Agricultural drought response to meteorological drought over different agro-climatic zones of the Ganga River basin

The effects of drought on the environment, agriculture, economy, and society are extensive as it is a complex extreme event having variable duration and severity levels. The scientific processes behind the origin, propagation, and termination of drought are perplexing aspects and poorly understood in the world of research as drought can occur in all types of climatic zones at different forms with different intensities and duration (Mishra & Singh 2010). According to certain studies, droughts in India cause an average annual economic loss of a significant amount of GDP and it also depends on the drought category (e.g., mild, moderate, and intense droughts). Extreme climatic events in India are showing an increasing trend in recent years in which anthropogenic activities also have great influence. The land use change, the over-exploitation of water resources, etc. affects the hydrological balance of the planet which also contributes to extreme climatic events. The propagation of drought is a complex process impacted by several interconnected elements, including temperature, precipitation, evapotranspiration, humidity, climatic characteristics, land use practices, topography, etc. making it challenging to predict and understand its full extent (Wilhite & Glantz 1985). There are four basic categories for drought, firstly, meteorological drought, which occurs whenever precipitation is insufficient when compared to long-term normal; second, agricultural drought, which is a result of reduced soil moisture to support plant growth; thirdly, hydrological drought, which occurs as a result of reduced streamflow, and finally, socio-economic drought, where the human activities are affected due to the inadequate water availability (Das et al. 2022).

Considering food security in a region, it is important to monitor and predict natural disaster events like droughts as these can severely affect agricultural productivity (Mishra & Singh 2010). If the meteorological drought propagates into the agricultural drought, it affects the crop yield which can lead to insufficient food production (Hameed et al. 2020). As drought propagates from the atmosphere to agricultural areas, competition for limited water resources intensifies, and, therefore, in developing nations, the agricultural sector is one of those most impacted by drought. Drought is a major threat to India as well, causing increased susceptibility of crops to stress and various biotic and abiotic factors, and this condition is making it one of the most vulnerable countries in the world to drought. According to some studies, drought affects 68% of India's agricultural area, with 33% classified as ‘chronically drought-prone’ (Kaur et al. 2022). One of the main driving elements for triggering soil moisture stress is precipitation deficiency (Li et al. 2022) and hence studying the transition from meteorological to agricultural drought is crucial. Similar research work is of significant importance in any country as the propagation of drought from atmospheric conditions to agricultural drought can have several significant multifaceted effects such as increased evapotranspiration and competition for scarce water resources, serious soil moisture deficit, crop stress, and vulnerability, impacts on agricultural income and food security, impacts on rural livelihood and associated economic losses and other environmental impacts such as habitat degradation and biodiversity losses (Wei et al. 2022).

Drought propagation is a very complex and non-linear process, hence the study of the progression of conditions of negative rainfall anomalies to soil moisture water stress is intricately intertwined. Nevertheless, researchers have used a variety of techniques such as propagation threshold, correlation coefficients, aggregate drought indexes, standardized indicators, multivariate indicators, and copula-based approaches. Some researchers have combined the application of hydrological models and land surface process models with the intensity and duration of drought propagation (e.g., JULES land surface model). Some studies have proved that the application of hydrological and land surface models can help researchers understand how meteorological drought conditions propagate through the hydrological system, affecting soil moisture, river flows, and groundwater availability. To assess drought propagation nature in a simple and realistic manner, several standardized drought indices are developed and employed (Li et al. 2022) and these indices helped to derive new knowledge on drought characteristics like frequency, duration, and intensity. Meteorological drought can be assessed using various indices like the Standardized Precipitation Evapotranspiration Index (SPEI), Standardized Precipitation Index (SPI), etc. (Mckee et al. 1993). The Normalized Difference Vegetation Index (NDVI) (Cai et al. 2023), Vegetation Condition Index (Al Kafy et al. 2023), Vegetation Health Index (VHI) (Weng et al. 2023), and Standardized Soil Moisture Index (SSMI) (Wei et al. 2022) are the most commonly used indices for agricultural drought assessment. The lag time to agricultural drought from meteorological drought can be assessed effectively using correlation analysis. Traditionally, Pearson's correlation analysis has been used in many studies to assess drought propagation time (DPT) (Zhou & Shi 2021). Previous studies on meteorological drought to agricultural drought have shown very interesting but sometimes contrasting results. For example, the duration of drought propagation from meteorological to agricultural drought, in humid regions, typically does not exceed 3 months, whereas over the arid as well as semi-arid regions of China, it varies between 1 and 8 months. A study by Xu et al. (2023) identified that agricultural droughts are getting triggered on an average time of 48 days from the onset of meteorological drought when Improved Soil Moisture Anomaly Percentage Index (ISMAPI) and SPI were employed in China's Yangtze River Basin. The observed differences in DPT between regions may be attributed to limitations associated with the linear nature of correlation analysis, potential biases arising from the temporal resolution of the developed indices, or the absence of sub-monthly drought indices which can provide a more comprehensive representation of spatial and temporal distributions in the time for the propagation of water stress conditions. Regarding the propagation dynamics of agricultural drought in the Indian setting, there is a significant research gap. Globally, only a few studies have been conducted clearly pointing to the drought propagation values in terms of days or month, and these studies are primarily focusing on catchment or river basin scales. A recent study by Das et al. (2023) made an important contribution by suggesting that the DPT varies within the range of 5–7 months for the onset of drought and 9–15 months for the occurrence of drought peak in India, utilizing a non-stationary based approach. However, additional study is required to completely comprehend the complexities of drought propagation in India and close the current research gap. Several studies exist on agricultural and hydrological characteristics of the Ganga River basin (see Supplementary materials Table S1). The major gap in the existing literature on Ganga basin exists in terms of extensive analysis on transition from meteorological drought to agricultural drought.

Apart from drought propagation analysis, another important aspect from the agricultural drought point of view is the drivers in terms of hydro-climatic variables and anthropogenic activities. Climatic elements such as precipitation and vapor pressure deficit are identified as the major driver for the agricultural drought by many previous researchers (Dai et al. 2022), whereas, human alterations such as altered cropping patterns and supplementary irrigation can affect the development of agricultural drought (Ding et al. 2020; Shah et al. 2021). Wu et al. (2023) have attempted to assess the relative contribution of the climatic factors (temperature, precipitation, wind, and solar radiation) and human actions using the simple multiple linear regression (MLR) model. A similar approach has been used in the present study with the limited climate variables, details of which are discussed in the method section.

Motivated by the above mentioned aspects, this study is addressing the existing research gap regarding drought propagation and influencing factors in the Indian scenario, particularly in the Ganga River basin. Since the high vulnerability to drought in India due to its agrarian nature and reliance on agriculture, understanding the propagation from meteorological to agricultural drought along with its driving factors is crucial. This study aims to quantify meteorological drought using the SPI, agricultural drought using the SSMI, and vegetation health using the NDVI within the Ganga River basin. Additionally, to assess the time for drought propagation from meteorological to agricultural drought Pearson's correlation coefficient-based lag time analysis is employed. As different weather conditions and cropping patterns significantly affect the drought propagation mechanism, the present study also investigates the relative contribution elements (climatic and human factors) over the regions with different agro-climatic settings using an MLR-based approach.

The specific objectives of this study are as follows:

• To characterize the meteorological and agricultural drought patterns over the Ganga River basin from 2001 to 2020.

• To assess the DPT and rates in different agro-climatic regions with diverse climatic conditions within the Ganga River basin.

• To quantify the relative contribution of climatic factors and human activities to the agricultural drought for the 20 years of time span (2001–2020).

By achieving these objectives, the study aims to enhance the understanding of agricultural drought propagation dynamics and the influence of climatic and human factors from the Indian context, specifically within the Ganga River basin, and provide valuable insights for drought management and mitigation measures in the region.

The Ganga River basin is one of the major basins of India with 65.57% of the area contributing to agriculture. Ganga River basin lies between 22°30′ and 31°30′ north latitude and 73°30′ and 89°00′ east longitude in northern India. It is one of the most active hydrological systems in the world (Moench 2010). Almost 26% of India's total area is comprised of the Ganga River basin, which is 0.86 million km² (Anand et al. 2018). Gangotri Glacier in the western Himalayas is the source of the Ganga River. As it traverses different topographies, it joins the Bay of Bengal after 275.8 km (Sharma et al. 2021). Rainfall in the basin varies throughout the year and is present for a few months only. Most of this occurs during the monsoon season in the months ranging from June to October. Ganga is the largest contributor to India's agricultural sector because of the region's fertile soil (Ghirardelli et al. 2021). Food insecurity in the area may result from unfavorable changes in the meteorological variables and diminishing trends in the potential on-farm yield of rice (Mishra et al. 2013).

The largest portion (31%) of the Ganga basin is contributed by the CPHR. CPHR is distributed in the states of Rajasthan, Madhya Pradesh, and Uttar Pradesh. Both Uttar Pradesh and Uttarakhand are included in the Upper Gangetic, while Uttar Pradesh and Bihar are included in the MGP. The Eastern Plateau and Hills Region includes the states of Madhya Pradesh, Jharkhand, West Bengal, and Chhattisgarh. The states of West Bengal fully encompass the Lower Gangetic Plain Region and Eastern Himalayan Region. Madhya Pradesh and Rajasthan are the states that make up the Western Plateau and Hills Region. The Western Himalayan Region includes Himachal Pradesh and Uttarakhand. Only 2% of the basin's overall area is made up of the Trans-Gangetic Plain region, which is located in parts of Delhi and Haryana. The state of Rajasthan contains the Western Dry Region, the smallest of all the agro-climatic zones.

The daily gridded precipitation data from the India Meteorological Department (https://www.imdpune.gov.in/cmpg/Griddata/Rainfall_25_NetCDF.html) are used for the assessment of meteorological drought. The precipitation data having a spatial resolution of 0.25° × 0.25° was used from 2001 to 2020. Monthly precipitation time series was prepared from daily data to calculate the SPI at 1-month, 2-month, 3-month, ….., 12-month timescales. Monthly area-averaged (for agro-climatic regions) soil moisture data time series from ERA5 was extracted using the Google Earth Engine (GEE) platform for 20 years from 2001 to 2020. In the present study, soil moisture data from ERA5-Land (https://developers.google.com/earth-engine/datasets/catalog/ECMWF_ERA5_LAND_DAILY_RAW) were used which is originally at 0.1° × 0.1° spatial and 1 hourly temporal resolution. MOD13Q1.006 Terra Vegetation Indices 16-Day Global 250 m (https://developers.google.com/earth-engine/datasets/catalog/MODIS_061_MOD13Q1) was used in this study for assessing the agricultural drought. The PET data at 8-Day Global 500 m resolution are downloaded from GEE platform with MOD16A2 version for the required duration (https://developers.google.com/earth-engine/datasets/catalog/MODIS_061_MOD16A2). MODIS data were available only from the year 2001 and hence the time period of the study was selected from 2001. Two decades during 2001 to 2020 were chosen for the study for this reason.

The SPI's multiple timescales can accurately quantify water scarcity in soil moisture and streamflow, making it an ideal tool for modeling drought events (Bonaccorso et al. 2003). Many previous studies have demonstrated that spatial variation of drought severities can be captured based on different severity classes of the SPI (Mckee et al. 1993; Spinoni et al. 2019; Pachore & Remesan 2022) such as extreme, severe, moderate, and near-normal dry conditions.

where n is equal to the number of years equal to 20 years in the present study.

The Soil Moisture Index (SSMI) measures soil water stress by analyzing its interaction with hydro-meteorological variables like precipitation and evapotranspiration (Afshar et al. 2022). Soil moisture being a significant factor for crop growth, SSMI is widely used for the assessment of agricultural drought (Um et al. 2022). SSMI, with its similar calculation principle to the SPI, offers all the advantages of the SPI, including the ability to analyze agricultural drought at multiple timescales (Dai et al. 2022).

To assess the time of propagation from the meteorological to agricultural drought, the SPI timescale having the highest correlation with the monthly NDVI anomaly was used. The SPI_n was calculated for multiple timescales (n = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12) at 1-month intervals for the period from 2001 to 2020 (20 years). Similarly, the NDVI anomaly was also calculated at 1-month timescale with 1-month intervals from 2001 to 2020. To study the effects of agro-climatic conditions on drought propagation, the study was done independently over the different agro-climatic regions of the Ganga basin. The timescale of the SPI showing the highest correlation with the NDVI anomaly was considered to be the lag time from the meteorological drought to the agricultural drought in terms of vegetation dynamics.

The propagation relation between various types of droughts was examined using the MPCC method. Many studies (Xu et al. 2019; Ding et al. 2021) have successfully demonstrated that the PCC is effective in showing the response of one type of drought to the other. Since soil moisture also gives significant information regarding the incidents of agricultural drought, the SPI_n calculated for multiple timescales was correlated with the 1-month SSMI. Propagation time taken by soil-based agricultural drought for responding to meteorological drought was computed using the SPI timescale having the highest correlation with the SSMI-1 time series. Also, the effect of soil moisture variation on the vegetation dynamics was studied by correlating the SSMI-n calculated for multiple timescales (n = 1, 2, 3, 4, 5, 6, 9, 12) at 1-month intervals with 1-month NDVI anomaly. We chose this analysis as the MPCC is a widely demonstrated and accepted analysis for addressing drought response time and propagation, as evidenced by studies (Zhou & Shi 2021; Weng et al. 2023) spanning transitions from meteorological to agricultural or hydrological droughts. Shi et al. (2022) conducted a comprehensive global evaluation of drought propagation using spatial distributions of MPCC. Additionally, studies such as Das et al. (2022) and Zhou et al. (2021) emphasize its utility in quantifying the relationship between SPI and other drought indices across different timescales, including the decadal scale and lengths. These studies highlight the robustness and reliability of MPCC, offering insights into the potential impact of meteorological drought on hydrological drought propagation and elucidating the impacts of human activities.

Here, RC (i) is the relative contribution of the term ⁠, where i belongs to the predicted (climatic factors) and residual (human activities and other factors) SSMI factors, whereas, denominator of the Equation (10) gives the overall trend of the SSMI contributed by each term.

Based on heatmaps shown in Figure 3, it is evident that, in the CPHR, the years 2002, 2007, and 2018 were the major drought years, with 2007 being the year having prominent moderate drought conditions. The Eastern Himalayan region showed more intense drought conditions than the CPHR, under the category of ‘Severely Dry’ with the year 2014 being more prominent. For the Eastern Plateau and Hills region, the year 2010 falls under severe drought conditions, whereas, in the Lower Gangetic Plain region, severe drought was observed in 2012. In the MGP, moderately dry condition was experienced in 2010 whereas the Trans-Gangetic Plain experienced near-normal condition in 2009 and 2016. In the Upper Gangetic Plain and Western Plateau and Hills region, moderately dry conditions were experienced in 2002 and 2016, respectively. The Western Dry Region and Western Himalayan Region experienced severely dry conditions in 2002 and 2009, respectively.

Different regions experienced meteorological drought in different degrees but how the vegetation responded to this depends on various factors including the agro-climatic and hydrological conditions of the region (Ding et al. 2021). The agricultural drought can be determined effectively by the soil moisture-based SSMI or vegetation-based indices like NDVI and VHI. As the soil moisture is strongly related to crop conditions and will give reliable information regarding the vegetation characteristics, SSMI is being used in the present analysis. As the health of vegetation can be determined by the reflectance of the vegetation and hence the NDVI which represents the reflectance of vegetation in red and NIR bands has also been used as an effective tool for the assessment of the vegetation condition of agro-climatic regions for the present analysis.

The variable severity of drought conditions in each region is influenced by agro-climatic factors as well as the soil characteristics of each region. The Upper Gangetic region with rich soil and water resources experienced the least drought conditions among all. The use of groundwater for irrigation practices can also be a reason for the fair soil moisture conditions in this region (Ganga Basin Report 2014).

The agricultural drought assessment using NDVI anomaly is depicted in Figure 5(b). The CPHR experienced a vegetation-based agricultural drought in the year 2003 with an NDVI anomaly of −1.76, which corresponds with the soil moisture-based agricultural drought of the same region in the year 2002. Agricultural drought in the Eastern Himalayan Region was noticeable in 2007 with an NDVI anomaly of −3.85. The Eastern Plateau and Hills region and Lower Gangetic Plain region, with NDVI anomaly of −1.74 and −1.03, faced agricultural drought in 2016 and 2019. The MGP region and Trans-Gangetic Plain region with NDVI anomaly of −1.68 went through an agricultural drought in the year 2001. The Upper Gangetic Plain in the year 2003 had drought conditions with an NDVI anomaly of −1.93. The Western Dry region with an NDVI anomaly of −1.46 experienced the drought condition in 2010. In the Western Himalayan Region and Western Plateau and Hills region, drought condition was evident with NDVI anomalies of −1.51 and −1.76 in the years 2003 and 2014, respectively.

Studies done in the past have highlighted that the occurrence of agricultural drought depends on various factors like meteorological conditions of the area, soil and crop characteristics, etc. Results of the present study have led to similar observations, as the agricultural drought of a higher degree, when compared to all other regions, was experienced in the Eastern Himalayan region which is suffering from soil erosions and degradation with NDVI anomaly of −3.85 in 2007. However, the Lower Gangetic Plain region, which is rich in soil and water resources, experienced the least agricultural drought conditions with NDVI anomaly of −1.03 in the year 2019.

To assess the agricultural drought response (propagation time) to meteorological drought, this study uses the PCC-based lag time analysis. Many studies on drought propagation (Weng et al. 2023; Zhou & Shi 2021) from one type to another (e.g., meteorological to agricultural, meteorological to hydrological) have successfully demonstrated the reliability of MPCC on assessing the DPT. Through MPCC, the linear correlation between the meteorological drought and agricultural drought was assessed using SPI-NDVI and SPI-SSMI correlation. The main focus of the study was assessing the DPT and DPR and the effect of agro-climatic conditions on these two variables in the Ganga River basin of India. Twenty-year data was found to be good to get the trend and draw conclusions for the assessment of DPT and DPR.

To assess the propagation characteristics from meteorological to agricultural drought, the MPCC between the SPI-SSMI-1 and SPI-NDVI anomaly for each agro-climatic region was assessed. The SPI-n (n = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12) at multiple timescales was correlated with 1-month SSMI and NDVI anomaly as depicted in Figure 6. It was observed that SPI-3 of EPHR, LGP, MGP, UGP, and WHR showed the highest correlation with SSMI-1. SPI-1 of EHR, SPI-4 of TGP, SPI-9 of WPHR, SPI-10 of CPHR, and SPI-11 of WDR showed the highest correlation with SSMI-1. Similarly, based on the MPCC of SPI-NDVI, CPHR, EPHR, and WHR, SPI-8 showed the highest correlation with the 1-month SSMI for the corresponding regions. The LGP, WDR, and WPHR showed a higher correlation at SPI-11 while MGP and UGP showed a higher correlation at SPI-1. SPI-5 and SPI-3 showed the highest strength of association in EHR and TGP.

The MPCC between SPI and the NDVI anomaly varied from 0.02 to 0.186 for different agro-climatic regions. In CPHR, EPHR, and WHR, SPI-8 showed the highest correlation with the monthly NDVI anomaly, with MPCCs of 0.186, 0.06, and 0.07, respectively. Similarly, SPI-11 showed the highest correlation with NDVI anomaly for LGP, WDR, and WPHR with MPCCs of 0.023, 0.22, and 0.20, respectively. SPI-5 in EHR showed the highest correlation of 0.06 with NDVI anomaly and in TGP, SPI-3 showed the highest correlation with MPCC of 0.14. In MGP and UGP, SPI-1 showed the highest correlation with MPCC values of 0.10 and 0.13, respectively. Li et al. (2023) have found the non-linear and complex connection of agricultural drought with meteorological drought for different climatic regions in their study. The findings of the present analysis also support the same with varied strength of PCC for regions with different agro-climatic conditions, highlighting the control of climate on drought propagation.

Agricultural drought (or soil moisture stress) typically responds quickly, usually within a few months, to meteorological drought. Based on the region's climate and type of ecosystem, this study found that the effects of drought varied greatly. The drought resistance of various ecosystem types may vary due to varying climatic and vegetational characteristics (Cao et al. 2022). Agricultural drought is generally driven by meteorological drought and follows it with a certain amount of delay. Many researchers in the past have acknowledged that a particular period marks the transition for meteorological drought to get reflected into agricultural and hydrological drought which is termed as propagation or response time. Correlation analysis is one of the most popular approaches for determining the propagation time between different drought types. The present study uses the Pearson correlation coefficients-based method to examine the correlations between the SPI at multiple timescales and SSMI-1, along with NDVI anomaly to determine when a meteorological drought transition can take place into an agricultural drought.

Similar to SPI-SSMI, the lag time analysis was also done based on SPI-NDVI to find the DPT. Based on the results, the lag time between the two droughts under consideration, i.e., meteorological (SPI) and agricultural (NDVI) drought, was observed to be 11 months in the LGP, WDR, and WPHR. In the CPHR, EPHR, and WHR, the time for drought propagation was observed to be 8 months. The EHR and TGP regions showed a DPT of 5 months and 3 months, respectively. The MGP and UGP showed a DPT of 1 month in both regions. The highest DPT was observed in LGP, WDR, and WPHR whereas the least was observed in MGP and UGP. SPI-NDVI based analysis also showed a similar result to that of SPI-SSMI computations for agricultural DPT quantification with similar evidence for control of climatic conditions on DPT as shown in Figure 7(b). MPCC analysis indicated statistically insignificant SPI-NDVI correlations. Consequently, for DPT and DPR assessment, SPI-SSMI correlations were employed, assigning 100% weightage to SSMI for agricultural drought assessment and 0% weightage to the NDVI.

Many researchers (Xu et al. 2023; Sattar et al. 2019) have used the propagation rate as a crucial indicator of how sensitive the agricultural drought is to meteorological dryness. Similarly, in the current study, the percentage of propagation between meteorological and soil moisture-based agricultural droughts was expressed as DPR. Due to the effect of soil characteristics and crop conditions or other factors, not all meteorological droughts will result in agricultural drought, and not all agricultural droughts will correspond to meteorological drought. The DPR was quantified based on meteorological drought events characterized by SPI (timescale selected from the MPCC-based drought propagation analysis) and agricultural drought events by SSMI-1.

The drought propagation rates were different for all the agro-climatic regions. The highest DPR was observed in the Eastern Himalayan Region and Western Plateau and Hills region as shown in Figure 8. Mostly, the meteorological drought is not the only cause of agricultural drought but various factors like the initial condition of the catchment, irrigation, climate type, crop characteristics, and soil characteristics also play an important role in the transition toward agricultural drought (Li et al. 2022; Xu et al. 2023). Apart from the meteorological factors, human activities and climate change (CC) play crucial roles in the variations of DPT.

Overall, it is observed that human activities are more dominant as compared to climatic factors on agricultural drought occurrence over most parts of the basin. This can be attributed to the perennial nature of the Ganga River which makes it rich in surface and ground water resources even during the non-monsoon season with better irrigation facilities, resulting in the weaker climate control on agricultural drought occurrence (Swarnkar et al. 2021). These inferences are useful from the CC point of view, demonstrating that the changing climatic conditions may have fewer impacts than human activities on agricultural drought over most of the Ganga basin provided it is not affecting the rainfall and overall annual water balance over the basin. As discussed by Wu et al. (2023), people in general identify agricultural drought primarily on the basis of precipitation amount and its variations, human influences pose significance influences on areas with lower or variable rainfall just as reported in the TGP region of Ganges.

The diverse agro-climatic zones within the Ganga River basin exhibit distinct agricultural practices and human interventions in terms of irrigation, etc., leading to variations in agricultural drought propagation rates and the relative contributions of CC and human activities (HA). This heterogeneity is evident in the findings of Singh et al. (2023), who noted diverse behavior in extreme precipitation events across different agro-climatic regions, attributed to variations in basin topography, ranging from 8 to 7798 m in elevation. Additionally, Bhatla et al. (2020) highlighted significant diversity in cropping patterns across the 10 agro-climatic regions, with the lower Gangetic plain predominantly cultivating rice, while other regions favor wheat as the dominant crop. This variation in cropping patterns and irrigation conditions can markedly influence evapotranspiration rates, leading to variable rates of soil moisture decline and subsequently impacting DPT and rate. Sun et al. (2023) supported these findings, emphasizing the variability in DPTs associated with different vegetation types characterized by varying crop root zone depths in different regions.

While this study focuses on the two-decadal analyses of drought propagation, the findings underscore potential future repercussions across various sectors such as energy, water supply and sanitation, agriculture, industry, fisheries, forestry, and other ecosystems in the study region due to anticipated increases in drought frequency and severity, presenting challenges that are intricate to comprehend (Teutschbein et al. 2023). This highlights the need for future studies that focus on drought scenarios, employing robust climate projections and scenarios utilizing appropriate General Circulation Models (GCMs) and Regional Climate Models (RCMs). Such investigations are crucial for quantifying the impending impacts of future droughts and formulating effective mitigation strategies through climate adaptation (Mukherjee et al. 2018). Several studies have emphasized key strategies essential for future drought mitigation across diverse contexts, including: (i) strategic interventions in land use and land, (ii) the augmentation of soil water retention, (iii) the adoption of infield water conservation techniques, and (iv) the promotion of public awareness regarding droughts, their consequences, and the imperative for potential countermeasures. Implementation of such measures can collectively contribute to a comprehensive framework for effective future agricultural drought mitigation in the Ganga River basin.

The Ganga basin stands out as one of the most dynamic river basins in terms of water availability, intricately connected to drought patterns across regions with diverse agro-climatic conditions. Though the present study effectively reported relative contribution of climatic as well as human elements on agricultural droughts on various agro-climatic zones, it is crucial to acknowledge potential limitations inherent in the chosen approaches, models and datasets. However, the results of drought propagation have revealed distinct strengths of association between the SPI and both the NDVI and SSMI. The sensitivity of soil and crop health to meteorological drought variations may extend its influence to other variables not encompassed in this study, possibly by adding uncertainties to the reported results. Even though the initial calculations were on a pixel-based approach, it is important to highlight that area-averaged time series are employed to yield an average condition for each agro-climatic region. Furthermore, the whole analysis was performed on annual time series, and this could potentially mitigate the influence of rainfall seasonality especially in the monsoon season. Above mentioned caveats might have led to the averaging out of certain seasonal information in the Ganga basin as a dominant factor in India's regions characterized by seasonal climate and especially monsoon.

Various studies have shown additional influencing factors such as local climatic conditions, land use practices, irrigation facilities, and rainfall seasonality. Ding et al. (2020) demonstrated that cropland, forest, grassland, and desert exhibit diverse responses to meteorological drought due to variations in vegetation characteristics. Our study does not consider these specific factors, potentially introducing uncertainties into the results. Our study has employed relatively simple but very effective residual trend method initially proposed by Jin et al. (2020) and subsequently adapted for drought assessments by Wu et al. (2023) with the help of a simple MLR model which may have ignored the non-linear interactions of the variables considered. The relationship of different climatic factors to drought responses may be non-linear and this may be making interpretations of MLR model results difficult (Seddon et al. 2016). Furthermore, the MLR model relied solely on SPI and PET due to limited data availability for ground variables (such as supplementary irrigation and local agricultural practices). This limitation might have led to an exaggerated relative contribution of human elements in certain regions, such as EPHR, TGP, and WHR.

Despite the challenges outlined in the above paragraph, the present study contributes fresh insights into the variable nature of meteorological conditions affecting the propagation time and rate of agricultural drought. This extends to sub-regional spatial scales and highlights the interconnections with various features of agro-climatic zones within the Ganga River basin. The quantified understanding of the relative impact of climatic factors and human activities on agricultural drought can prove valuable for policymakers in formulating future strategies, particularly with a focus on agro-climatic zones in the basin.

Ten agro-climatic regions of the Ganga responded differently to the meteorological drought and exhibited different lag times between the drought types (meteorological and agricultural) along with varying propagation rates. This variability may be due to the soil characteristics, crop characteristics, and climatic conditions of the region along with the human influences like irrigation water use and land use alterations. The present study has analyzed the drought propagation mechanism along with the relative contribution of the climatic as well as human factors and drawn the following specific conclusions:

• The regions rich in water resources having fertile soil, like the Upper Gangetic Region, Lower Gangetic Plain Region, Trans-Gangetic Plain Region, and Eastern Himalayan Region, experienced a lesser degree of agricultural drought with DPT varying from 1 to 4 months, whereas the regions with shallow soil conditions experienced a higher degree of agricultural drought.

• For the CPHR, with an MPCC value of 0.72, the DPT was found to be 6 months.

• The Eastern Himalayan Region and Upper Gangetic Plain Region showed comparatively lesser sensitivity of SSMI to the SPI with MPCC values of 0.23 and 0.24 corresponding to the DPT of 1 month and 3 months, respectively.

• The drier regions like Western Dry Region, Western Plateau and Hills region, etc. responded very slowly to changes in the meteorological drought conditions with a DPT of 11 and 9 months, respectively.

• Different agro-climatic regions showed different drought propagation rates varying from 29.03% to 73.33% under the influence of different agricultural and climatic conditions.

• The changing climatic conditions may not affect the agricultural drought as most of the regions of the Ganga basin show the dominance of human activities in agricultural drought development.

In the future work of this research team on Ganga basin analysis, enhanced outcomes could be achieved by integrating an advanced regression model that considers additional variables, providing a more accurate representation of local catchment conditions (Dai et al. 2022). Thus the assessment of DPT and rate could be extended to future climate scenarios by employing suitable GCMs and RCMs to reveal and anticipate potential future instances of agricultural drought in the region.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.