Different parts of world are experiencing extreme hydrological hazards such as droughts, flooding and other related events. Droughts are associated with absence of rainfall occurrence over an extended period. According to the United Nations (2022), the frequency and intensity of drought events in the last two decades has increased by 29%. These figures are expected to increase further in the coming years due to climate change (Gunathilake et al., 2020). 

Recent drought forecast estimates show that by 2050 over 75% of the world’s population may be affected by droughts (United Nations, 2022). This further emphasizes that droughts are continuously a hazard of concern for many regions, especially arid and semi-arid regions.  Droughts have significant socio-economic impacts on vulnerable communities in both rural and urban areas due to the lack of preparedness and awareness of the dire conditions that water resources are under. The resulting dry conditions caused serious water stress especially for the agricultural sector resulting in drastic reduction in crop harvest and livestock production contributing to further malnutrition and food insecurity. In addition, droughts can increase biodiversity loss, poverty, and loss of human life (Wade, 1974; Abbaspour et al., 2009).

Types of droughts

Drought monitoring is essential for drought management, policy guidance, and tackling other drought-related challenges. To effectively monitor droughts, it is important to be able to distinguish between types of droughts. Droughts are mainly classified into four types: meteorological, agricultural, hydrological, and socioeconomic droughts. 

  1. Meteorological drought is associated with degree or duration of dry weather, which is related absence of rainfall.  Rainfall intensity and duration are therefore dominant factors that are compared to normal or average conditions. For example, meteorological drought occurs when rainfall falls below a specific threshold over a given period, e.g., 15 days.  However, the threshold used to define this type of drought is not applicable in all instances since it depends on a specific region, as atmospheric conditions vary depending on the location.
  2. Agricultural drought happens when drought affects soil moisture, which reduces available water for crops. The first impact of drought is on soil moisture, as a result, available water for plant will be affected and trigger agriculture drought. This drought account for susceptibility of crop at various growth stages. At early growth stage, crops require less water compared to the latter stages. When the soil moisture is not sufficient at latter growth stages, crops yield will be reduced.    
  3. Hydrological drought is mainly due to effect of rainfall shortfall in contrast to meteorological drought that is associated with rainfall absence or agricultural drought due to soil moisture deficit. Prolonged drought will impact surface and sub surface water supplies (streamflow, reservoir and lake levels, groundwater), which occurs after many months of meteorological drought. The severity and frequency of this drought is defined by discharge at river basin scale. Usually, dry years have discharge that fall below average of long-term records. In addition, hydrological drought takes long time to recover because it takes long time to raise reservoir water level/increase ground water level. On the other hand, agricultural drought will quickly recover as the precipitation falls and soil moisture becomes available. 
  4. The socioeconomic drought relates the supply and demand of various commodities to drought. When demand (e.g., per capita water use) exceeds available supplies (e.g., reservoirs or other water resources) cause another drought which is associated with human interventions or as the consequences of such interventions (Maciej Serda et al., 2013). 

Detecting and characterizing meteorological droughts

Given the limited data availability particularly in the Global South, satellite and remotely sensed data are frequently utilized for various applications. From these data sets several indices can be derived for varying applications such as drought. For example, Standardized Precipitation Index (SPI) often used for meteorological drought assessments, Normalized Difference Vegetation Index (NDVI) and Standardized Vegetation Index (SVI) are applied to assess agricultural droughts, and Soil Moisture Deficit Index (SMDI) is used for hydrological drought analysis (Narasimhan and Srinivasan 2005; Musolino, Massarutto, and de Carli 2015). 

Standardized Precipitation Index

SPI is the most used indicator worldwide for detecting and characterizing meteorological droughts on a range of time series and is recommended by World Meteorological Organization for drought studies (Svoboda et al., 2012). The SPI equals the deviation of precipitation from the mean, therefore, negative SPI values indicate dry conditions (i.e., drought events), whereas positive SPI values indicate wet conditions (i.e., flooding events). SPI is calculated monthly for a moving window of rainfall accumulation periods depending on the desired study outcome. The rainfall accumulation periods are typically one, three, six, nine, 12 or 48 months. Time periods ranging from one to three months are usually used for meteorological studies, three to six months for agriculture and six to 12 months for hydrological drought studies, respectively (Svoboda et al., 2012). 

SPI is a powerful drought index that only requires precipitation as input. However, it cannot account for evapotranspiration which is also an important factor, especially when using long-term rainfall and temperature time series to assess climate impacts. Due to this limitation, SPI is not recommended to assess drought under changing climate, which requires >30-year time period. A longer time series on other hand, is important for SPI derivation. Because longer records capture different weather conditions (droughts) which are not identifiable within a smaller time span.  

Climate Hazards Group InfraRed Precipitation with Station data

The Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) is a 35+ year quasi-global rainfall dataset developed by scientists at the University of California, Santa Barbara and the U.S. Geological Survey Earth Resources Observation and Science Center under the direction of Famine Early Warning Systems Network (FEWS NET). It provides near-real time precipitation from 1981 to the present (Shukla et al., 2017). The geographic extent of datasets is 50°S-50°N and 180°W-180°E.  Data is available at monthly or daily time series and in two sets of spatial resolutions i.e., 0.25° × 0.25° and 0.05° × 0.05° (Shukla et al., 2017).

The CHIRPS dataset is developed based on a blend of three data sources: (i) the Climate Hazards Precipitation Climatology (CHPclim), a global precipitation climatology at 0.05° latitude and longitude resolution (estimated for each month based on station data, averaged satellite observations, elevation, latitude and longitude); (ii) quasi-global geostationary Thermal Infrared Radiation (TIR) satellite observations, TMPA 3B42 product, and (iii) atmospheric model precipitation fields from the National Oceanic and Atmospheric Administration (NOAA) Climate Forecast System (CFS) version 2.0 (Funk et al., 2015).

CHIRPS is preferred over many other satellite precipitation datasets since it has low latency, high resolution, low bias, and a long period of records (Funk et al., 2015). It has been widely applied in various assessments such as drought impact assessment due to its feasibility in the absence of gauged data and climate model downscaling (Shukla et al., 2017; Dinh, 2020; Hordofa et al., 2021). In addition, applications of CHIRPS in Africa have outperformed other precipitation data derived from space technology such as Global Precipitation Measurement (GPM) and Agrometeorological Indicators from ERA5 (AgERA5) (Dinh, 2020; Hordofa et al., 2021).  

Case study: Real-time droughts monitoring in Harirud River Basin, Afghanistan

A real-time drought monitoring study aimed at evaluating meteorological drought impacts in both spatial and temporal dimensions using SPI derived from Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) which is readily available at the University of California, Santa Barbara website was conducted in Harirud River Basin, Afghanistan. 

The study implements a one-month SPI (SPI 1) to assess meteorological drought. SPI values are normalized using Pearson Type III distribution. Since precipitation is not a normally distributed variable, it is recommended to apply either Gamma or Pearson Type III distribution. In addition, Guttman (1999) suggests a Pearson Type III distribution over the Gamma distribution. For the sake of this case study, the below described calculations are performed in Python. To calculate the SPI, follow the steps below:

  1. The necessary Python packages and libraries for this calculation can be seen in Figure 1. Import these packages and libraries. Note that xarray, geopandas, and regionmask  require installation before importing. To install a package or library, use the ‘pip install’ command in the Command Prompt in Windows or the Terminal on Mac OS. 
    Figure 1: Importing required python packages for this study
    Figure 1: Importing required python packages for this study


  2. Slice the global CHIRPS month precipitation to the approximate region of interest and time frame to reduce the data size and change the attributes of the data to the format of the climate-indices package. This package requires data to be in Network Common Data Form (NetCDF) format. NetCDF is a set of software libraries and machine-independent data formats that support the creation, access, and sharing of array-oriented scientific data.
    Figure 2: Preprocessing CHIRPS precipitation and formatting according to climate-indices package
    Figure 2: Preprocessing CHIRPS precipitation and formatting according to climate-indices package


  3. Climate-indices packages must be installed in an anaconda environment. Anaconda supports multiple versions of Python and associated packages. An environment within anaconda generally includes one version of Python or R language and some packages. Once climate-indices package is installed in your anaconda environment, change directory to the location where data from step two is saved. Then type the following script in your Command Prompt/Terminal to calculate the SPI (Pearson and gamma). It will save SPI dataset in NetCDF format.
    Figure 3: Script to calculate SPI using climate-indices
    Figure 3: Script to calculate SPI using climate-indices 


  4. The output of step three can be further processed, the clipping of xarray dataset. The xarray dataset resembles an in-memory representation of a NetCDF file, and consists of variables, coordinates and attributes which together form a self-describing dataset. Here, the SPI values are clipped to basin boundary (Harirud Basin) and other relevant time series calculations could be performed such as seasonal and monthly distribution of SPI over this region.
    Figure 4:Clipping an xarray dataset (NetCDF data) to a shapefile boundary
    Figure 4: Clipping an xarray dataset (NetCDF data) to a shapefile boundary


Figure 5: SPI Temporal variation from 1981 to 2021 – monthly time series grouped under each season – Harirud basin, Afghanistan.
Figure 5: SPI Temporal variation from 1981 to 2021 – monthly time series grouped under each season – Harirud Basin, Afghanistan.


In this study, SPI was evaluated from 1981 to 2010 using CHIRPS-v2 at a monthly time series. The SPI is evaluated temporally for each season (Figure 5). Winter and summer SPI decrease at rate of 0.0002% and 0.001% respectively. While spring and autumn respectively increase at rate of 0.007% and 0.003%, based on a linear slope trend line. The summer shows a higher precipitation decrease which means an increasing drought risk. However, an opposite trend is revealed in spring, which can pose a high flood risk. In addition, all seasons show a significant fluctuation range in recent years. This implies that recent years are experiencing more severe events (increased flood and drought), especially in summer. The order of seasons according to their dryness (driest to wettest) is autumn, spring, summer and winter, with a respective mean SPI of -0.04, -0.01, 0, and 0.01 over the given time span.

The time series in Figure 5 was derived by averaging SPI values along longitude and latitude dimensions. Then each season was filtered from the resulting time series as shown below. 

Figure 6: Extracting seasonal SPI from an xarray dataset
Figure 6: Extracting seasonal SPI from an xarray dataset


The spatial distribution of SPI is also assessed under different time scales, seasonal and monthly (Figures 8 and 9). Seasonal analysis (Figure 8) suggests autumn as the driest season while summer as the wettest season. Spatial distribution indicates uniform variation for winter and spring (SPI ranges between -0.04 to 0.04). In contrast, summer and autumn show high spatial variation, meaning variable rainfall pattern across different location. In general, the southwest regions of basin show lower SPI values, especially during summer and autumn indicating higher drought risk. 

The analysis was performed by grouping monthly values under each season and month and then averaged them over 1981-2010, as shown in following script (Figure 7). 

Figure 7: Calculating Mean SPI over each season
Figure 7: Calculating Mean SPI over each season


Figure 8: SPI Spatial variation for each season, averaged over 1981-2021 – Harirud basin, Afghanistan.
Figure 8: SPI Spatial variation for each season, averaged over 1981-2021 – Harirud Basin, Afghanistan. 

The color bar shows SPI ranges; dark blue (SPI< -0.16), blue (-0.12>SPI>-0.16), light blue (-0.08>SPI>-0.12), green and yellow (0.08>SPI>-0.08), light red (0.12>SPI>0.08), red (0.16>SPI>0.12), and dark red (SPI>0.16).


Monthly analysis (Figure 9) indicates October as the driest month while June and July as the wettest months. During July, SPI goes above 0 for most areas. Higher SPI range is noticeable in the northwest region (> 0.16). The same area is therefore subject to high flood risk. Conversely, during October SPI falls below 0. The south of the basin is highly susceptible to drought given the lower SPI range (< -0.16). 

SPI has higher spatial fluctuation from June to October. Over this period, different parts of the basin experience varied weather condition compared to other months, which show a uniform distribution. However, July and October show distinguishing characteristics that make them subject of interest for flood and drought analysis respectively. 

Figure 9: SPI spatial variation for each month, averaged over 1981-2021 – Harirud basin, Afghanistan.
Figure 9: SPI spatial variation for each month, averaged over 1981-2021 – Harirud Basin, Afghanistan. 

The color bar shows SPI ranges; dark blue (SPI< -0.32), blue (-0.24>SPI>-0.32), light blue (-0.16>SPI>-0.24), green and yellow (0.08>SPI>-0.08), gray (0.16>SPI>0.08), light red (0.24>SPI>0.16), red (0.32>SPI>0.24), and dark red (SPI>0.32).



The study at hand used an SPI index to evaluate spatiotemporal characteristics of drought. SPI computation is often a challenging task, especially when applying complicated probability distributions such as Gamma/Pearson. This study therefore shows an easy-to-use method that combines open-source Python packages to calculate SPI with different probability distributions using gridded datasets. A detailed guide is also provided on how to implement this procedure. To evaluate applicability of this method, the study involves a case study conducted over Harirud Basin, situated in Afghanistan’s west. 

In the case study, a 1-month SPI is calculated from CHIRPS monthly gridded rainfall over 1981-2021. Seasonal time series showed autumn as the driest while summer as the wettest season. Moreover, SPI had increasing trend during spring and autumn and decreasing trend during winter and summer. Spring had slightly higher increasing trend compared to other seasons. Since this season is moderately wet, this could cause flood. On the other hand, Summer had slightly higher decrease compared to other seasons, which means reduced water availability and drought risk. 

The monthly analysis revealed October as the driest month, while July was the wettest. The spatial analysis generally identified the southwest region of basin as the driest. Given the climate change impact on extreme events, drought and flood intensity is likely to become more severe in these months (October and July) and seasons (summer and spring). Therefore, it is recommended to explore climate change impact on these months and seasons to understand future trends. 

The above methodology can serve as a useful means for stakeholders to understand how drought events have evolved in the past. Disaster management planners and other involved stakeholders can therefore better prepare for extreme weather events and make informed decisions regarding drought preparedness, mitigation, response as well as recovery measures in order to reduce drought impacts. In addition, the study’s results can serve as an essential guidance for high-level decision-making and policy in the long run.


Abbaspour, Karim C., Monireh Faramarzi, Samaneh Seyed Ghasemi, and Hong Yang. 2009. “Assessing the Impact of Climate Change on Water Resources in Iran.” Water Resources Research 45 (10). https://doi.org/10.1029/2008WR007615.

Dinh, Ngoc Thuy Vy. 2020. “(2) (PDF) Evaluation and Comparison of Satellite-Based Rainfall Product CHIRPS and Reanalysis Product ERA5 in West Africa.” 2020. https://www.researchgate.net/publication/351334929_Evaluation_and_compa….
Funk, Chris, Pete Peterson, Martin Landsfeld, Diego Pedreros, James Verdin, Shraddhanand Shukla, Gregory Husak, et al. 2015. “The Climate Hazards Infrared Precipitation with Stations—a New Environmental Record for Monitoring Extremes.” Scientific Data 2015 2:1 2 (1): 1–21. https://doi.org/10.1038/sdata.2015.66.

Gunathilake, Miyuru B., Yasasna v. Amaratunga, Anushka Perera, Imiya M. Chathuranika, Anura S. Gunathilake, and Upaka Rathnayake. 2020. “Evaluation of Future Climate and Potential Impact on Streamflow in the Upper Nan River Basin of Northern Thailand.” Advances in Meteorology 2020. https://doi.org/10.1155/2020/8881118.

Guttman, Nathaniel B. 1999. “ACCEPTING THE STANDARDIZED PRECIPITATION INDEX: A CALCULATION ALGORITHM1.” JAWRA Journal of the American Water Resources Association 35 (2): 311–22. https://doi.org/10.1111/J.1752-1688.1999.TB03592.X.

Hordofa, Aster Tesfaye, Olkeba Tolessa Leta, Tena Alamirew, Nafyad Serre Kawo, and Abebe Demissie Chukalla. 2021. “Performance Evaluation and Comparison of Satellite-Derived Rainfall Datasets over the Ziway Lake Basin, Ethiopia.” Climate 2021, Vol. 9, Page 113 9 (7): 113. https://doi.org/10.3390/CLI9070113.

Maciej Serda, Fernando Gertum Becker, Michelle Cleary, R M Team, Helge Holtermann, Disclaimer The, National Agenda, et al. 2013. “Synteza i Aktywność Biologiczna Nowych Analogów Tiosemikarbazonowych Chelatorów Żelaza.” Edited by G. Balint, B. Antala, C. Carty, J-M. A. Mabieme, I. B. Amar, and A. Kaplanova. Uniwersytet Śląski 7 (1): 343–54. https://doi.org/10.2/JQUERY.MIN.JS.

Musolino, Dario, Antonio Massarutto, and Alessandro de Carli. 2015. “Ex-Post Evaluation of the Socio-Economic Impacts of Drought in Some Areas in Europe.” Drought: Research and Science-Policy Interfacing - Proceedings of the International Conference on Drought: Research and Science-Policy Interfacing, February, 71–78. https://doi.org/10.1201/B18077-13/DROUGHT-EFFECTS-RAINFED-AGRICULTURE-U….

Narasimhan, B., and R. Srinivasan. 2005. “Development and Evaluation of Soil Moisture Deficit Index (SMDI) and Evapotranspiration Deficit Index (ETDI) for Agricultural Drought Monitoring.” Agricultural and Forest Meteorology 133 (1–4): 69–88. https://doi.org/10.1016/J.AGRFORMET.2005.07.012.

Shukla, Shraddhanand, Chris Funk, Pete Peterson, Amy Mcnally, Tufa Dinku, Humberto Barbosa, Franklin Paredes-Trejo, Diego Pedreros, and Greg Husak. 2017. “The Climate Hazards Group InfraRed Precipitation with Stations (CHIRPS) Dataset and Its Applications in Drought Risk Management.” Geophysical Research Abstracts 19: 2017–11498.

Svoboda, Mark, Mark Svoboda, Michael Hayes, Deborah A. Wood, and World Meteorological Organization (WMO). 2012. Standardized Precipitation Index User Guide. WMO-No. 1090 ©. WMO. Geneva: WMO. http://library.wmo.int/opac/index.php?lvl=notice_display&id=13682.

United Nations. 2022. “World Day to Combat Desertification and Drought.” 2022. https://www.un.org/en/observances/desertification-day.

Wade, Nicholas. 1974. “Sahelian Drought: No Victory for Western Aid.” Science 185 (4147): 234–37. https://doi.org/10.1126/SCIENCE.185.4147.234/ASSET/9556252E-2FF1-439B-9….