 904.194 compared to a world without human-made greenhouse gas emissions, to 1.398%. Likewise, Afghanistan’s population exposed to droughts is projected to increase by a factor 1473.618, to 0.967%.
* Following the higher-emissions scenario (RCP6.0) which can entail over 3 degrees Celsius of global warming by the end of the century (2081-2100), this factor is projected to reach 1.234 for the land area affected to droughts (to 1.234%) and 1.158 for the population exposed to droughts (to 15.772%).
* Afghanistan ranks (ranking-value: land-rel-temp_AFG value: position temperature:2) with regards to its relative change in land area affected by droughts at 2 degrees Celsius of global warming in comparison to a situation without anthropogenic climate change. For the relative change in population exposed to droughts, Afghanistan’s ranking is (ranking-value: pop-rel-temp_AFG value: position temperature:2).
### Results
The figures below shows the relative change in Afghanistan’s land area affected by and population exposed to droughts in comparison to the time period before anthropogenic climate change. Results are shown both for the (link: glossary/historical-period text: historical time period) and for the (link: glossary/future-projections text: future projections) and can be visualized with regards to their change over time as well as with regards to their change in terms of global warming. Under the ‘change in terms of time’ setting, the change in land area affected by and population exposed to droughts over time is shown for 20-year (link: glossary/time-slices text: time-slices) from 1861-1880 until the end-of-the-century (2081-2100). Under the ‘change in terms of global warming’, the change in land area affected by and population exposed to droughts is shown for 20-year (link: glossary/time-slices text time-slices) representing increasing levels of global warming from 0 degrees Celsius up to 3 degrees Celsius compared to the time period before anthropogenic climate change started.
* Afghanistan ranks (ranking-value: land-abs-temp_AFG value: position temperature:2) with regards to absolute changes in land area affected by droughts (expressed as % of Afghanistan’s land area) at 2°C of global warming in comparison to a situation without climate change. For the absolute changes in population exposed to droughts (expressed as % of Afghanistan’s population), Afghanistan ranks (ranking-value: pop-abs-temp_AFG value: position temperature:2).
* At today’s levels of 1°C of global warming the simulated land area affected is already 12200 km<sup>2</sup> larger (2.0% of the national land area) than in a world without climate change where the annual area affected by droughts is 6000 km<sup>2</sup> (0.9% of Afghanistan’s land area). The number of people exposed is 0.4 million (1.1% of the national population) larger than without climate change where the annual number of people exposed to droughts was 0.2 million (0.7% of Afghanistan’s population).
* At 2°C of global warming the land area affected by droughts would increase by 24000 km<sup>2</sup> (3.7% of the national land area) compared to a world without climate change, to 4.4% of the country’s land area. Assuming present-day population patterns, Afghanistan’s population exposed to droughts would increase by 1.7 million, to 2.1% of the national population.
* Following the higher-emissions scenario (RCP6.0) which can entail over 3°C of global warming by the end of the century (2081-2100) ([Frieler et al. 2017](https://dx.doi.org/10.5194/gmd-10-4321-2017)) the land area affected by droughts would increase by 80500 km<sup>2</sup> (12.3% of the national land area) and reach 13.2% of the country’s land area. Assuming present-day population patterns the population exposed would reach 11.2% of Afghanistan’s population, and increase by 11.0 million compared to a situation without climate change.
### How have we got the results?
The assessment provided below is based on our paper “Change in exposure to climate impact events under global warming” submitted to the journal “Earth’s Future”. We have used simulations of 8 different (link: glossary/global-hydrological-models text: global hydrological models) to estimate changes in drought conditions assuming different levels of global mean temperature change. To this end the global hydrological models were forced by climate projections from 3 different global (link: glossary/climate-models text: climate models) following the ISIMIP2b scenario set-up. In the key messages we report the results for the median of the (link: glossary/model-ensemble text: model ensemble) of global hydrological models that ran simulations in (link: glossary/isimip2b text: ISIMIP2b). The median represents the middle of the ensemble, meaning that 50% of the ensemble members provide higher numbers and 50% provide lower numbers.
In both cases, results are shown for all possible combinations of (link: glossary/global-hydrological-models text: global hydrological models) and (link: glossary/global-climate-models text: global climate models) (normal lines) as well as for the median of the model ensemble (thick line). The influence of (link: glossary/inter-annual-variability text: inter-annual variability) on the results is visualized by means of the shaded area in black. This inter-annual variability is only shown for the median model combination. For the future projections and under the ‘change in terms of time’ settings only, an additional distinction is made between the results of two future (link: glossary/emissions-trajectories text: emissions trajectories): a low-emissions trajectory limiting global mean temperature in the 21st century to below 2 degrees Celsius compared to before anthropogenic climate change started, called RCP2.6 (blue lines), and a higher-emissions scenario leading global mean temperature to above 3 degrees Celsius by 2100, compared to pre-climate change conditions, called RCP6.0 (red lines). The shaded areas in blue and red indicate the variety (or spread) in results between all combinations of global hydrological models and global climate models for these two emissions trajectories. A filtering menu top-right of the graph allows the user to select individual emissions trajectories, individual global climate models, and individual global hydrological models for visualization.
Among the multiple concepts of “drought”, our study refers to a situation where monthly soil moisture (agricultural drought concept) is extremely low for at least 7 consecutive months. “Extremely low” means dry conditions occurring less than 5 times in 200 years under reference conditions without climate change. We were interested in changes in the land area affected by droughts and the number of people exposed to droughts that we have to expect at different (link: glossary/levels-of-global-warming text: levels of global warming) (1°C, 2°C, and 3°C) or during different future time periods.
By hovering over or clicking on a particular value in the figure additional details, such as the specific global climate model or hydrological used, behind the presented value become available. The visualization below the graph shows, for the selected model run under this time-period or temperature-change level, how Afghanistan ranks in comparison to other countries on its relative change in land area affected by or population exposed to droughts.
#### Land area affected by droughts
The ISIMIP2b simulations start in 1860 (before human greenhouse gas emissions started to change the climate) and end in 2100. They comprise a long run where weather only varies according to pre-industrial conditions without climate change. In addition, modellers have run two different future scenarios: one “business as usual” scenario (RCP6.0) reaching high levels of warming and one “low emissions scenario” (RCP2.6) with reduced levels of climate change.
As the “land area affected by drought” varies quite strongly from year to year we decided to average numbers over multiple years: for each year we calculate the deviation of the land area affected by droughts from the average reference level without climate change. These deviations are then averaged over a present-day period (2001-2020), a mid-century period (2041-2060) and an end-of-century period (2081-2100). To calculate the changes at different levels of global mean temperature change we do not average over years belonging to one of these periods but over all years with global mean warming levels close to 1°C, 2°C, and 3°C.
When calculating the “number of people exposed to droughts” we proceed in a similar way. We assume that only the rural population is exposed while the urban population does not necessarily “feel” any drought. For 1860 to 2005 we account for changes in population patterns ([Goldewijk et al., 2017](https://dx.doi.org/10.5194/essd-9-927-2017)). Afterwards the population data is considered constant -- not as a realistic assumption but an informative “thought experiment”: What would future climate change mean for present day societies? For all time periods, the pure effect of climate change on the “number of people exposed to droughts” is estimated by comparing the number of people exposed under climate change and a specific population pattern to the number of people exposed assuming the same population patterns but no climate change.
In the following we describe what we have found and how strongly Afghanistan is affected compared to other countries.
### What we have found
### Land area affected by droughts
(line-plot: land-abs-temp_AFG,land-abs-time_AFG first-temperature: 2 second-scenario: rcp26 second-time: 2041-2060)
Without human-made greenhouse gas emissions, 1.398% of Afghanistan’s land area would be affected by droughts each year, on average.
Our definition of “drought” is quite strict, such that, without climate change, only 0.9% of Afghanistan’s land area would be affected by droughts each year, on average.
However, at today’s level of 1°C global warming Afghanistan’s annual land area affected by droughts is, on average, already larger and amount to 12200 km<sup>2</sup> (1.9% of the national land area). The level of change ranges from 0.3% to 13.3% for the individual combinations of global hydrological models and global climate models. At 2°C of global warming, Afghanistan’s annual land area affected by droughts is projected to increase by 24000 km<sup>2</sup> (i.e. 3.7% of the national land area) on average in comparison to a world without climate change. Under these conditions, 4.4% of the national land area would be affected by droughts each year, on average. Across the individual combinations of global hydrological models and global climate models this expected level of change ranges from 1900 to 86500 km<sup>2</sup>.
Following the higher-emissions scenario (RCP6.0) the land area affected by droughts is expected to increase by 64200 km<sup>2</sup> (9.8% of the national land area) towards the end of the century (2081-2100). Following the low emission scenario (RCP2.6) the change would only reach 16800 km<sup>2</sup> (9.8% of the national land area). By the middle of the century, changes reach 31600 km<sup>2</sup> under RCP2.6 and 26100 km<sup>2</sup> under RCP6.0.
Afghanistan is the (ranking-value: land-abs-temp_AFG value: position temperature:2) strongest affected by droughts at 2°C of global warming. For the absolute change in land area affected by droughts towards the end of the century under a high-emissions scenario (RCP6.0), Afghanistan’s ranking is (ranking-value: land-abs-time_AFG value: position time:2081-2100 scenario:rcp60).
### Population exposed to droughts
(line-plot: pop-abs-temp_AFG,pop-abs-time_AFG first-temperature: 2 second-scenario: rcp26 second-time: 2041-2060)
Our definition of “drought” is quite strict, such that, without climate change, only 0.0% of Afghanistan’s population would be exposed to droughts each year, on average.
However, at today’s level of 1°C of global warming Afghanistan’s annual population exposed to droughts is, on average, already 0.4 million (i.e. 1.1% of the total population) higher than without climate change and amount to 1.5% of the total population. The level of change ranges from 0.3 % to 13.3% for the individual combinations of global hydrological models and global climate models. At 2°C of global warming, Afghanistan’s annual population exposed to droughts is projected to increase by 0 million (i.e. 1.7% of the population) on average in comparison to a world without climate change. Under these conditions, 2.1% of the total population would be affected by droughts, on average. Across the individual combinations of global hydrological models and global climate models this expected level of change ranges from 0.1 up to 3.7 million people.
Following the higher-emissions scenario (RCP6.0) the population exposed to droughts is expected to increase by 1.6 million (5.0% of the total population) towards the end of the century (2081-2100). Following the low emission scenario (RCP2.6) the change would expose 0.2 million people (0.7% of the total population). By mid of the century changes expose 0.4 million people under RCP2.6 and 0.5 million people under RCP6.0.
Afghanistan is the (ranking-value: pop-abs-temp_AFG value: position temperature:2) strongest affected by droughts at 2°C of global warming. For the absolute change in population exposed to droughts towards the end of the century under a higher-emissions scenario (RCP6.0), Afghanistan’s ranking is (ranking-value: pop-abs-time_AFG value: position time:2081-2100 scenario:rcp60).
### How is soil moisture calculated?
Soil moisture is defined as the water stored in the soil in liquid or frozen form. It is calculated as the difference between the amount of water coming in as rain or snow, and the amount of water going out, either through evapotranspiration from the surface and vegetation, runoff, or through percolation of water towards the deeper groundwater layers. Global hydrological models (GHMs) include different representations of the (link: glossary/soil-water-column text: soil water column) with different numbers of soil water layer(s) (between 1 and 15 layers), and with different total depth of soil layers (between 1 and 42.1 meter).
Here, we use (link: glossary/rootzone-soil-moisture text: rootzone soil moisture) estimates, the portion of soil moisture that is within the rooting depth of plants, when directly provided by the GHMs. For GHMs that do not directly provide root zone soil moisture conditions, we approximated this variable by integrating soil moisture across multiple soil water layers in order to reach a depth of ~1 meter. Daily rootzone soil moisture values are finally aggregated into monthly average rootzone soil moisture conditions per grid cell.
### Where are the most relevant gaps in our knowledge?
Models are simplified representations of reality, hence model simulations come with limitations and uncertainties that have to be kept in mind ([Döll et al. 2016](https://dx.doi.org/10.1007/s10712-015-9343-1)). Regarding our drought simulations we mainly see the following issues.
* Representation of direct human influences
Under the ‘change in terms of global warming’ setting the figure shows that at today’s level of 1 degrees Celsius of global warming Afghanistan’s land area affected by droughts is, on average, already 3.097 times (or 209.721%) lower and amount to 4.563% of the total land area. This level of change ranges from 5.585% to 13.256% for the individual combinations of global hydrological models and global climate models. At 2 degrees Celsius of global warming, Afghanistan’s land area affected by droughts is projected to change by a factor of 10.042 (or 11.346%) in comparison to a world without human-made greenhouse gas emissions. Under these conditions, 11.63% of the total land area would be affected by droughts on a yearly basis, on average. Across the individual combinations of global hydrological models and global climate models this expected level of change ranges from 5.585% up to 13.256%.
When presenting the ‘change in terms of time’, we find that when following the higher-emissions scenario (RCP6.0) towards the end-of-the-century (equivalent to over 3 degrees Celsius change) would result in a change in the land area affected by droughts of, on average, a factor 33.454 (22.705%), towards: 23.405% of the total land area. Following the (link: glossary/climate-mitigation-emissions-scenario text: climate-mitigation emissions scenario)(RCP2.6) towards the end-of-the-century (entailing an average 2.5 degrees Celsius change) would result in a foreseen change in the land area affected by droughts of, on average, a factor 7.198 (5.41%), 6.283% of the total land area being affected.
The presented drought projections reflect the isolated effect of climate change, while their local manifestations are expected to also be influenced by direct human drivers like changes in land use, land cover, and irrigation. Even in the historical period our simulations only account for water abstractions (with varying sectors among the models like irrigation, industry, domestic, livestock, desalinization) from groundwater and surface waters, reservoir management while other processes such as inter-basin water transfers and a more realistic reservoir management are also expected to be important. Including these drivers into our simulation would greatly improve the simulations ([Veldkamp et al. 2018](https://doi.org/10.1088/1748-9326/aab96f)). That it is not done is mainly an issue of missing knowledge about them.
Globally, the land area affected by droughts each year, on average, is 2.377% in a situation without human-made greenhouse gas emissions. At 2 degrees Celsius global warming or by the end-of-the-century under a higher-emissions scenario (RCP6.0) these values are foreseen to increase globally by a factor 3.106 and 7.014%, respectively.
As such, Afghanistan ranks (ranking-value: land-rel-temp_AFG value: position temperature:2) with regards to its relative change in land area affected by droughts at 2 degrees Celsius of global warming in comparison to a situation without anthropogenic climate change. For the relative change in land area affected by droughts towards the-end-of-the-century under a higher-emissions scenario (RCP6.0), Afghanistan’s ranking is (ranking-value: land-rel-time_AFG value: position time:2081-2100 scenario: rcp60).
* Representation of the CO2 fertilization effect
#### Population exposed to droughts
(line-plot: pop-temp_AFG,pop-time_AFG first-temperature: 2 second-scenario: rcp26 second-time: 2041-2060)
Another challenge lies in the representation of how vegetation cover affects evapotranspiration, a relationship that is subject to change under rising temperature and CO2 levels. For example, the effect of CO2 fertilization – the phenomenon through which photosynthesis, hence plant growth, should be enhanced in a CO2-richer atmosphere – is not represented in most global hydrological models as the vegetation components in those models are often simplified. Plants need less water to assimilate the same amount of carbon in a CO2-richer atmosphere. This effect alone would lead to a decrease of evapotranspiration. Yet the more efficient carbon assimilation lets plants grow better, which leads to bigger or more plants. This effect alone would lead to an increase of evapotranspiration. The overall CO2 fertilization effect on evapotranspiration, soil moisture conditions and drought risk is uncertain (Gerten et al., 2014; [Prudhomme et al., 2014](https://dx.doi.org/10.1073/pnas.1222473110); [Döll et al., 2016](https://dx.doi.org/10.1007/s10712-015-9343-1); [Kuzyakov et al., 2019](https://dx.doi.org/10.1016/j.soilbio.2018.10.005)).
Without human-made greenhouse gas emissions, 0.967% of Afghanistan’s population would be exposed to droughts each year, on average.
Under the ‘change in terms of global warming’ setting the figure shows that at today’s level of 1 degrees Celsius of global warming Afghanistan’s population exposed to droughts is, on average, already 8.32 times (or 731.995%) lower: 2.932% of the total land area. This level of change ranges from 1.567% to 3.437% for the individual combinations of global hydrological models and global climate models. At 2 degrees Celsius of global warming, Afghanistan’s population exposed to droughts is projected to change by a factor of 15.736 (or 7.317%) in comparison to a world without human-made greenhouse gas emissions. Under these conditions, 7.508% of the total population would be exposed to droughts on a yearly basis, on average. Across the individual combinations of global hydrological models and global climate models this expected level of change ranges from 4.072% up to 11.832%.
* Representation of the soil compartment
When presenting the ‘change in terms of time’, we find that when following the higher-emissions scenario (RCP6.0) towards the end-of-the-century (equivalent to over 3 degrees Celsius change) would result in a foreseen change in the population exposed to droughts of, on average, a factor 57.518 (15.512%),15.772% of the total population. Following the climate-mitigation emissions scenario (RCP2.6) towards the end-of-the-century (entailing an average 2.5 degrees Celsius change) would result in a foreseen change in the population exposed to droughts of, on average, a factor 6.503 (2.349%), 2.776% of the total population being exposed.
The representation of the soil compartment vary with the total depth of soil layers (between 1 and 42.1 m), the number of soil layers (ranging from 1 up to 15) and the conceptualization of soil water processes. In many global hydrological models, we use here root zone soil moisture (related to agricultural needs) as an input for our drought assessment. However, in other models, we have to approximate this variable by integrating soil moisture across multiple water layers. Differences in the representation of the soil column and soil water processes can lead to significant differences in estimates of soil water availability, soil water saturation and drought conditions but this effect could not be investigated for the models used in this assessment.
Globally, the population exposed to droughts each year, on average, is 0.601% in a situation without human-made greenhouse gas emissions. At 2 degrees of global warming or by the end-of-the-century under a higher-emissions scenario (RCP6.0) these values are foreseen to increase globally by 3.264% and 10.391% respectively.
As such, Afghanistan ranks (ranking-value: pop-rel-temp_AFG value: position temperature:2) with regards to its relative change in population exposed to droughts at 2 degrees Celsius of global warming in comparison to a situation without anthropogenic climate change. For the relative change in population exposed to droughts towards the-end-of-the-century under a business-as-usual emissions scenario, Afghanistan’s ranking is (ranking-value: pop-rel-time_AFG value: position time:2081-2100 scenario: rcp60).
* Representation of evapotranspiration in general
### Methodology
Much of the differences between the individual hydrological model simulations considered here is assumed to be due to different calculation of potential and actual evapotranspiration. Whereas some models directly calculates actual evapotranspiration (AET, the amount of water that is transferred through plants, bare soil, open water bodies or from water stored at canopies) directly using water transfer schemes or turbulent fluxes, the majority of the models use the potential evapotranspiration (PET) concept. By using various approaches with different demand on meteorological input variables, the potential water demand of the atmosphere (PET as upper limit) is calculated. The availability of water in the storages (e.g. soil water storage) can reduce the amount of water that is actually evapotranspired to the atmosphere (AET). Unfortunately, currently, we are still not able to decide which of the representations is most appropriate but first assessments of those representations have been done ([Wartenburger et al. 2018](https://doi.org/10.1088/1748-9326/aac4bb)). As a result, estimates of actual evapotranspiration can differ significantly across models despite having used uniform inputs from the global climate models, which may affect estimates of drought conditions much more than for example flooding which mainly depends on the amount of precipitation.
#### What do we analyze?
We analyze the land area affected by droughts and the number of people exposed to droughts at different levels of global warming (0, 1, 2, and 3 degrees Celsius) and during different time periods. We cover the time before anthropogenic climate change started (1850), present-day (2001-2020), mid-century (2041-2060) and end-of-century (2081-2100). We compare the later periods against the climate conditions before anthropogenic climate change (1850).
#### How do we calculate where and when a drought occurs?
The calculation has several steps: 1) calculate (link: glossary/soil-moisture text: soil moisture); 2) calculate the 2.5th percentile (glossary/link: variable-monthly-threshold text: variable monthly threshold) value for monthly soil moisture conditions; and 3) evaluate where and when a grid cell is exposed to drought conditions for at least 7 consecutive months. All these steps are detailed below, after a short explanation of the spatial structure on which our models operate.
### Disclaimer
#### What is the spatial structure of the models?
The models we use cover the whole globe’s land area. The land area is divided into a grid. Each grid cell has a size of 0.5 degrees × 0.5 degrees (latitude by longitude). At the equator, this equals roughly 55 × 55 km; towards the North Pole or South Pole, where the land area covered per grid cell becomes smaller.
Note, that although this report is based on an article published in a peer-reviewed scientific journal, the original publication focused on a global analysis. Therefore, the results presented here for 200+ countries have not been reviewed individually. ISIpedia provides data on country level for convenience, but cannot be held responsible for any issues with the data. Please contact the ISIpedia editorial team (isipedia.editorial.team@pik-potsdam.de) for more information or questions about this report.
#### How is soil moisture calculated?
The calculation is done for each grid cell and each day. Information about temperature, precipitation, solar radiation, and other weather indicators is taken from the global climate models (GCMs) and used as input for the ISIMIP global hydrological models (GHMs). Additional spatial data, such as soil, land cover and water bodies are also inputted. Soil moisture, i.e. the water stored in the soil in liquid or frozen form, is calculated as the difference between the amount of water coming in as rain or snow, and the amount of water going out, either through evapotranspiration from the surface and vegetation, or through percolation of water towards the deeper ground water layers. Global hydrological models include a different representation of the (link: glossary/soil-water-column text: soil water column) with different numbers of soil compartments with varying depths of the soil water layer(s), ranging from one soil water layer with a depth of between 0.1 - 4 meter up to 15 fully resolved soil water layers with a total depth up to 45 meter.
### References
Döll, P., Douville, H., Güntner, A., Müller Schmied, H., Wada, Y. (2016) Modelling Freshwater Resources at the Global Scale: Challenges and Prospects. Surveys in Geophysics, 37(2), 195–221. https://dx.doi.org/10.1007/s10712-015-9343-1
Here, we use (link: glossary/rootzone-soil-moisture text: rootzone soil moisture) estimates, the portion of soil moisture that is within the rooting depth of plants, when directly provided by the global hydrological models. For global hydrological models that do not directly provide rootzone soil moisture conditions, we approximated this variable by integrating soil moisture across multiple soil water layers in order to reach a depth of ~1 meter.
Daily rootzone soil moisture values were finally aggregated into monthly average rootzone soil moisture conditions per grid cell.
Frieler, K. and Lange, S. and Piontek, F. and Reyer, C. P. O. and Schewe, J. and Warszawski, L. and Zhao, F. and Chini, L. and Denvil, S. and Emanuel, K. and Geiger, T. and Halladay, K. and Hurtt, G. and Mengel, M. and Murakami, D. and Ostberg, S. and Popp, A. and Riva, R. and Stevanovic, M. and Suzuki, T. and Volkholz, J. and Burke, E. and Ciais, P. and Ebi, K. and Eddy, T. D. and Elliott, J. and Galbraith, E. and Gosling, S. N. and Hattermann, F. and Hickler, T. and Hinkel, J. and Hof, C. and Huber, V. and J\"agermeyr, J. and Krysanova, V. and Marc\'e, R. and M\"uller Schmied, H. and Mouratiadou, I. and Pierson, D. and Tittensor, D. P. and Vautard, R. and van Vliet, M. and Biber, M. F. and Betts, R. A. and Bodirsky, B. L. and Deryng, D. and Frolking, S. and Jones, C. D. and Lotze, H. K. and Lotze-Campen, H. and Sahajpal, R. and Thonicke, K. and Tian, H. and Yamagata, Y. (2017). Assessing the impacts of 1.5 °C global warming -- simulation protocol of the Inter-Sectoral Impact Model Intercomparison Project (ISIMIP2b). Geoscientific Model Development,10 (12), 4321--4345. https://dx.doi.org/10.5194/gmd-10-4321-2017
#### How is the 2.5th percentile variable monthly threshold calculated?
To distinguish between periods of drought and periods of no drought we applied a (link: glossary/variable-monthly-threshold text: variable monthly threshold) level method over the monthly rootzone soil moisture conditions, using a 2.5th percentile value. Using the pre-industrial scenario as a reference period, we identified at the grid-level, per global hydrological model and global climate model, and for each month individually, the monthly rootzone soil moisture value that represents the 2.5th percentile value. This can be interpret as the monthly rootzone soil moisture condition that is exceeded 39 out of 40 times throughout the full time-period under study. The use of a variable monthly threshold to distinguish drought from periods of no drought allows accounting for seasonal climatology, which is relevant for the management of water resources.
#### How do we know whether a drought occurs?
With these obtained 2.5th percentile variable monthly threshold values, we then identify for the different climate change scenarios how often monthly rootzone soil moisture conditions fall under this threshold. To identify prolonged drought events, and omit incidental occurrences of droughts lasting only a short period of time, we applied a six-month threshold. Only droughts that last longer than six-months were taken into account in the analysis.
Gerten, D. and Betts, R. and Döll, P. (2014). Cross-chapter box on the active role of vegetation in altering water flows under climate change. In: Field CB, Barros VR, Dokken DJ, Mach KJ, Mastrandrea MD, Bilir TE, Chatterjee M, Ebi KL, Estrada YO, Genova RC, Girma B, Kissel ES, Levy AN, MacCracken S, Mastrandrea PR, White LL (eds) Climate Change 2014: impacts, adaptation, and vulnerability. Part A: global and sectoral aspects. Contribution of working group II to the 5th assessment report of the intergovernmental panel on climate change. Cambridge University Press, Cambridge, pp 157–161
#### How are the indicators land area affected by droughts and number of people exposed to droughts calculated?
To estimate the land area affected by and population exposed to drought we checked for each year whether a grid cell was exposed for at least one period of prolonged drought. If so, the grid cell is accounted as ‘being exposed to drought’ for that respective year. We then calculate the total land area affected by drought for Afghanistan by adding up all 0.5 degree x 0.5 degree grid cells belonging to Afghanistan. To calculate the number of people exposed to droughts, we first multiply the each grid cell by the number of people living in that grid cell to estimate the number of people exposed to droughts, and then add up those numbers across all grid cells belonging to Afghanistan.
#### What else should I know about the methodology?
We assume that the number of people living in an area, as well as the area’s land use and land cover (what fractions of the area are used for settlement, cropland, or pasture, or covered by forest), remain constant at the levels of year 2005 throughout the whole simulation. This is not meant to be realistic, but to isolate the influence of climate change from the influence of other changes.
Goldewijk, K., Beusen, A., Doelman, J., Stehfest, E. (2017) Anthropogenic land use estimates for the Holocene – HYDE 3.2. Earth Syst. Sci. Data, 9, 927–953. https://dx.doi.org/10.5194/essd-9-927-2017
In the key messages we report the results for the median (the model in the middle) of the group of global hydrological models that ran simulations in (link: glossary/isimip2b text: ISIMIP2b). All model-specific results are presented in the figures and a discussion of the limitations of the analysis is included at the end of the report.
A ranking of countries for each indicator is also provided to compare countries on the basis of their relative change in land area affected or number of people exposed under different levels of global warming and for various time horizons, compared to conditions without climate change. A ranking of 1 implies that the country experiences a change in the indicator higher than all other ranked countries, for example the strongest impacts from climate change; whereas a lower ranking entails comparatively less impacts.
Kuzyakov, Y., Horwath, W. R., Dorodnikov, M. and Blagodatskaya, E. (2019). Review and synthesis of the effects of elevated atmospheric CO2 on soil processes: No changes in pools, but increased fluxes and accelerated cycles, Soil Biology and Biochemistry, 128, 66-78, https://dx.doi.org/10.1016/j.soilbio.2018.10.005
### Discussion
The model simulations used for this report build on science that has been established through many peer-reviewed studies: for example by [Sheffield and Wood (2008)](https://doi.org/10.1007/s00382-007-0340-z), [Trenberth, et al. (2014)](https://www.nature.com/articles/nclimate2067), [Berg, et al. (2017)](https://doi.org/10.1002/2016GL071921), and [Berg and Sheffield (2018)](https://doi.org/10.1007/s40641-018-0095-0).
Models are simplified representations of reality, hence model simulations come with limitations and uncertainties that have to be kept in mind. Using multiple global hydrological models with different representations and parameterizations of the hydrological components and processes may introduce biases in the estimation of soil moisture conditions and in the resulting assessment of droughts ([Veldkamp et al., 2018](https://doi.org/10.1088/1748-9326/aab96f)). Examples are the representation of evapotranspiration within global hydrological models and the representation of soil water column.
Lange et al. Change in exposure to climate impact events under global warming. Submitted to Earth’s Future.
Different global hydrological models apply different methods to estimate (link: glossary/potential-evapotranspiraton text: (potential) evapotranspiration). These vary from the (link: glossary/bulk-formula text: Bulk formula) and the (link: glossary/monin-obukhov-similarity-theory text: Monin-Obukhov similarity theory) to (link: glossary/penman-montheith text: Penman-Montheith), (link: glossary/hamon text: Hamon), and (link: glossary/priestley-taylor text: Priestley-Taylor). In result, estimates of (potential) evapotranspiration can differ significantly across models despite having used uniform inputs from the global climate models. This may affect estimates of soil moistures conditions. Another challenge lies in the representation of how vegetation cover affects evapotranspiration, a relationship that is subject to change within a given river basin, or with rising temperature and CO2 levels. For example, the effect of CO2 fertilization – the phenomenon through which photosynthesis, hence plant growth, should be enhanced in a CO2-richer atmosphere – is not represented in most global hydrological models although it would affect the plants’ water use, with significant consequences for soil moisture conditions and drought estimates.
Using multiple global hydrological models with different representations of the soil water column introduces additional biases in the estimation of soil moisture conditions and droughts. The global hydrological models applied here include a different numbers of soil compartments with varying depths of their soil water layer(s). These representations range from one soil water layer with a depth of between 0.1 - 4 meter up to 15 fully resolved soil water layers with a total depth up to 45 meter. When directly provided by the global hydrological models, we use here root zone soil moisture as an input for our drought assessment. For global hydrological models that did not provide root zone soil moisture directly, we approximated this variable by integrating soil moisture across multiple soil water layers in order to reach a depth of ~1 meter. Differences in the representation of the soil column and soil water processes can lead to significant differences in estimates of soil water availability, soil water saturation and drought conditions.
Prudhomme, C.; Giuntoli, I.; Robinson, E. L.; Clark, D. B.; Arnell, N. W.; Dankers, R.; Fekete, B. M.; Franssen, W.; Gerten, D.; Gosling, S. N.; Hagemann, S.; Hannah, D. M.; Kim, H.; Masaki, Y.; Satoh, Y.; Stacke, T.; Wada, Y. & Wisser, D. (2014). Hydrological droughts in the 21st century, hotspots and uncertainties from a global multimodel ensemble experiment, Proceedings of the National Academy of Sciences, National Academy of Sciences, 111, 3262-3267, https://dx.doi.org/10.1073/pnas.1222473110
The performance of the global hydrological models can be tested by using (link: glossary/observed-historical-weather-information text: observed historical weather information) as inputs, and comparing the simulated soil moisture to soil moisture observations, locally measured or remotely sensed.
Apparent conflicting results of how droughts develop due climate change highlight the sensitivity of drought estimates to the choice of indicator and hydrological variable used ([Trenberth, et al. (2014)](https://www.nature.com/articles/nclimate2067), [Berg, et al. (2017)](https://doi.org/10.1002/2016GL071921)). Here, we use a variable monthly percentile threshold in combination with root zone soil moisture conditions to distinguish drought conditions from no drought conditions. Alternative indicators representing soil moisture droughts, such as the (link: glossary/palmer-drought-severity-index text: Palmer Drought Severity Index (PDSI)), the (link: glossary/standardized-soil-moisture-index text: Standardized Soil Moisture Index (SSMI)), or indicators representing a different part of the (link: glossary/hydrological-cycle text: hydrological cycle) (for example the (link: glossary/standardized-precipitation-index text: standardized precipitation index (SPI)), the (link: glossary/standardized-precipitation-and-evapotranspiration-index text: standardized precipitation and evapotranspiration index (SPEI)), or the (link: glossary/standardized-runoff-index text: standardized runoff index (SRI))) may result in different conclusions on the absolute and relative changes in land area affected by and population exposed to drought due to climate change.
Veldkamp, T. I.E., Zhao, F., Ward, P.J., de Moel, H., Aerts, J.C.J.H., Müller Schmied, H., Portmann, F.T., Masaki, Y., Pokhrel, Y., Liu, X., Satoh, Y., Gerten, D. Gosling, S.N., Zaherpour, J., Wada, Y. (2018) Human impact parameterizations in global hydrological models improve estimates of monthly discharges and hydrological extremes: a multimodel validation study. Environ. Res. Lett. 13 055008, https://doi.org/10.1088/1748-9326/aab96f
The presented drought projections reflect the isolated effect of climate change, while their local manifestations are expected to also be influenced by direct human drivers like changes in land use, land cover, and irrigation. The representation of these drivers is limited to (link: glossary/present-day-conditions text: present-day conditions), while their future changes are not explicitly modelled in the presented simulations. This serves the purpose of isolating the effect of anthropogenic climate change from other factors, but it means that actual drought exposure in the future could be more or less acute, depending on non-climatic human drivers, such as irrigation or adaptation.
Regarding climate models and global hydrological models, it is known that generally both types of models contribute substantially to the overall spread in projected climate change impacts on water-related variables. Because this report presents the results of a combination of climate models and global hydrological models that ran simulations based on the same experiment protocol, and whose outputs are thus directly comparable, it captures at least some of this spread in the projections.
Wartenburger R., and Sonia I Seneviratne and Martin Hirschi and Jinfeng Chang and Philippe Ciais and Delphine Deryng and Joshua Elliott and Christian Folberth and Simon N Gosling and Lukas Gudmundsson and Alexandra-Jane Henrot and Thomas Hickler and Akihiko Ito and Nikolay Khabarov and Hyungjun Kim and Guoyong Leng and Junguo Liu and Xingcai Liu and Yoshimitsu Masaki and Catherine Morfopoulos and Christoph Müller and Hannes Müller Schmied and Kazuya Nishina and Rene Orth and Yadu Pokhrel and Thomas A M Pugh and Yusuke Satoh and Sibyll Schaphoff and Erwin Schmid and Justin Sheffield and Tobias Stacke and Joerg Steinkamp and Qiuhong Tang and Wim Thiery and Yoshihide Wada and Xuhui Wang and Graham P Weedon and Hong Yang and Tian Zhou. (2018) Evapotranspiration simulations in ISIMIP2a—Evaluation of spatio-temporal characteristics with a comprehensive ensemble of independent datasets. Environmental Research Letters,13 (7), 075001. https://dx.doi.org/10.1088/1748-9326/aac4bb
\ No newline at end of file
This (link: glossary/climate-impact-assessment text: climate impact) assessment presents (link: glossary/projections text: projections) of the land area affected by droughts and the number of people exposed to droughts in Angola, based on 1 (link: glossary/global-hydrological-models text: global hydrological model) (GHMs) and 3 different global (link: glossary/climate-models text: climate models)(GCMs) participating in [ISIMIP](https://www.isimip.org/).
### Areas affected by - people exposed to drought
Here, (link: glossary/droughts text: droughts) refer to a situation where the monthly (link: glossary/soil-moisture text: soil moisture) conditions fall short of the 2.5th percentile (glossary/link: variable-monthly-threshold text: variable monthly threshold) for at least 7 consecutive months. To calculate the total land area affected by drought per country the land area exposed to drought is summed up across all grid cells belonging to Angola. Multiplying, at each grid cell, with the number of people living in the grid cell yields the number of people exposed to drought.
Authors: Stefan Lange<sup>1</sup>, Ted Veldkamp<sup>2</sup>, Matthias Mengel<sup>1</sup>, Hannes Müller Schmied<sup>3</sup>, Katja Frieler<sup>1</sup>
We analyze the land area affected by droughts and the number of people exposed to droughts at different (link: glossary/levels-of-global-warming text: levels of global warming)(0, 1, 2, and 3 degrees Celsius) and during different time periods. We cover the time before (link: glossary/anthropogenic-climate-change text: anthropogenic climate change) started (conventionally set at 1850), present-day (2001-2020), mid-century (2041-2060) and end-of-century (2081-2100). We compare the later periods against the climate conditions before anthropogenic climate change started (1850). A ranking of countries for each indicator is provided to allow comparison between countries on the basis of their (link: glossary/relative-change text: relative change) in the land area affected or the number of people exposed under different levels of global warming and for various time horizons, compared to conditions before anthropogenic climate change started. Here, a ranking of 1 implies that the country experiences a change in the (link: glossary/indicator text: indicator) that is more than all other ranked countries, whereas a lower ranking entails comparatively less effects due to climate change.
It is important to note that our (link: glossary/future-projections text: future projections) assume that the number of people living in an area, as well as the area’s (link: glossary/land-use text: land use) and (link: glossary/land-cover text: land cover), remain constant to that of the year 2005. This is not necessarily meant to be realistic, but to isolate the influence of climate change from the influence of other changes.
Affilitations:
1. Potsdam Institute for Climate Impact Research, Germany
2. Vrije Universiteit Amsterdam, The Netherlands
3. Goethe University Frankfurt, Germany
Published: [to be added]
Doi: [to be added]
In the key messages we report the results for the (link: glossary/median text: median) of the (link: glossary/model-ensemble text: model ensemble) of global hydrological models providing data to ISIMIP. The median represents the middle of the ensemble, meaning that 50% of the ensemble members provide higher numbers and 50% provide lower numbers. After the key message, which only references the median, you can find the model-specific results in the figures, a discussion of the methodology and the restrictions of the analysis at the end of the article.
### Key messages
* Without human-made greenhouse gas emissions, 0.328% of Angola’s land area and 0.152% of Angola’s population would be affected by droughts each year, on average.
* At today’s levels of 1 degrees Celsius of global warming the land area affected is 1.722 times as much: 72.175% of the total land area, while the number of people affected is 1.529 times as much: 0.243% of the total population.
* At 2 degrees Celsius of global warming, Angola’s land area affected by droughts is projected to increase by a factor of 102.678 compared to a world without human-made greenhouse gas emissions, to 0.328%. Likewise, Angola’s population exposed to droughts is projected to increase by a factor 44.434, to 0.152%.
* Following the higher-emissions scenario (RCP6.0) which can entail over 3 degrees Celsius of global warming by the end of the century (2081-2100), this factor is projected to reach 1.004 for the land area affected to droughts (to 1.004%) and 1.001 for the population exposed to droughts (to 0.14%).
* Angola ranks (ranking-value: land-rel-temp_AGO value: position temperature:2) with regards to its relative change in land area affected by droughts at 2 degrees Celsius of global warming in comparison to a situation without anthropogenic climate change. For the relative change in population exposed to droughts, Angola’s ranking is (ranking-value: pop-rel-temp_AGO value: position temperature:2).
### Results
The figures below shows the relative change in Angola’s land area affected by and population exposed to droughts in comparison to the time period before anthropogenic climate change. Results are shown both for the (link: glossary/historical-period text: historical time period) and for the (link: glossary/future-projections text: future projections) and can be visualized with regards to their change over time as well as with regards to their change in terms of global warming. Under the ‘change in terms of time’ setting, the change in land area affected by and population exposed to droughts over time is shown for 20-year (link: glossary/time-slices text: time-slices) from 1861-1880 until the end-of-the-century (2081-2100). Under the ‘change in terms of global warming’, the change in land area affected by and population exposed to droughts is shown for 20-year (link: glossary/time-slices text time-slices) representing increasing levels of global warming from 0 degrees Celsius up to 3 degrees Celsius compared to the time period before anthropogenic climate change started.
* Angola ranks (ranking-value: land-abs-temp_AGO value: position temperature:2) with regards to absolute changes in land area affected by droughts (expressed as % of Angola’s land area) at 2°C of global warming in comparison to a situation without climate change. For the absolute changes in population exposed to droughts (expressed as % of Angola’s population), Angola ranks (ranking-value: pop-abs-temp_AGO value: position temperature:2).
* At today’s levels of 1°C of global warming the simulated land area affected is already 6800 km<sup>2</sup> larger (1.0% of the national land area) than in a world without climate change where the annual area affected by droughts is 4100 km<sup>2</sup> (0.3% of Angola’s land area). The number of people exposed is 0.0 million (0.1% of the national population) larger than without climate change where the annual number of people exposed to droughts was 0.0 million (0.2% of Angola’s population).
* At 2°C of global warming the land area affected by droughts would increase by 12000 km<sup>2</sup> (1.0% of the national land area) compared to a world without climate change, to 0.6% of the country’s land area. Assuming present-day population patterns, Angola’s population exposed to droughts would increase by 0.1 million, to 0.1% of the national population.
* Following the higher-emissions scenario (RCP6.0) which can entail over 3°C of global warming by the end of the century (2081-2100) ([Frieler et al. 2017](https://dx.doi.org/10.5194/gmd-10-4321-2017)) the land area affected by droughts would increase by 20200 km<sup>2</sup> (1.6% of the national land area) and reach 1.8% of the country’s land area. Assuming present-day population patterns the population exposed would reach 0.4% of Angola’s population, and increase by 0.3 million compared to a situation without climate change.
### How have we got the results?
The assessment provided below is based on our paper “Change in exposure to climate impact events under global warming” submitted to the journal “Earth’s Future”. We have used simulations of 1 different (link: glossary/global-hydrological-models text: global hydrological models) to estimate changes in drought conditions assuming different levels of global mean temperature change. To this end the global hydrological models were forced by climate projections from 3 different global (link: glossary/climate-models text: climate models) following the ISIMIP2b scenario set-up. In the key messages we report the results for the median of the (link: glossary/model-ensemble text: model ensemble) of global hydrological models that ran simulations in (link: glossary/isimip2b text: ISIMIP2b). The median represents the middle of the ensemble, meaning that 50% of the ensemble members provide higher numbers and 50% provide lower numbers.
Among the multiple concepts of “drought”, our study refers to a situation where monthly soil moisture (agricultural drought concept) is extremely low for at least 7 consecutive months. “Extremely low” means dry conditions occurring less than 5 times in 200 years under reference conditions without climate change. We were interested in changes in the land area affected by droughts and the number of people exposed to droughts that we have to expect at different (link: glossary/levels-of-global-warming text: levels of global warming) (1°C, 2°C, and 3°C) or during different future time periods.
The ISIMIP2b simulations start in 1860 (before human greenhouse gas emissions started to change the climate) and end in 2100. They comprise a long run where weather only varies according to pre-industrial conditions without climate change. In addition, modellers have run two different future scenarios: one “business as usual” scenario (RCP6.0) reaching high levels of warming and one “low emissions scenario” (RCP2.6) with reduced levels of climate change.
As the “land area affected by drought” varies quite strongly from year to year we decided to average numbers over multiple years: for each year we calculate the deviation of the land area affected by droughts from the average reference level without climate change. These deviations are then averaged over a present-day period (2001-2020), a mid-century period (2041-2060) and an end-of-century period (2081-2100). To calculate the changes at different levels of global mean temperature change we do not average over years belonging to one of these periods but over all years with global mean warming levels close to 1°C, 2°C, and 3°C.
When calculating the “number of people exposed to droughts” we proceed in a similar way. We assume that only the rural population is exposed while the urban population does not necessarily “feel” any drought. For 1860 to 2005 we account for changes in population patterns ([Goldewijk et al., 2017](https://dx.doi.org/10.5194/essd-9-927-2017)). Afterwards the population data is considered constant -- not as a realistic assumption but an informative “thought experiment”: What would future climate change mean for present day societies? For all time periods, the pure effect of climate change on the “number of people exposed to droughts” is estimated by comparing the number of people exposed under climate change and a specific population pattern to the number of people exposed assuming the same population patterns but no climate change.
In the following we describe what we have found and how strongly Angola is affected compared to other countries.
### What we have found
### Land area affected by droughts
(line-plot: land-abs-temp_AGO,land-abs-time_AGO first-temperature: 2 second-scenario: rcp26 second-time: 2041-2060)
Our definition of “drought” is quite strict, such that, without climate change, only 0.3% of Angola’s land area would be affected by droughts each year, on average.
However, at today’s level of 1°C global warming Angola’s annual land area affected by droughts is, on average, already larger and amount to 6800 km<sup>2</sup> (0.5% of the national land area). The level of change ranges from -0.1% to 12.6% for the individual combinations of global hydrological models and global climate models. At 2°C of global warming, Angola’s annual land area affected by droughts is projected to increase by 12000 km<sup>2</sup> (i.e. 1.0% of the national land area) on average in comparison to a world without climate change. Under these conditions, 0.6% of the national land area would be affected by droughts each year, on average. Across the individual combinations of global hydrological models and global climate models this expected level of change ranges from -800 to 157300 km<sup>2</sup>.
Following the higher-emissions scenario (RCP6.0) the land area affected by droughts is expected to increase by 14300 km<sup>2</sup> (1.1% of the national land area) towards the end of the century (2081-2100). Following the low emission scenario (RCP2.6) the change would only reach 14000 km<sup>2</sup> (1.1% of the national land area). By the middle of the century, changes reach 4700 km<sup>2</sup> under RCP2.6 and 18400 km<sup>2</sup> under RCP6.0.
Angola is the (ranking-value: land-abs-temp_AGO value: position temperature:2) strongest affected by droughts at 2°C of global warming. For the absolute change in land area affected by droughts towards the end of the century under a high-emissions scenario (RCP6.0), Angola’s ranking is (ranking-value: land-abs-time_AGO value: position time:2081-2100 scenario:rcp60).
### Population exposed to droughts
(line-plot: pop-abs-temp_AGO,pop-abs-time_AGO first-temperature: 2 second-scenario: rcp26 second-time: 2041-2060)
Our definition of “drought” is quite strict, such that, without climate change, only 0.0% of Angola’s population would be exposed to droughts each year, on average.
However, at today’s level of 1°C of global warming Angola’s annual population exposed to droughts is, on average, already 0.0 million (i.e. 0.1% of the total population) higher than without climate change and amount to 0.2% of the total population. The level of change ranges from -0.1 % to 12.6% for the individual combinations of global hydrological models and global climate models. At 2°C of global warming, Angola’s annual population exposed to droughts is projected to increase by 0 million (i.e. 0.1% of the population) on average in comparison to a world without climate change. Under these conditions, 0.1% of the total population would be affected by droughts, on average. Across the individual combinations of global hydrological models and global climate models this expected level of change ranges from -0.0 up to 1.0 million people.
Following the higher-emissions scenario (RCP6.0) the population exposed to droughts is expected to increase by 0.1 million (0.2% of the total population) towards the end of the century (2081-2100). Following the low emission scenario (RCP2.6) the change would expose 0.0 million people (0.2% of the total population). By mid of the century changes expose 0.0 million people under RCP2.6 and 0.0 million people under RCP6.0.
Angola is the (ranking-value: pop-abs-temp_AGO value: position temperature:2) strongest affected by droughts at 2°C of global warming. For the absolute change in population exposed to droughts towards the end of the century under a higher-emissions scenario (RCP6.0), Angola’s ranking is (ranking-value: pop-abs-time_AGO value: position time:2081-2100 scenario:rcp60).
### How is soil moisture calculated?
In both cases, results are shown for all possible combinations of (link: glossary/global-hydrological-models text: global hydrological models) and (link: glossary/global-climate-models text: global climate models) (normal lines) as well as for the median of the model ensemble (thick line). The influence of (link: glossary/inter-annual-variability text: inter-annual variability) on the results is visualized by means of the shaded area in black. This inter-annual variability is only shown for the median model combination. For the future projections and under the ‘change in terms of time’ settings only, an additional distinction is made between the results of two future (link: glossary/emissions-trajectories text: emissions trajectories): a low-emissions trajectory limiting global mean temperature in the 21st century to below 2 degrees Celsius compared to before anthropogenic climate change started, called RCP2.6 (blue lines), and a higher-emissions scenario leading global mean temperature to above 3 degrees Celsius by 2100, compared to pre-climate change conditions, called RCP6.0 (red lines). The shaded areas in blue and red indicate the variety (or spread) in results between all combinations of global hydrological models and global climate models for these two emissions trajectories. A filtering menu top-right of the graph allows the user to select individual emissions trajectories, individual global climate models, and individual global hydrological models for visualization.
Soil moisture is defined as the water stored in the soil in liquid or frozen form. It is calculated as the difference between the amount of water coming in as rain or snow, and the amount of water going out, either through evapotranspiration from the surface and vegetation, runoff, or through percolation of water towards the deeper groundwater layers. Global hydrological models (GHMs) include different representations of the (link: glossary/soil-water-column text: soil water column) with different numbers of soil water layer(s) (between 1 and 15 layers), and with different total depth of soil layers (between 1 and 42.1 meter).
Here, we use (link: glossary/rootzone-soil-moisture text: rootzone soil moisture) estimates, the portion of soil moisture that is within the rooting depth of plants, when directly provided by the GHMs. For GHMs that do not directly provide root zone soil moisture conditions, we approximated this variable by integrating soil moisture across multiple soil water layers in order to reach a depth of ~1 meter. Daily rootzone soil moisture values are finally aggregated into monthly average rootzone soil moisture conditions per grid cell.
By hovering over or clicking on a particular value in the figure additional details, such as the specific global climate model or hydrological used, behind the presented value become available. The visualization below the graph shows, for the selected model run under this time-period or temperature-change level, how Angola ranks in comparison to other countries on its relative change in land area affected by or population exposed to droughts.
#### Land area affected by droughts
### Where are the most relevant gaps in our knowledge?
Models are simplified representations of reality, hence model simulations come with limitations and uncertainties that have to be kept in mind ([Döll et al. 2016](https://dx.doi.org/10.1007/s10712-015-9343-1)). Regarding our drought simulations we mainly see the following issues.
(line-plot: land-rel-temp_AGO,land-rel-time_AGO first-temperature: 2 second-scenario: rcp26 second-time: 2041-2060)
Without human-made greenhouse gas emissions, 0.328% of Angola’s land area would be affected by droughts each year, on average.
* Representation of direct human influences
Under the ‘change in terms of global warming’ setting the figure shows that at today’s level of 1 degrees Celsius of global warming Angola’s land area affected by droughts is, on average, already 1.722 times (or 72.175%) lower and amount to 0.573% of the total land area. This level of change ranges from 0.194% to 0.342% for the individual combinations of global hydrological models and global climate models. At 2 degrees Celsius of global warming, Angola’s land area affected by droughts is projected to change by a factor of 2.027 (or 0.332%) in comparison to a world without human-made greenhouse gas emissions. Under these conditions, 0.631% of the total land area would be affected by droughts on a yearly basis, on average. Across the individual combinations of global hydrological models and global climate models this expected level of change ranges from 0.194% up to 0.342%.
When presenting the ‘change in terms of time’, we find that when following the higher-emissions scenario (RCP6.0) towards the end-of-the-century (equivalent to over 3 degrees Celsius change) would result in a change in the land area affected by droughts of, on average, a factor 1.993 (0.224%), towards: 0.449% of the total land area. Following the (link: glossary/climate-mitigation-emissions-scenario text: climate-mitigation emissions scenario)(RCP2.6) towards the end-of-the-century (entailing an average 2.5 degrees Celsius change) would result in a foreseen change in the land area affected by droughts of, on average, a factor 1.733 (0.248%), 0.613% of the total land area being affected.
The presented drought projections reflect the isolated effect of climate change, while their local manifestations are expected to also be influenced by direct human drivers like changes in land use, land cover, and irrigation. Even in the historical period our simulations only account for water abstractions (with varying sectors among the models like irrigation, industry, domestic, livestock, desalinization) from groundwater and surface waters, reservoir management while other processes such as inter-basin water transfers and a more realistic reservoir management are also expected to be important. Including these drivers into our simulation would greatly improve the simulations ([Veldkamp et al. 2018](https://doi.org/10.1088/1748-9326/aab96f)). That it is not done is mainly an issue of missing knowledge about them.
Globally, the land area affected by droughts each year, on average, is 2.377% in a situation without human-made greenhouse gas emissions. At 2 degrees Celsius global warming or by the end-of-the-century under a higher-emissions scenario (RCP6.0) these values are foreseen to increase globally by a factor 3.106 and 7.014%, respectively.
As such, Angola ranks (ranking-value: land-rel-temp_AGO value: position temperature:2) with regards to its relative change in land area affected by droughts at 2 degrees Celsius of global warming in comparison to a situation without anthropogenic climate change. For the relative change in land area affected by droughts towards the-end-of-the-century under a higher-emissions scenario (RCP6.0), Angola’s ranking is (ranking-value: land-rel-time_AGO value: position time:2081-2100 scenario: rcp60).
* Representation of the CO2 fertilization effect
#### Population exposed to droughts
(line-plot: pop-rel-temp_AGO,pop-rel-time_AGO first-temperature: 2 second-scenario: rcp26 second-time: 2041-2060)
Another challenge lies in the representation of how vegetation cover affects evapotranspiration, a relationship that is subject to change under rising temperature and CO2 levels. For example, the effect of CO2 fertilization – the phenomenon through which photosynthesis, hence plant growth, should be enhanced in a CO2-richer atmosphere – is not represented in most global hydrological models as the vegetation components in those models are often simplified. Plants need less water to assimilate the same amount of carbon in a CO2-richer atmosphere. This effect alone would lead to a decrease of evapotranspiration. Yet the more efficient carbon assimilation lets plants grow better, which leads to bigger or more plants. This effect alone would lead to an increase of evapotranspiration. The overall CO2 fertilization effect on evapotranspiration, soil moisture conditions and drought risk is uncertain (Gerten et al., 2014; [Prudhomme et al., 2014](https://dx.doi.org/10.1073/pnas.1222473110); [Döll et al., 2016](https://dx.doi.org/10.1007/s10712-015-9343-1); [Kuzyakov et al., 2019](https://dx.doi.org/10.1016/j.soilbio.2018.10.005)).
Without human-made greenhouse gas emissions, 0.152% of Angola’s population would be exposed to droughts each year, on average.
Under the ‘change in terms of global warming’ setting the figure shows that at today’s level of 1 degrees Celsius of global warming Angola’s population exposed to droughts is, on average, already 1.529 times (or 52.863%) lower: 0.243% of the total land area. This level of change ranges from 0.017% to 0.264% for the individual combinations of global hydrological models and global climate models. At 2 degrees Celsius of global warming, Angola’s population exposed to droughts is projected to change by a factor of 1.444 (or 0.028%) in comparison to a world without human-made greenhouse gas emissions. Under these conditions, 0.091% of the total population would be exposed to droughts on a yearly basis, on average. Across the individual combinations of global hydrological models and global climate models this expected level of change ranges from -0.017% up to 0.108%.
* Representation of the soil compartment
When presenting the ‘change in terms of time’, we find that when following the higher-emissions scenario (RCP6.0) towards the end-of-the-century (equivalent to over 3 degrees Celsius change) would result in a foreseen change in the population exposed to droughts of, on average, a factor 3.456 (0.1%),0.14% of the total population. Following the climate-mitigation emissions scenario (RCP2.6) towards the end-of-the-century (entailing an average 2.5 degrees Celsius change) would result in a foreseen change in the population exposed to droughts of, on average, a factor 2.104 (0.045%), 0.147% of the total population being exposed.
The representation of the soil compartment vary with the total depth of soil layers (between 1 and 42.1 m), the number of soil layers (ranging from 1 up to 15) and the conceptualization of soil water processes. In many global hydrological models, we use here root zone soil moisture (related to agricultural needs) as an input for our drought assessment. However, in other models, we have to approximate this variable by integrating soil moisture across multiple water layers. Differences in the representation of the soil column and soil water processes can lead to significant differences in estimates of soil water availability, soil water saturation and drought conditions but this effect could not be investigated for the models used in this assessment.
Globally, the population exposed to droughts each year, on average, is 0.601% in a situation without human-made greenhouse gas emissions. At 2 degrees of global warming or by the end-of-the-century under a higher-emissions scenario (RCP6.0) these values are foreseen to increase globally by 3.264% and 10.391% respectively.
As such, Angola ranks (ranking-value: pop-rel-temp_AGO value: position temperature:2) with regards to its relative change in population exposed to droughts at 2 degrees Celsius of global warming in comparison to a situation without anthropogenic climate change. For the relative change in population exposed to droughts towards the-end-of-the-century under a business-as-usual emissions scenario, Angola’s ranking is (ranking-value: pop-rel-time_AGO value: position time:2081-2100 scenario: rcp60).
* Representation of evapotranspiration in general
### Methodology
Much of the differences between the individual hydrological model simulations considered here is assumed to be due to different calculation of potential and actual evapotranspiration. Whereas some models directly calculates actual evapotranspiration (AET, the amount of water that is transferred through plants, bare soil, open water bodies or from water stored at canopies) directly using water transfer schemes or turbulent fluxes, the majority of the models use the potential evapotranspiration (PET) concept. By using various approaches with different demand on meteorological input variables, the potential water demand of the atmosphere (PET as upper limit) is calculated. The availability of water in the storages (e.g. soil water storage) can reduce the amount of water that is actually evapotranspired to the atmosphere (AET). Unfortunately, currently, we are still not able to decide which of the representations is most appropriate but first assessments of those representations have been done ([Wartenburger et al. 2018](https://doi.org/10.1088/1748-9326/aac4bb)). As a result, estimates of actual evapotranspiration can differ significantly across models despite having used uniform inputs from the global climate models, which may affect estimates of drought conditions much more than for example flooding which mainly depends on the amount of precipitation.
#### What do we analyze?
We analyze the land area affected by droughts and the number of people exposed to droughts at different levels of global warming (0, 1, 2, and 3 degrees Celsius) and during different time periods. We cover the time before anthropogenic climate change started (1850), present-day (2001-2020), mid-century (2041-2060) and end-of-century (2081-2100). We compare the later periods against the climate conditions before anthropogenic climate change (1850).
#### How do we calculate where and when a drought occurs?
The calculation has several steps: 1) calculate (link: glossary/soil-moisture text: soil moisture); 2) calculate the 2.5th percentile (glossary/link: variable-monthly-threshold text: variable monthly threshold) value for monthly soil moisture conditions; and 3) evaluate where and when a grid cell is exposed to drought conditions for at least 7 consecutive months. All these steps are detailed below, after a short explanation of the spatial structure on which our models operate.
### Disclaimer
#### What is the spatial structure of the models?
The models we use cover the whole globe’s land area. The land area is divided into a grid. Each grid cell has a size of 0.5 degrees × 0.5 degrees (latitude by longitude). At the equator, this equals roughly 55 × 55 km; towards the North Pole or South Pole, where the land area covered per grid cell becomes smaller.
Note, that although this report is based on an article published in a peer-reviewed scientific journal, the original publication focused on a global analysis. Therefore, the results presented here for 200+ countries have not been reviewed individually. ISIpedia provides data on country level for convenience, but cannot be held responsible for any issues with the data. Please contact the ISIpedia editorial team (isipedia.editorial.team@pik-potsdam.de) for more information or questions about this report.
#### How is soil moisture calculated?
The calculation is done for each grid cell and each day. Information about temperature, precipitation, solar radiation, and other weather indicators is taken from the global climate models (GCMs) and used as input for the ISIMIP global hydrological models (GHMs). Additional spatial data, such as soil, land cover and water bodies are also inputted. Soil moisture, i.e. the water stored in the soil in liquid or frozen form, is calculated as the difference between the amount of water coming in as rain or snow, and the amount of water going out, either through evapotranspiration from the surface and vegetation, or through percolation of water towards the deeper ground water layers. Global hydrological models include a different representation of the (link: glossary/soil-water-column text: soil water column) with different numbers of soil compartments with varying depths of the soil water layer(s), ranging from one soil water layer with a depth of between 0.1 - 4 meter up to 15 fully resolved soil water layers with a total depth up to 45 meter.
### References
Döll, P., Douville, H., Güntner, A., Müller Schmied, H., Wada, Y. (2016) Modelling Freshwater Resources at the Global Scale: Challenges and Prospects. Surveys in Geophysics, 37(2), 195–221. https://dx.doi.org/10.1007/s10712-015-9343-1
Here, we use (link: glossary/rootzone-soil-moisture text: rootzone soil moisture) estimates, the portion of soil moisture that is within the rooting depth of plants, when directly provided by the global hydrological models. For global hydrological models that do not directly provide rootzone soil moisture conditions, we approximated this variable by integrating soil moisture across multiple soil water layers in order to reach a depth of ~1 meter.
Daily rootzone soil moisture values were finally aggregated into monthly average rootzone soil moisture conditions per grid cell.
Frieler, K. and Lange, S. and Piontek, F. and Reyer, C. P. O. and Schewe, J. and Warszawski, L. and Zhao, F. and Chini, L. and Denvil, S. and Emanuel, K. and Geiger, T. and Halladay, K. and Hurtt, G. and Mengel, M. and Murakami, D. and Ostberg, S. and Popp, A. and Riva, R. and Stevanovic, M. and Suzuki, T. and Volkholz, J. and Burke, E. and Ciais, P. and Ebi, K. and Eddy, T. D. and Elliott, J. and Galbraith, E. and Gosling, S. N. and Hattermann, F. and Hickler, T. and Hinkel, J. and Hof, C. and Huber, V. and J\"agermeyr, J. and Krysanova, V. and Marc\'e, R. and M\"uller Schmied, H. and Mouratiadou, I. and Pierson, D. and Tittensor, D. P. and Vautard, R. and van Vliet, M. and Biber, M. F. and Betts, R. A. and Bodirsky, B. L. and Deryng, D. and Frolking, S. and Jones, C. D. and Lotze, H. K. and Lotze-Campen, H. and Sahajpal, R. and Thonicke, K. and Tian, H. and Yamagata, Y. (2017). Assessing the impacts of 1.5 °C global warming -- simulation protocol of the Inter-Sectoral Impact Model Intercomparison Project (ISIMIP2b). Geoscientific Model Development,10 (12), 4321--4345. https://dx.doi.org/10.5194/gmd-10-4321-2017
#### How is the 2.5th percentile variable monthly threshold calculated?
To distinguish between periods of drought and periods of no drought we applied a (link: glossary/variable-monthly-threshold text: variable monthly threshold) level method over the monthly rootzone soil moisture conditions, using a 2.5th percentile value. Using the pre-industrial scenario as a reference period, we identified at the grid-level, per global hydrological model and global climate model, and for each month individually, the monthly rootzone soil moisture value that represents the 2.5th percentile value. This can be interpret as the monthly rootzone soil moisture condition that is exceeded 39 out of 40 times throughout the full time-period under study. The use of a variable monthly threshold to distinguish drought from periods of no drought allows accounting for seasonal climatology, which is relevant for the management of water resources.
#### How do we know whether a drought occurs?
With these obtained 2.5th percentile variable monthly threshold values, we then identify for the different climate change scenarios how often monthly rootzone soil moisture conditions fall under this threshold. To identify prolonged drought events, and omit incidental occurrences of droughts lasting only a short period of time, we applied a six-month threshold. Only droughts that last longer than six-months were taken into account in the analysis.
Gerten, D. and Betts, R. and Döll, P. (2014). Cross-chapter box on the active role of vegetation in altering water flows under climate change. In: Field CB, Barros VR, Dokken DJ, Mach KJ, Mastrandrea MD, Bilir TE, Chatterjee M, Ebi KL, Estrada YO, Genova RC, Girma B, Kissel ES, Levy AN, MacCracken S, Mastrandrea PR, White LL (eds) Climate Change 2014: impacts, adaptation, and vulnerability. Part A: global and sectoral aspects. Contribution of working group II to the 5th assessment report of the intergovernmental panel on climate change. Cambridge University Press, Cambridge, pp 157–161
#### How are the indicators land area affected by droughts and number of people exposed to droughts calculated?
To estimate the land area affected by and population exposed to drought we checked for each year whether a grid cell was exposed for at least one period of prolonged drought. If so, the grid cell is accounted as ‘being exposed to drought’ for that respective year. We then calculate the total land area affected by drought for Angola by adding up all 0.5 degree x 0.5 degree grid cells belonging to Angola. To calculate the number of people exposed to droughts, we first multiply the each grid cell by the number of people living in that grid cell to estimate the number of people exposed to droughts, and then add up those numbers across all grid cells belonging to Angola.
#### What else should I know about the methodology?
We assume that the number of people living in an area, as well as the area’s land use and land cover (what fractions of the area are used for settlement, cropland, or pasture, or covered by forest), remain constant at the levels of year 2005 throughout the whole simulation. This is not meant to be realistic, but to isolate the influence of climate change from the influence of other changes.
Goldewijk, K., Beusen, A., Doelman, J., Stehfest, E. (2017) Anthropogenic land use estimates for the Holocene – HYDE 3.2. Earth Syst. Sci. Data, 9, 927–953. https://dx.doi.org/10.5194/essd-9-927-2017
In the key messages we report the results for the median (the model in the middle) of the group of global hydrological models that ran simulations in (link: glossary/isimip2b text: ISIMIP2b). All model-specific results are presented in the figures and a discussion of the limitations of the analysis is included at the end of the report.
A ranking of countries for each indicator is also provided to compare countries on the basis of their relative change in land area affected or number of people exposed under different levels of global warming and for various time horizons, compared to conditions without climate change. A ranking of 1 implies that the country experiences a change in the indicator higher than all other ranked countries, for example the strongest impacts from climate change; whereas a lower ranking entails comparatively less impacts.
Kuzyakov, Y., Horwath, W. R., Dorodnikov, M. and Blagodatskaya, E. (2019). Review and synthesis of the effects of elevated atmospheric CO2 on soil processes: No changes in pools, but increased fluxes and accelerated cycles, Soil Biology and Biochemistry, 128, 66-78, https://dx.doi.org/10.1016/j.soilbio.2018.10.005
### Discussion
The model simulations used for this report build on science that has been established through many peer-reviewed studies: for example by [Sheffield and Wood (2008)](https://doi.org/10.1007/s00382-007-0340-z), [Trenberth, et al. (2014)](https://www.nature.com/articles/nclimate2067), [Berg, et al. (2017)](https://doi.org/10.1002/2016GL071921), and [Berg and Sheffield (2018)](https://doi.org/10.1007/s40641-018-0095-0).
Models are simplified representations of reality, hence model simulations come with limitations and uncertainties that have to be kept in mind. Using multiple global hydrological models with different representations and parameterizations of the hydrological components and processes may introduce biases in the estimation of soil moisture conditions and in the resulting assessment of droughts ([Veldkamp et al., 2018](https://doi.org/10.1088/1748-9326/aab96f)). Examples are the representation of evapotranspiration within global hydrological models and the representation of soil water column.
Lange et al. Change in exposure to climate impact events under global warming. Submitted to Earth’s Future.
Different global hydrological models apply different methods to estimate (link: glossary/potential-evapotranspiraton text: (potential) evapotranspiration). These vary from the (link: glossary/bulk-formula text: Bulk formula) and the (link: glossary/monin-obukhov-similarity-theory text: Monin-Obukhov similarity theory) to (link: glossary/penman-montheith text: Penman-Montheith), (link: glossary/hamon text: Hamon), and (link: glossary/priestley-taylor text: Priestley-Taylor). In result, estimates of (potential) evapotranspiration can differ significantly across models despite having used uniform inputs from the global climate models. This may affect estimates of soil moistures conditions. Another challenge lies in the representation of how vegetation cover affects evapotranspiration, a relationship that is subject to change within a given river basin, or with rising temperature and CO2 levels. For example, the effect of CO2 fertilization – the phenomenon through which photosynthesis, hence plant growth, should be enhanced in a CO2-richer atmosphere – is not represented in most global hydrological models although it would affect the plants’ water use, with significant consequences for soil moisture conditions and drought estimates.
Using multiple global hydrological models with different representations of the soil water column introduces additional biases in the estimation of soil moisture conditions and droughts. The global hydrological models applied here include a different numbers of soil compartments with varying depths of their soil water layer(s). These representations range from one soil water layer with a depth of between 0.1 - 4 meter up to 15 fully resolved soil water layers with a total depth up to 45 meter. When directly provided by the global hydrological models, we use here root zone soil moisture as an input for our drought assessment. For global hydrological models that did not provide root zone soil moisture directly, we approximated this variable by integrating soil moisture across multiple soil water layers in order to reach a depth of ~1 meter. Differences in the representation of the soil column and soil water processes can lead to significant differences in estimates of soil water availability, soil water saturation and drought conditions.
Prudhomme, C.; Giuntoli, I.; Robinson, E. L.; Clark, D. B.; Arnell, N. W.; Dankers, R.; Fekete, B. M.; Franssen, W.; Gerten, D.; Gosling, S. N.; Hagemann, S.; Hannah, D. M.; Kim, H.; Masaki, Y.; Satoh, Y.; Stacke, T.; Wada, Y. & Wisser, D. (2014). Hydrological droughts in the 21st century, hotspots and uncertainties from a global multimodel ensemble experiment, Proceedings of the National Academy of Sciences, National Academy of Sciences, 111, 3262-3267, https://dx.doi.org/10.1073/pnas.1222473110
The performance of the global hydrological models can be tested by using (link: glossary/observed-historical-weather-information text: observed historical weather information) as inputs, and comparing the simulated soil moisture to soil moisture observations, locally measured or remotely sensed.
Apparent conflicting results of how droughts develop due climate change highlight the sensitivity of drought estimates to the choice of indicator and hydrological variable used ([Trenberth, et al. (2014)](https://www.nature.com/articles/nclimate2067), [Berg, et al. (2017)](https://doi.org/10.1002/2016GL071921)). Here, we use a variable monthly percentile threshold in combination with root zone soil moisture conditions to distinguish drought conditions from no drought conditions. Alternative indicators representing soil moisture droughts, such as the (link: glossary/palmer-drought-severity-index text: Palmer Drought Severity Index (PDSI)), the (link: glossary/standardized-soil-moisture-index text: Standardized Soil Moisture Index (SSMI)), or indicators representing a different part of the (link: glossary/hydrological-cycle text: hydrological cycle) (for example the (link: glossary/standardized-precipitation-index text: standardized precipitation index (SPI)), the (link: glossary/standardized-precipitation-and-evapotranspiration-index text: standardized precipitation and evapotranspiration index (SPEI)), or the (link: glossary/standardized-runoff-index text: standardized runoff index (SRI))) may result in different conclusions on the absolute and relative changes in land area affected by and population exposed to drought due to climate change.
Veldkamp, T. I.E., Zhao, F., Ward, P.J., de Moel, H., Aerts, J.C.J.H., Müller Schmied, H., Portmann, F.T., Masaki, Y., Pokhrel, Y., Liu, X., Satoh, Y., Gerten, D. Gosling, S.N., Zaherpour, J., Wada, Y. (2018) Human impact parameterizations in global hydrological models improve estimates of monthly discharges and hydrological extremes: a multimodel validation study. Environ. Res. Lett. 13 055008, https://doi.org/10.1088/1748-9326/aab96f
The presented drought projections reflect the isolated effect of climate change, while their local manifestations are expected to also be influenced by direct human drivers like changes in land use, land cover, and irrigation. The representation of these drivers is limited to (link: glossary/present-day-conditions text: present-day conditions), while their future changes are not explicitly modelled in the presented simulations. This serves the purpose of isolating the effect of anthropogenic climate change from other factors, but it means that actual drought exposure in the future could be more or less acute, depending on non-climatic human drivers, such as irrigation or adaptation.
Regarding climate models and global hydrological models, it is known that generally both types of models contribute substantially to the overall spread in projected climate change impacts on water-related variables. Because this report presents the results of a combination of climate models and global hydrological models that ran simulations based on the same experiment protocol, and whose outputs are thus directly comparable, it captures at least some of this spread in the projections.
\ No newline at end of file
Wartenburger R., and Sonia I Seneviratne and Martin Hirschi and Jinfeng Chang and Philippe Ciais and Delphine Deryng and Joshua Elliott and Christian Folberth and Simon N Gosling and Lukas Gudmundsson and Alexandra-Jane Henrot and Thomas Hickler and Akihiko Ito and Nikolay Khabarov and Hyungjun Kim and Guoyong Leng and Junguo Liu and Xingcai Liu and Yoshimitsu Masaki and Catherine Morfopoulos and Christoph Müller and Hannes Müller Schmied and Kazuya Nishina and Rene Orth and Yadu Pokhrel and Thomas A M Pugh and Yusuke Satoh and Sibyll Schaphoff and Erwin Schmid and Justin Sheffield and Tobias Stacke and Joerg Steinkamp and Qiuhong Tang and Wim Thiery and Yoshihide Wada and Xuhui Wang and Graham P Weedon and Hong Yang and Tian Zhou. (2018) Evapotranspiration simulations in ISIMIP2a—Evaluation of spatio-temporal characteristics with a comprehensive ensemble of independent datasets. Environmental Research Letters,13 (7), 075001. https://dx.doi.org/10.1088/1748-9326/aac4bb
\ No newline at end of file
This (link: glossary/climate-impact-assessment text: climate impact) assessment presents (link: glossary/projections text: projections) of the land area affected by droughts and the number of people exposed to droughts in Albania, based on 1 (link: glossary/global-hydrological-models text: global hydrological model) (GHMs) and 3 different global (link: glossary/climate-models text: climate models)(GCMs) participating in [ISIMIP](https://www.isimip.org/).
### Areas affected by - people exposed to drought
Here, (link: glossary/droughts text: droughts) refer to a situation where the monthly (link: glossary/soil-moisture text: soil moisture) conditions fall short of the 2.5th percentile (glossary/link: variable-monthly-threshold text: variable monthly threshold) for at least 7 consecutive months. To calculate the total land area affected by drought per country the land area exposed to drought is summed up across all grid cells belonging to Albania. Multiplying, at each grid cell, with the number of people living in the grid cell yields the number of people exposed to drought.
Authors: Stefan Lange<sup>1</sup>, Ted Veldkamp<sup>2</sup>, Matthias Mengel<sup>1</sup>, Hannes Müller Schmied<sup>3</sup>, Katja Frieler<sup>1</sup>
We analyze the land area affected by droughts and the number of people exposed to droughts at different (link: glossary/levels-of-global-warming text: levels of global warming)(0, 1, 2, and 3 degrees Celsius) and during different time periods. We cover the time before (link: glossary/anthropogenic-climate-change text: anthropogenic climate change) started (conventionally set at 1850), present-day (2001-2020), mid-century (2041-2060) and end-of-century (2081-2100). We compare the later periods against the climate conditions before anthropogenic climate change started (1850). A ranking of countries for each indicator is provided to allow comparison between countries on the basis of their (link: glossary/relative-change text: relative change) in the land area affected or the number of people exposed under different levels of global warming and for various time horizons, compared to conditions before anthropogenic climate change started. Here, a ranking of 1 implies that the country experiences a change in the (link: glossary/indicator text: indicator) that is more than all other ranked countries, whereas a lower ranking entails comparatively less effects due to climate change.
It is important to note that our (link: glossary/future-projections text: future projections) assume that the number of people living in an area, as well as the area’s (link: glossary/land-use text: land use) and (link: glossary/land-cover text: land cover), remain constant to that of the year 2005. This is not necessarily meant to be realistic, but to isolate the influence of climate change from the influence of other changes.
Affilitations:
1. Potsdam Institute for Climate Impact Research, Germany
2. Vrije Universiteit Amsterdam, The Netherlands
3. Goethe University Frankfurt, Germany
Published: [to be added]
Doi: [to be added]
In the key messages we report the results for the (link: glossary/median text: median) of the (link: glossary/model-ensemble text: model ensemble) of global hydrological models providing data to ISIMIP. The median represents the middle of the ensemble, meaning that 50% of the ensemble members provide higher numbers and 50% provide lower numbers. After the key message, which only references the median, you can find the model-specific results in the figures, a discussion of the methodology and the restrictions of the analysis at the end of the article.
### Key messages
* Without human-made greenhouse gas emissions, 0% of Albania’s land area and 0% of Albania’s population would be affected by droughts each year, on average.
* At today’s levels of 1 degrees Celsius of global warming the land area affected is 1.0 times as much: 0% of the total land area, while the number of people affected is 1.0 times as much: 0% of the total population.
* At 2 degrees Celsius of global warming, Albania’s land area affected by droughts is projected to increase by a factor of 0 compared to a world without human-made greenhouse gas emissions, to 0%. Likewise, Albania’s population exposed to droughts is projected to increase by a factor 0, to 0%.
* Following the higher-emissions scenario (RCP6.0) which can entail over 3 degrees Celsius of global warming by the end of the century (2081-2100), this factor is projected to reach 1.0 for the land area affected to droughts (to 1.0%) and 1.0 for the population exposed to droughts (to 0%).
* Albania ranks (ranking-value: land-rel-temp_ALB value: position temperature:2) with regards to its relative change in land area affected by droughts at 2 degrees Celsius of global warming in comparison to a situation without anthropogenic climate change. For the relative change in population exposed to droughts, Albania’s ranking is (ranking-value: pop-rel-temp_ALB value: position temperature:2).
### Results
The figures below shows the relative change in Albania’s land area affected by and population exposed to droughts in comparison to the time period before anthropogenic climate change. Results are shown both for the (link: glossary/historical-period text: historical time period) and for the (link: glossary/future-projections text: future projections) and can be visualized with regards to their change over time as well as with regards to their change in terms of global warming. Under the ‘change in terms of time’ setting, the change in land area affected by and population exposed to droughts over time is shown for 20-year (link: glossary/time-slices text: time-slices) from 1861-1880 until the end-of-the-century (2081-2100). Under the ‘change in terms of global warming’, the change in land area affected by and population exposed to droughts is shown for 20-year (link: glossary/time-slices text time-slices) representing increasing levels of global warming from 0 degrees Celsius up to 3 degrees Celsius compared to the time period before anthropogenic climate change started.
* Albania ranks (ranking-value: land-abs-temp_ALB value: position temperature:2) with regards to absolute changes in land area affected by droughts (expressed as % of Albania’s land area) at 2°C of global warming in comparison to a situation without climate change. For the absolute changes in population exposed to droughts (expressed as % of Albania’s population), Albania ranks (ranking-value: pop-abs-temp_ALB value: position temperature:2).
* At today’s levels of 1°C of global warming the simulated land area affected is already 100 km<sup>2</sup> larger (0.0% of the national land area) than in a world without climate change where the annual area affected by droughts is 0 km<sup>2</sup> (0% of Albania’s land area). The number of people exposed is 0.0 million (0.1% of the national population) larger than without climate change where the annual number of people exposed to droughts was 0.0 million (0% of Albania’s population).
* At 2°C of global warming the land area affected by droughts would increase by 200 km<sup>2</sup> (0.9% of the national land area) compared to a world without climate change, to 0.1% of the country’s land area. Assuming present-day population patterns, Albania’s population exposed to droughts would increase by 0.4 million, to 0.0% of the national population.
* Following the higher-emissions scenario (RCP6.0) which can entail over 3°C of global warming by the end of the century (2081-2100) ([Frieler et al. 2017](https://dx.doi.org/10.5194/gmd-10-4321-2017)) the land area affected by droughts would increase by 1300 km<sup>2</sup> (4.9% of the national land area) and reach 2.5% of the country’s land area. Assuming present-day population patterns the population exposed would reach 1.5% of Albania’s population, and increase by 2.6 million compared to a situation without climate change.
### How have we got the results?
The assessment provided below is based on our paper “Change in exposure to climate impact events under global warming” submitted to the journal “Earth’s Future”. We have used simulations of 1 different (link: glossary/global-hydrological-models text: global hydrological models) to estimate changes in drought conditions assuming different levels of global mean temperature change. To this end the global hydrological models were forced by climate projections from 3 different global (link: glossary/climate-models text: climate models) following the ISIMIP2b scenario set-up. In the key messages we report the results for the median of the (link: glossary/model-ensemble text: model ensemble) of global hydrological models that ran simulations in (link: glossary/isimip2b text: ISIMIP2b). The median represents the middle of the ensemble, meaning that 50% of the ensemble members provide higher numbers and 50% provide lower numbers.
Among the multiple concepts of “drought”, our study refers to a situation where monthly soil moisture (agricultural drought concept) is extremely low for at least 7 consecutive months. “Extremely low” means dry conditions occurring less than 5 times in 200 years under reference conditions without climate change. We were interested in changes in the land area affected by droughts and the number of people exposed to droughts that we have to expect at different (link: glossary/levels-of-global-warming text: levels of global warming) (1°C, 2°C, and 3°C) or during different future time periods.
The ISIMIP2b simulations start in 1860 (before human greenhouse gas emissions started to change the climate) and end in 2100. They comprise a long run where weather only varies according to pre-industrial conditions without climate change. In addition, modellers have run two different future scenarios: one “business as usual” scenario (RCP6.0) reaching high levels of warming and one “low emissions scenario” (RCP2.6) with reduced levels of climate change.
As the “land area affected by drought” varies quite strongly from year to year we decided to average numbers over multiple years: for each year we calculate the deviation of the land area affected by droughts from the average reference level without climate change. These deviations are then averaged over a present-day period (2001-2020), a mid-century period (2041-2060) and an end-of-century period (2081-2100). To calculate the changes at different levels of global mean temperature change we do not average over years belonging to one of these periods but over all years with global mean warming levels close to 1°C, 2°C, and 3°C.
When calculating the “number of people exposed to droughts” we proceed in a similar way. We assume that only the rural population is exposed while the urban population does not necessarily “feel” any drought. For 1860 to 2005 we account for changes in population patterns ([Goldewijk et al., 2017](https://dx.doi.org/10.5194/essd-9-927-2017)). Afterwards the population data is considered constant -- not as a realistic assumption but an informative “thought experiment”: What would future climate change mean for present day societies? For all time periods, the pure effect of climate change on the “number of people exposed to droughts” is estimated by comparing the number of people exposed under climate change and a specific population pattern to the number of people exposed assuming the same population patterns but no climate change.
In the following we describe what we have found and how strongly Albania is affected compared to other countries.
### What we have found
### Land area affected by droughts
(line-plot: land-abs-temp_ALB,land-abs-time_ALB first-temperature: 2 second-scenario: rcp26 second-time: 2041-2060)
Our definition of “drought” is quite strict, such that, without climate change, only 0% of Albania’s land area would be affected by droughts each year, on average.
However, at today’s level of 1°C global warming Albania’s annual land area affected by droughts is, on average, already larger and amount to 100 km<sup>2</sup> (0.2% of the national land area). The level of change ranges from 0.0% to 7.0% for the individual combinations of global hydrological models and global climate models. At 2°C of global warming, Albania’s annual land area affected by droughts is projected to increase by 200 km<sup>2</sup> (i.e. 0.9% of the national land area) on average in comparison to a world without climate change. Under these conditions, 0.1% of the national land area would be affected by droughts each year, on average. Across the individual combinations of global hydrological models and global climate models this expected level of change ranges from 0 to 1900 km<sup>2</sup>.
Following the higher-emissions scenario (RCP6.0) the land area affected by droughts is expected to increase by 300 km<sup>2</sup> (1.2% of the national land area) towards the end of the century (2081-2100). Following the low emission scenario (RCP2.6) the change would only reach 100 km<sup>2</sup> (1.2% of the national land area). By the middle of the century, changes reach 100 km<sup>2</sup> under RCP2.6 and 0 km<sup>2</sup> under RCP6.0.
Albania is the (ranking-value: land-abs-temp_ALB value: position temperature:2) strongest affected by droughts at 2°C of global warming. For the absolute change in land area affected by droughts towards the end of the century under a high-emissions scenario (RCP6.0), Albania’s ranking is (ranking-value: land-abs-time_ALB value: position time:2081-2100 scenario:rcp60).
### Population exposed to droughts
(line-plot: pop-abs-temp_ALB,pop-abs-time_ALB first-temperature: 2 second-scenario: rcp26 second-time: 2041-2060)
Our definition of “drought” is quite strict, such that, without climate change, only 0.0% of Albania’s population would be exposed to droughts each year, on average.
However, at today’s level of 1°C of global warming Albania’s annual population exposed to droughts is, on average, already 0.0 million (i.e. 0.1% of the total population) higher than without climate change and amount to 0% of the total population. The level of change ranges from 0.0 % to 7.0% for the individual combinations of global hydrological models and global climate models. At 2°C of global warming, Albania’s annual population exposed to droughts is projected to increase by 0 million (i.e. 0.4% of the population) on average in comparison to a world without climate change. Under these conditions, 0.0% of the total population would be affected by droughts, on average. Across the individual combinations of global hydrological models and global climate models this expected level of change ranges from 0.0 up to 0.1 million people.
Following the higher-emissions scenario (RCP6.0) the population exposed to droughts is expected to increase by 0.0 million (0.6% of the total population) towards the end of the century (2081-2100). Following the low emission scenario (RCP2.6) the change would expose 0.0 million people (0.1% of the total population). By mid of the century changes expose 0.0 million people under RCP2.6 and 0.0 million people under RCP6.0.
Albania is the (ranking-value: pop-abs-temp_ALB value: position temperature:2) strongest affected by droughts at 2°C of global warming. For the absolute change in population exposed to droughts towards the end of the century under a higher-emissions scenario (RCP6.0), Albania’s ranking is (ranking-value: pop-abs-time_ALB value: position time:2081-2100 scenario:rcp60).
### How is soil moisture calculated?
In both cases, results are shown for all possible combinations of (link: glossary/global-hydrological-models text: global hydrological models) and (link: glossary/global-climate-models text: global climate models) (normal lines) as well as for the median of the model ensemble (thick line). The influence of (link: glossary/inter-annual-variability text: inter-annual variability) on the results is visualized by means of the shaded area in black. This inter-annual variability is only shown for the median model combination. For the future projections and under the ‘change in terms of time’ settings only, an additional distinction is made between the results of two future (link: glossary/emissions-trajectories text: emissions trajectories): a low-emissions trajectory limiting global mean temperature in the 21st century to below 2 degrees Celsius compared to before anthropogenic climate change started, called RCP2.6 (blue lines), and a higher-emissions scenario leading global mean temperature to above 3 degrees Celsius by 2100, compared to pre-climate change conditions, called RCP6.0 (red lines). The shaded areas in blue and red indicate the variety (or spread) in results between all combinations of global hydrological models and global climate models for these two emissions trajectories. A filtering menu top-right of the graph allows the user to select individual emissions trajectories, individual global climate models, and individual global hydrological models for visualization.
Soil moisture is defined as the water stored in the soil in liquid or frozen form. It is calculated as the difference between the amount of water coming in as rain or snow, and the amount of water going out, either through evapotranspiration from the surface and vegetation, runoff, or through percolation of water towards the deeper groundwater layers. Global hydrological models (GHMs) include different representations of the (link: glossary/soil-water-column text: soil water column) with different numbers of soil water layer(s) (between 1 and 15 layers), and with different total depth of soil layers (between 1 and 42.1 meter).
Here, we use (link: glossary/rootzone-soil-moisture text: rootzone soil moisture) estimates, the portion of soil moisture that is within the rooting depth of plants, when directly provided by the GHMs. For GHMs that do not directly provide root zone soil moisture conditions, we approximated this variable by integrating soil moisture across multiple soil water layers in order to reach a depth of ~1 meter. Daily rootzone soil moisture values are finally aggregated into monthly average rootzone soil moisture conditions per grid cell.
By hovering over or clicking on a particular value in the figure additional details, such as the specific global climate model or hydrological used, behind the presented value become available. The visualization below the graph shows, for the selected model run under this time-period or temperature-change level, how Albania ranks in comparison to other countries on its relative change in land area affected by or population exposed to droughts.
#### Land area affected by droughts
### Where are the most relevant gaps in our knowledge?
Models are simplified representations of reality, hence model simulations come with limitations and uncertainties that have to be kept in mind ([Döll et al. 2016](https://dx.doi.org/10.1007/s10712-015-9343-1)). Regarding our drought simulations we mainly see the following issues.
(line-plot: land-rel-temp_ALB,land-rel-time_ALB first-temperature: 2 second-scenario: rcp26 second-time: 2041-2060)
Without human-made greenhouse gas emissions, 0% of Albania’s land area would be affected by droughts each year, on average.
* Representation of direct human influences
Under the ‘change in terms of global warming’ setting the figure shows that at today’s level of 1 degrees Celsius of global warming Albania’s land area affected by droughts is, on average, already 1.0 times (or 0%) lower and amount to 0% of the total land area. This level of change ranges from 0% to 0.306% for the individual combinations of global hydrological models and global climate models. At 2 degrees Celsius of global warming, Albania’s land area affected by droughts is projected to change by a factor of 1.0 (or 0.144%) in comparison to a world without human-made greenhouse gas emissions. Under these conditions, 0.144% of the total land area would be affected by droughts on a yearly basis, on average. Across the individual combinations of global hydrological models and global climate models this expected level of change ranges from 0% up to 0.306%.
When presenting the ‘change in terms of time’, we find that when following the higher-emissions scenario (RCP6.0) towards the end-of-the-century (equivalent to over 3 degrees Celsius change) would result in a change in the land area affected by droughts of, on average, a factor 1.0 (0%), towards: 0% of the total land area. Following the (link: glossary/climate-mitigation-emissions-scenario text: climate-mitigation emissions scenario)(RCP2.6) towards the end-of-the-century (entailing an average 2.5 degrees Celsius change) would result in a foreseen change in the land area affected by droughts of, on average, a factor 1.0 (0%), 0% of the total land area being affected.
The presented drought projections reflect the isolated effect of climate change, while their local manifestations are expected to also be influenced by direct human drivers like changes in land use, land cover, and irrigation. Even in the historical period our simulations only account for water abstractions (with varying sectors among the models like irrigation, industry, domestic, livestock, desalinization) from groundwater and surface waters, reservoir management while other processes such as inter-basin water transfers and a more realistic reservoir management are also expected to be important. Including these drivers into our simulation would greatly improve the simulations ([Veldkamp et al. 2018](https://doi.org/10.1088/1748-9326/aab96f)). That it is not done is mainly an issue of missing knowledge about them.
Globally, the land area affected by droughts each year, on average, is 2.377% in a situation without human-made greenhouse gas emissions. At 2 degrees Celsius global warming or by the end-of-the-century under a higher-emissions scenario (RCP6.0) these values are foreseen to increase globally by a factor 3.106 and 7.014%, respectively.