Drought, a recurrent and persistent natural phenomenon, is an unavoidable force that can affect regions across the scale of arid to humid. Its onset remains vague, making it one of the most unpredictable yet least understood calamities to confront humanity (Fung et al., 2020; Huang et al., 2017; Mallya et al., 2015; Mishra and Singh, 2011a; Poonia et al., 2021a). At its core, drought emerges from a prolonged water availability deficit which disturbs water balance. This meteorological disturbance sets in motion a complex sequence of events that impacts water resources, the hydrological cycle, and, ultimately, food security (Aggarwal and Singh, 2010; Cook et al., 2020). Drought's commencement hinges on shifting hydrological patterns, marked by a disturbing lack of precipitation coupled with heightened evaporation and transpiration rates. A series of event unfolds from this meteorological drought, giving rise to a hydrological drought, culminating in an agricultural crisis (Gyamfi et al., 2019). Drought episodes in South Indian River basins are more common but less severe than those in Western and Central India, where they are more severe and lasts longer (Poonia et al., 2021a).
This study is an exploration of the complexity of drought, aiming to explain its diverse indicators. Ranging from meteorological droughts, influenced by precipitation deficits, to hydrologic droughts exhibiting stream flow anomalies, and groundwater droughts marked by declining aquifers, we meticulously examined the aspects of drought (Thomas et al., 2016). Moreover, we theoretically explored agricultural drought, which exacts a toll on the soil moisture essential for sustaining crop yields (Xu et al., 2021).
It has been aimed to cultivate a deeper understanding of drought's complexity and provide insights that can aid mitigation and adaptation efforts. In a world where climate change amplifies the frequency and intensity of droughts, this exploration is not just academic but enlightening for our collective survival. Drought being a multivariate event (Yang et al., 2018) is characterized by duration, magnitude, and intensity whose overall impact depends on spatio-temporal factors viz. area and frequency (Cook et al., 2020; Fung et al., 2020; Mishra and Singh, 2011b). The risk analysis of drought can be done with various modeling components like Probabilistic characterization of drought, Spatio-temporal drought analysis, Climate change impact on drought, Drought forecasting and management, and Land data assimilation (Kim et al., 2021; Mishra and Desai, 2005; Sreeparvathy and Srinivas, 2022).
Linear regression models are prepared by linear relationships between dependent variables and independent variables (Kim et al., 2020). Here the dependent variable is a quantifying drought parameter, for example, a drought index, whereas the independent variables are explanatory variables for the quantifying drought parameter (i.e., precipitation and temperature). The main drawback of this method is the assumption that there is a linearity between predictor and predictand variables, which becomes quite uncertain in the case of complex studies like drought (Ghosh and Mujumdar, 2007a). To counter this drawback, various time series models came into the picture, among which Seasonal Autoregressive Integrated Moving Average (SARIMA) and Autoregressive Integrated Moving Average (ARIMA) are popular ones, which effectively consider serial linear correlation among observations (Ghosh and Mujumdar, 2007a; Xu et al., 2022).
Since there is a complexity in quantifying uncertainties related to hydro-meteorological variables producing droughts, Drought modeling can also be done with probabilistic models where Markov chains are applied on monthly drought indices to large-scale air circulation patterns that might forecast droughts (Fung et al., 2020; Poonia et al., 2021b). Deep learning models can act as a bridge for the computational gaps in time series and probabilistic modeling. It requires less formal statistical training and can detect complex nonlinear relationships between dependent and independent variables (Prodhan et al., 2022). The benefits of both stochastic and deep learning models to combine a linear stochastic model and a nonlinear model to forecast can be employed for drought forecasting using hybrid models more accurately (Fung et al., 2020; Mishra and Singh, 2011a; Prodhan et al., 2022).
There is a passive ignorance regarding the inclusion of evapotranspiration in research works related to hydroclimatic extremes. Moreover, there is a preference of comparative simpler methods for the calculation of evapotranspiration rather than using the most accurate one till date i.e., the Penman-Monteith (PM) equation-FAO 56 (Allen et al., 2005; McNaughton and Jarvis, 1984; Nandagiri and Kovoor, 2006). It is reasonable to believe that the rise in temperature will have significant effects on drought conditions (Shafiei Shiva et al., 2022).
Drought indices help us to understand how severe a drought is and how it is impacting different sectors, which allows us to develop and implement effective drought management strategies (Mukherjee et al., 2018). Numerous well-established drought indices, such as Palmer Drought Index (PDI) (Rao and Padmanabhan, 1984; Vasiliades and Loukas, 2009) for prolonged meteorological drought assessment, Crop Moisture Index (CMI) (Palmer, 1968) for quick responses to changing soil moisture during growing seasons, Palmer Hydrological Drought Index (PHDI) (Vasiliades and Loukas, 2009) for long-term hydrological impact quantification, and Standardized Precipitation Index (SPI) (Keyantash, 2021; Memon and Shah, 2019) for probabilistic monthly precipitation-focused evaluations, serve as essential tools for comprehending and monitoring drought conditions. Additionally, Vegetation Health Index (VHI) (Bento et al., 2018) combines chlorophyll and moisture content, Objective Blended Drought Index Percentiles (OBDI) (Svoboda et al., 2002) offers a balanced view of short and long-term drought conditions, Palmer Drought Severity Index (PDSI) (Dai, 2011) employs a water balance model, Streamflow Drought Index (SDI) (Tabari et al., 2013) examines streamflow volumes, Reconnaissance Drought Index (RDI) assesses water deficit more accurately, and Standardized Precipitation Evapotranspiration Index (SPEI) blends sensitivity and multitemporal aspects, collectively providing a versatile toolkit for drought assessment and management, catering to various temporal and thematic requirements in drought analysis.
The Standardized Precipitation Evapotranspiration Index (SPEI) builds on the SPI's framework but adds a temperature component, which is a key factor in drought conditions for forecasting the progression of a drought. SPEI provides a multiscale methodology that enables drought assessments over a range of hydrological systems and time scales (Beguería et al., 2014a; Ghasemi et al., 2021; Mokhtar et al., 2021; Pyarali et al., 2022; Vicente-Serrano et al., 2010).