Global Drought Experimental Dataset Development and Analysis using Multi-Scenario Fusion Methods (2015–2100)

CUI Jingwen^1,2  LIU Yangxiaoyue^1*  XI Yongshi³  ZHAO Guanfeng⁴

1. Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, China;

2. University of Chinese Academy of Sciences, Beijing 100049, China;

3. Xizang Autonomous Region Agro-products Quality and Safety Inspection and Testing Center, Lhasa 850000, China;

4. Xizang Autonomous Region Agricultural Technology Extension Service Center, Lhasa 850000, China

Abstract: Global climate change is intensifying droughts in terms of frequency, spatial coverage, and severity, making them a prominent environmental challenge that threatens the stability of natural ecosystems and human society. To establish a reliable global multi-scenario future drought prediction dataset, this study adopts 18 Earth system models from the Coupled Model Intercomparison Project Phase 6 (CMIP6), selecting 4 core climate variables — precipitation (mm/month), moisture in the upper portion of the soil column (% volumetric), near-surface air temperature (℃), and potential evapotranspiration (PET, mm/month). Based on unified temporal and spatial resolutions, 4 representative drought indices were calculated: the Standardized Precipitation Evapotranspiration Index (SPEI), the Drought Severity Index (DSI), the Temperature Condition Index (TCI), and the Standardized Soil Moisture Index (SSMI). By validating the predictive performance of each index against 5 major historical drought events, DSI and SSMI were found to perform best in characterizing drought intensity and its spatial distribution, and could reasonably reproduce the spatiotemporal evolution of future droughts. In terms of numerical interpretation, lower SPEI, SSMI, and DSI values indicate more severe drought, with values below −1.0, −1.5, and −2.0 indicating at least moderate drought, at least severe drought, and extreme drought, respectively. Lower TCI values indicate greater heat stress, with TCI < 50% indicating high heat stress. The dataset covers monthly values of the four drought indices under the SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5 scenarios from 2015 to 2100, at a spatial resolution of 0.5°, archived in .tif format, and consists of 16,512 data files with data size of 7.45 GB (compressed into 8 files, 4.86 GB).

Keywords: drought prediction; multi-scenario projections; drought indices; global; data fusion

DOI: https://doi.org/10.3974/geodp.2026.05.05

Dataset Availability Statement:

The dataset supporting this paper was published and is accessible through the Digital Journal of Global Change Data Repository at: https://doi.org/10.3974/geodb.2026.05.02.V1.

1 Introduction

As a typical natural hazard, drought has shown trends of increasing frequency, expanding extent and intensifying severity under global climate change, and has become an important climatic risk factor affecting ecosystem health, agricultural production, water resource security and the sustainable development of regional socio-economic systems^[1–3]. With the intensification of global warming, the continued rise in air temperature interacts with chan­g­es in the spatiotemporal patterns of precipitation, making drought events more complex and uncertain and imposing higher demands on the scientific assessment and accurate moni-  t­oring of drought^[4].

To characterize the evolution of drought and its multidimensional impacts more comprehensively, the academic community has developed a variety of drought indices, which are widely used in drought identification, monitoring and risk assessment. Among these, the Standardized Precipitation Evapotranspiration Index (SPEI), the Drought Severity Index (DSI), the Temperature Condition Index (TCI) and the Standardized Soil Moisture Index (SSMI) are representative. SPEI integrates the effects of precipitation and potential evapotranspiration and is therefore more sensitive to the influence of temperature change on the water balance^[5]. SSMI mainly assesses soil moisture deficits and is commonly regarded as an important indicator of agricultural drought^[6]. TCI is based on remotely sensed land surface temperature (LST) data and assesses vegetation thermal stress, making it applicable to large-scale and high-frequency drought monitoring^[7]. DSI, in turn, can more compre-  he­nsi­vely reflect the ecological and hydrological responses during drought events. Each of these four indices has its own strengths, and their combined use can improve the timeliness and accuracy of drought identification and enhance the capacity to respond to different drought types.

In recent years, the Coupled Model Intercomparison Project Phase 6 (CMIP6) has provided abundant climate simulation data for drought monitoring research^[8,9]. Compared with traditional observational and reanalysis datasets (e.g., ERA5, CRU and GPCC), CMIP6 offers advantages such as broader global coverage, longer temporal span, richer variables and scientifically designed scenarios, and can provide climate projections under a range of socio-economic development pathways. The Shared Socioeconomic Pathways in CMIP6 (SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5) cover a range of future scenarios from low to high emissions and are suitable for analyzing drought risk under multi-scenario comparison.

Although drought indices and CMIP6 data have been widely applied in various studies, a global drought monitoring dataset that systematically integrates multi-source drought indices with CMIP6 multi-scenario climate data is still lacking. To some extent, this has limited cross-comparison among drought indices, risk assessment across different regions, and multi-perspective analysis and integrated interpretation of drought trends under future climate change scenarios. Therefore, based on the four drought indices SPEI, DSI, TCI and SSMI, and combining CMIP6 historical simulations with future climate scenario data, this study constructs a dataset with global coverage, multiple spatiotemporal scales and a unified processing workflow, aiming to provide a high-quality basic data product for research on the spatiotemporal evolution of drought, risk assessment and decision support under climate change. The release of this dataset will provide data support for global drought monitoring research and will also serve as an important reference for research on extreme climate responses and for the development of regional climate adaptation strategies.

2 Metadata of the Dataset

Table 1 summarizes the metadata for Global drought experimental dataset using multi-sce­nario fusion methods (2015–2100)^[10]. It includes the dataset title, authors, geographic coverage, year of the dataset, temporal resolution, spatial resolution, dataset files, etc.

Table 1  Metadata summary for Global drought experimental dataset using multi-scenario fusion methods (2015–2100)

3 Methods

3.1 Data Sources

Under the four Shared Socioeconomic Pathway scenarios (SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5), this study selected 4 key variables from 18 CMIP6 models for drought index calculation and distinguished the model sources by variable. Precipitation (mm/month), moisture in the upper portion of the soil column (% volumetric) and near-surface air temperature (℃) were provided by the following 16 models: ACCESS-CM2 (Commonwealth Scientific and Industrial Research Organization, Australia), BCC-CSM2- MR (Beijing Climate Center, China Meteorological Administration), CAMS-CSM1-0 (Chinese Academy of Meteorological Sciences), CanESM5-CanOE (Canadian Centre for Climate Modelling and Analysis), CESM2 (National Center for Atmospheric Research, USA), CMCC-CM2-SR5 (Centro Euro-Mediterraneo sui Cambiamenti Climatici, Italy), CNRM-CM6-1-HR (Centre National de Recherches Météorologiques, France), CNRM- ESM2-1 (Centre National de Recherches Météorologiques, France), KACE-1-0-G (National Institute of Meteorological Sciences, Korea Meteorological Administration), MIROC6 (Japan Agency for Marine-Earth Science and Technology), MIROC-ES2L (Japan Agency for Marine-Earth Science and Technology), MPI-ESM1-2-LR (Max-Planck-Institut für Meteorologie, Germany), MRI-ESM2-0 (Meteorological Research Institute, Japan), NorESM2-MM (Norwegian Climate Center, Norway), TaiESM1 (Academia Sinica, Taiwan, China) and UKESM1-0-LL (Met Office Hadley Centre, UK); potential evapotranspiration (PET, mm/month) was provided by CNRM-CM6-1 (Centre National de Recherches Météorologiques, France), IPSL-CM6A-LR (Institut Pierre-Simon Laplace, France) and CNRM-ESM2-1 (Centre National de Recherches Météorologiques, France). All datasets were harmonized to a spatial resolution of 0.5°×0.5° and a monthly temporal resolution to ensure comparability across models and variables.

3.2 Technical Workflow

This study aims to develop a stable and reliable CMIP6-based global drought projection data product. First, all gridded datasets were converted to TIFF format and standardized to a spatial resolution of 0.5°×0.5° and a monthly temporal resolution. To improve the robustness of each variable, pixel-wise ensemble means were calculated across the available climate models, subject to a minimum threshold for the number of valid models. Specifically, at least 8 valid models were required for variables available from 16 models, while at least 2 valid models were required for variables available from 3 models. Pixels that did not meet the corresponding threshold were assigned a NoData value of −9999. Second, 4 drought indices were calculated, and drought severity categories were defined according to the corresponding index values. Third, the performance of the four drought indices was evaluated against documented severe drought events to assess their ability to reproduce observed drought conditions.

3.3 Drought Index Calculation

This study calculated 4 drought indices, including SPEI, SSMI, TCI, and DSI.

3.3.1 SPEI

                                                                                                                                                  (1)

where S is the water balance value (mm/month), P is the precipitation (mm/month), PET is the potential evapotranspiration (mm/month).

                                                                                                                              (2)

where S_cumul represents the cumulative water balance over the selected time scale (mm/month), n represents the time scale (e.g., n=3 for 3 months), S_i is the water balance value for the i-th month (mm/month).

                                                                                                  (3)

where f(x) represents the probability density function of the cumulative water balance, x is the cumulative water balance value (mm/month), α, β, and γ are the scale parameter (mm/month), the shape parameter (dimensionless), and the location parameter (mm/month), respectively. These parameters were estimated using maximum likelihood.

                                                                                                                              (4)

where F(x) represents the cumulative probability for a given Scumul​ (dimensionless).

                                                                                                                                      (5)

where Φ⁻¹ is the inverse of the standard normal distribution (probit function), SPEI values follow a standard normal distribution with a mean of 0 and a standard deviation of 1. The SPEI values are classified to determine drought or wetness severity based on the following thresholds. SPEI values are interpreted according to the classification in Table 2.

3.3.2 SSMI

                                                           (6)

where X represents the soil moisture value at a specific time step (%), and μ refers to the mean of the entire time series (%), σ is the standard deviation of the entire time series (%).

                                                                                                                                                  (7)

where N is the total number of time steps (dimensionless), and Xᵢ is the soil moisture value at the i-th time step (%).

                                                                                                                          (8)

The SSMI values are interpreted according to Table 3.

3.3.3 TCI

TCI is a widely used drought indicator based on land surface temperature (LST), and is designed to quantify thermal stress levels on vegetation within specific regions. The TCI reflects the relative position of current surface temperature within its historical range, providing valuable insights into the impact of heat stress on vegetation.

                              (9)

where T_i refers to the LST (℃) at time i. T_max is the maximum LST during the study period, and T_min is the minimum LST (℃) during the same period.

The TCI values range from 0% to 100%, providing a direct measure of heat stress on vegetation. High TCI values (TCI≥50%): Indicate that the current temperature is closer to the historical minimum, suggesting low heat stress, which is beneficial for vegetation growth. Low TCI values (TCI<50%): Indicate that the current temperature is closer to the historical maximum, reflecting severe heat stress that may inhibit vegetation growth.

3.3.4 DSI

                                                                                                                                               (10)

where X is the water balance value (mm/month), P is the precipitation (mm/month), PET is the potential evapotranspiration (mm/month). This represents the net water available in the system after accounting for losses due to evapotranspiration.

                                                                                                                                              (11)

where X is the current value of water balance (mm/month). μ is the long-term mean of the variable (mm/month), and σ is the long-term standard deviation of the variable (mm/month). Table 4 presents the classification of drought and wetness severity based on DSI values.

                                                                                                                                                (12)

                                                                                                                        (13)

4 Data Results and Validation

4.1 Dataset Composition

The dataset includes monthly global drought data for 2015 to 2100 predicted under the SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5 scenarios using four drought indices. The spatial resolution is 0.5° and files are named as ssp126_yyyy.mm_DSI.tif.

4.2 Data Validation

This study examined 5 high-profile drought events in recent years: the severe drought in Europe in 2022, the prolonged drought in the Horn of Africa from 2020 to 2023, the “megadrought” in central Chile since 2010, the historic drought in the western United States in 2022, and the extreme drought in Southern China in 2022. Under the four Shared Socioeconomic Pathway (SSP) scenarios, we conducted a comprehensive analysis of these events using 4 drought indices: SPEI, TCI, DSI and SSMI. The results show marked differences among the indices. Compared with SPEI and TCI, DSI and SSMI perform better in identifying the severity and spatial extent of major droughts. Although SPEI can distinguish a spectrum of drought categories from normal to extreme, it often underestimates the actual intensity and affected area. TCI assesses heat stress using near-surface air temperature but does not account for key hydrometeorological factors such as precipitation and soil moisture. Its classification scheme is relatively simple, typically limited to “high” and “low” heat-stress levels, resulting in weaker performance for complex or compound droughts. By contrast, DSI and SSMI yield more stable and accurate results. In these cases, their assessments align closely with in situ observations and media reports, indicating stronger practical applicability. These findings underscore the importance of selecting appropriate drought indices in the context of climate change. Given their consistent performance across regions and emission scenarios, we recommend DSI and SSMI as core tools for global drought monitoring to enhance drought risk management and support the development of local response strategies.

4.3 Data Results Analysis

Using July 2030 as an example of the CMIP6 global drought projection data product, Figure 1 presents the global drought projections derived from different drought indices under the four CMIP6 Shared Socioeconomic Pathway (SSP) scenarios (SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5). As shown in Figure 1, for a given drought index, the global spatial distributions under the four SSP scenarios in July 2030 are broadly similar, although regional differences in drought severity are evident. By contrast, the spatial patterns differ markedly among the drought indices. DSI and SSMI display relatively distinct regional

Figure 1  Maps of the global drought prediction data of four drought indices under different scenarios (July 2030)

patterns, SPEI shows comparatively limited spatial variation, and TCI yields a more generalized distribution defined by high and low heat-stress categories. Overall, this example provides an intuitive illustration of the global spatial characteristics represented by the dataset across different scenarios and drought indices.

5 Discussion and Conclusion

Drought has become a pressing global climate and environmental challenge. Advancing drought research is essential for improving early-warning capabilities and safeguarding ecological and socioeconomic sustainability. Long-term, high-quality global drought data are critical for understanding drought patterns, and reliable future-scenario datasets are indispensable for accurately assessing its spatiotemporal evolution under climate change. From a data-fusion perspective, this study integrates 4 drought indices under SSP1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5, combining outputs from multiple CMIP6 models, to produce a global monthly drought projection dataset for the period 2015–2100 at a spatial resolution of 0.5°×0.5°. Validation against 5 major drought events indicates that DSI and SSMI better capture drought severity and spatial extent, suggesting that the dataset can effectively characterize future global drought patterns and their evolving trends.

Author Contributions

Liu, Y. X. Y. conducted the overall design for the dataset development; Cui, J. W. collected and processed the data; Xi, Y. S. designed the models and algorithms; Zhao, G. F. performed the data validation; Cui, J. W. wrote the data paper.

Conflicts of Interest

The authors declare no conflicts of interest.

References

[1]        Allen, C. D., Macalady, A. K., Chenchouni, H., et al. A global overview of drought and heat-induced tree mortality reveals emerging climate change risks for forests [J]. Forest Ecology and Management, 2010, 259(4): 660–684.

[2]        Lesk, C., Rowhani, P., Ramankutty, N. Influence of extreme weather disasters on global crop production [J]. Nature, 2016, 529(7584): 84–87.

[3]        He, X. G., Pan, M., Wei, Z. W., et al. A global drought and flood catalogue from 1950 to 2016 [J]. Bulletin of the American Meteorological Society, 2020, 101(5): E508–E535.

[4]        Wang, L., Chen, W., Fu, Q., et al. Super droughts over East Asia since 1960 under the impacts of global warming and decadal variability [J]. International Journal of Climatology, 2022, 42(9): 4508–4521.

[5]        Vicente-Serrano, S. M., Beguería, S., López-Moreno, J. I. A multiscalar drought index sensitive to global warming: the standardized precipitation evapotranspiration index [J]. Journal of Climate, 2010, 23(7): 1696–1718.

[6]        Wang, W., Wang, P., Cui, W. A comparison of terrestrial water storage data and multiple hydrological data in the Yangtze River basin [J]. Advances in Water Science, 2015, 26(6): 759–768.

[7]        Kogan, F. N. Application of vegetation index and brightness temperature for drought detection [J]. Advances in Space Research, 1995, 15(11): 91–100.

[8]        Eyring, V., Bony, S., Meehl, G. A., et al. Overview of the coupled model intercomparison project phase 6 (CMIP6) experimental design and organization [J]. Geoscientific Model Development, 2016, 9(5): 1937– 1958.

[9]        Tokarska, K. B., Stolpe, M. B., Sippel, S., et al. Past warming trend constrains future warming in CMIP6 models [J]. Science Advances, 2020, 6(12): eaaz9549.

[10]     Cui, J. W., Liu, Y. X. Y., Xi, Y. S., et al. Global drought experimental dataset using multi-scenario fusion methods (2015–2100) [J/DB/OL]. Digital Journal of Global Change Data Repository, 2026. https://doi. org/10.3974/geodb.2026.05.02.V1.

[11]     GCdataPR Editorial Office. GCdataPR data sharing policy [OL]. https://doi.org/10.3974/dp.policy.2014.05 (Updated 2017).