Drought frequency and intensity have increased in recent decades and are expected to escalate in future under the changing climate scenario. However, a wide range of uncertainty exists regarding the risk, variability and severity of the drought. This study evaluates the future drought and associated uncertainty over homogeneous regions of India using the suites of CMIP5 global climate models (GCMs) and CORDEX-SA regional climate models (RCMs). The drought characteristics and its projected future changes are analysed using probability density functions derived from hydroclimatic parameters, including the standardized precipitation index (SPI) and the standardized precipitation-evapotranspiration index (SPEI). Besides, uncertainties from various sources such as inter-model variability, indices type, timescale, adapted methods, and time-slices are explored under the RCP8.5 emission scenario to the end of the 21st century. Our study reveals large biases in the individual model; however, both the multi model ensembles (GCM and RCM) generally demonstrate better performance with respect to observation. In particular, the RCM ensemble showed limitations in capturing the regional precipitation pattern while temperature and potential evapotranspiration (PET) showed considerable enhancement concerning GCMs. SPI (SPEI) generally exhibited enhanced wetness (dryness) derived from increased precipitation (PET), although a few discrepancies were noticed. The regional heterogeneity was also found to exist, although some robust changes were noticed in drought frequency and severity with different return periods. Our finding underscores a wide range of uncertainties in drought projection, with maximum contribution from indices selection followed by model variability whereas other sources have the least contribution. The primary drivers for all these uncertainty sources arise due to variations among models simulated hydroclimatic variables that need to be parameterized more precisely for sustainable drought management.

中文翻译：

---

印度同质地区的未来干旱变化和相关不确定性：一种多模型方法

近几十年来，干旱的频率和强度有所增加，并且在不断变化的气候情景下，预计未来会升级。然而，关于干旱的风险、可变性和严重程度存在广泛的不确定性。本研究使用 CMIP5 全球气候模型 (GCM) 和 CORDEX-SA 区域气候模型 (RCM) 套件评估印度同质地区未来的干旱和相关不确定性。使用从水文气候参数得出的概率密度函数分析干旱特征及其预测的未来变化，包括标准化降水指数（SPI）和标准化降水-蒸散指数（SPEI）。此外，来自各种来源的不确定性，例如模型间变异性、指数类型、时间尺度、适应方法、在 RCP8.5 排放情景下探索到 21 世纪末的时间片。我们的研究揭示了个体模型存在很大的偏差；然而，多模型集成（GCM 和 RCM）通常在观察方面表现出更好的性能。特别是，RCM 集合在捕获区域降水模式方面表现出局限性，而温度和潜在蒸散量 (PET) 在 GCM 方面表现出显着增强。SPI (SPEI) 通常表现出因降水量增加 (PET) 而增加的湿度（干燥度），尽管注意到了一些差异。区域异质性也被发现存在，尽管在干旱频率和严重程度随着不同的回归期出现了一些强烈的变化。我们的发现强调了干旱预测的广泛不确定性，指数选择的贡献最大，其次是模型变异性，而其他来源的贡献最小。所有这些不确定性来源的主要驱动因素是由于模型模拟的水文气候变量之间的差异，这些变量需要更精确地参数化以进行可持续干旱管理。