A Spatiotemporal Analysis of Drought Conditions Framework in Vast Paddy Cultivation Areas of Thung Kula Ronghai, Thailand

Abstract

This study presents an integrated spatiotemporal assessment of drought conditions in the Thung Kula Ronghai region of Northeastern Thailand from 2001 to 2023. Multiple satellite-derived drought indices, including SPI, SPEI, RDI, and AI, together with NDVI anomalies, were used to detect seasonal and long-term drought dynamics affecting rainfed Hom Mali rice production. The results show that dry season droughts now affect up to 17 percent of the region’s agricultural land in some years, while severe drought zones persist across more than 2.5 million hectares over the 20-year period. In the most recent 5 years, approximately 50 percent of cultivated areas experienced moderate to severe drought conditions. The RDI showed the strongest correlation with NDVI anomalies (r = 0.22), indicating its relative value for assessing vegetation response to moisture deficits. The combined index approach delineated high-risk sub-regions, particularly in central Thung Kula Ronghai and lower Surin, where drought frequency and severity have intensified. These findings underscore the region’s increasing exposure to dry-season water stress and highlight the need for site-specific irrigation development and adaptive cropping strategies. The methodological framework demonstrated here provides a practical basis for improving drought monitoring and early warning systems to support the resilience of Thailand’s high-value rice production under changing climate conditions.

1. Introduction

1.1. Background and Rationale

Rice cultivation is a strategic cornerstone of Thailand’s food security and national economy, securing the country’s longstanding position among the world’s top rice exporters [1]. Oryza sativa L. var. Khao Hom Mali (Thai jasmine rice) is globally recognized for its superior fragrance and grain quality, commanding premium prices in international markets [2]. This aromatic rice variety is predominantly cultivated in Northeastern Thailand, where it provides a primary source of income for millions of smallholder households and plays a pivotal role in sustaining rural employment and community livelihoods [3]. The production and export of Hom Mali rice contribute significantly to Thailand’s agricultural GDP and foreign exchange earnings, underpinning both local economic stability and national trade competitiveness [4].

Thung Kula Ronghai, spanning parts of Roi Et, Surin, Maha Sarakham, Si Sa Ket, and Yasothon provinces, is among the largest and most critical production zones for Hom Mali rice [5]. This lowland plain is characterized by sandy, nutrient-poor soils with inherently low water retention capacity, which, combined with minimal irrigation infrastructure, renders the region highly vulnerable to interannual climatic variability and recurrent droughts [1,5]. Periodic droughts have repeatedly disrupted rice cultivation cycles, diminished yields, and caused considerable economic losses for farming households dependent on this high-value crop [3]. Climate projections further indicate that rising temperatures and increasingly erratic monsoon patterns will likely intensify drought frequency and severity across the region, amplifying existing production risks [2].

Given its premium market value and irreplaceable role in sustaining local livelihoods, any climate-induced decline in Hom Mali rice production directly threatens both household economic security and Thailand’s reputation as a leading exporter of high-quality rice. To address these vulnerabilities, an integrated, spatiotemporal drought monitoring framework that can capture localized drought dynamics and support timely water resource management is urgently required for this climate-sensitive rice cultivation system.

1.2. Drought Monitoring Through Environmental Indices and Remote Sensing

Effective drought monitoring in rainfed agricultural regions increasingly depends on multi-source environmental indices derived from meteorological data and satellite-based remote sensing [6,7]. Drought is a complex phenomenon driven by interactions among precipitation deficits, elevated evapotranspiration demand, soil moisture depletion, and vegetation stress [8]. Relying on a single index often fails to capture this complexity in full. Therefore, integrating multiple complementary indices enhances detection accuracy and supports early warning and adaptive management in climate-sensitive cropping systems such as rainfed rice cultivation in northeastern Thailand [9].

The Standardized Precipitation Index (SPI) is widely recognized for characterizing meteorological drought through the analysis of precipitation anomalies over various timescales [10]. However, because the SPI considers only precipitation, it may underestimate drought severity during periods of elevated temperatures that increase evapotranspiration. To address this limitation, the Standardized Precipitation Evapotranspiration Index (SPEI) incorporates potential evapotranspiration (PET), providing a more robust representation of climatic water balance in warming environments [6].

The Reconnaissance Drought Index (RDI) also integrates precipitation and PET, using aggregated climatic water balance to assess seasonal or interannual drought patterns, which is particularly suitable for semi-arid and tropical regions [8]. The Aridity Index (AI) represents the long-term ratio of precipitation to PET and is widely used to classify climatic zones and to contextualize background drought risk at broad spatial and temporal scales [11]. Although the AI is not sensitive to short-term anomalies, it remains useful for understanding structural aridity that shapes long-term agricultural vulnerability [12].

In addition to meteorological and climatic indicators, vegetation-based indices derived from optical satellite sensors capture the biophysical response of crops to drought stress. The Normalized Difference Vegetation Index (NDVI) anomaly measures deviations in vegetation greenness relative to long-term baselines and provides a practical proxy for assessing moisture stress and crop health [13]. Although the NDVI may react with a time lag due to crop phenology, its spatial coverage complements meteorological indices by illustrating the ecological impacts of drought conditions on rice-growing areas [3]. In this study, NDVI anomalies were analyzed using seasonal composites, which partially account for this lag by aggregating vegetation response over the main cropping cycle [3,13].

Using SPI, SPEI, RDI, AI and NDVI anomaly in a combined spatiotemporal framework allows for a more comprehensive characterization of drought onset, intensity, persistence, and agricultural impacts. This multi-index approach addresses the limitations of single-index methods, captures compound drivers and seasonal dynamics, and bridges the gap between meteorological monitoring and practical agricultural drought assessment. Such integration is critical for early warning and targeted drought risk management in regions like Thung Kula Ronghai, where rainfed rice cultivation remains highly sensitive to climatic variability.

1.3. Global to Regional Perspectives on Remote Sensing-Based Drought Monitoring

Drought monitoring has undergone substantial development in recent decades due to advancements in remote sensing and GIS technologies. These tools now allow researchers to observe, quantify, and interpret drought dynamics across diverse ecosystems and climatic conditions. This section synthesizes the key literature spanning global, regional (Asia), and national (Thailand) contexts to establish a coherent foundation for applying a remote sensing index-based approach in the Thung Kula Ronghai region.

Globally, recent studies demonstrate the expanding role of satellite-derived biophysical indices and advanced analytical techniques in capturing drought phenomena with increasing accuracy. For example, Savinelli et al. investigated how Sentinel-2 imagery could be applied to monitor changes in forest biophysical traits under severe drought conditions in Northern Italy, highlighting species-specific sensitivities and the importance of local site factors [14]. Similarly, Hu et al. demonstrated the value of thermal infrared data by developing the Temperature Rise Index (TRI) derived from Himawari-8 satellite observations in Australia. The TRI showed a strong correlation with soil moisture and crop yield, providing early drought warnings and complementing optical vegetation indices [15].

In the study by Chere, a Combined Agricultural Drought Index (CADI) was constructed to enhance monitoring precision for Ethiopia. This was achieved by integrating multi-source satellite data, including the normalized difference vegetation index (NDVI), precipitation, soil moisture, land surface temperature (LST), and evapotranspiration. The study confirmed the index’s validity by demonstrating a strong correlation with agricultural crop yield records, with the highest correlation being r = 0.87 in one region. This underscores the index’s value for effectively monitoring agricultural drought and its potential use in an early warning system [16]. Łągiewska and Bartold extended the scope of remote sensing applications by combining satellite-derived drought indices with Multi-Criteria Decision Analysis to map high-risk drought areas in Poland, underscoring the mitigating role of forest cover [17].

In the Asian context, similar methodological innovations have been adapted to the region’s complex monsoon climate and diverse agricultural systems. Neeti et al. developed a high-resolution drought monitoring framework in India by downscaling CHIRPS precipitation data using Random Forest algorithms and multi-sensor satellite covariates, producing the Effective Meteorological Drought Index (EMDI) [18]. In Vietnam, Binh et al. analyzed the combined impacts of drought and salinity intrusion on land use change in the Mekong Delta, demonstrating how increased drought frequency drives agricultural vulnerability [19]. In China, Wu et al. constructed a dynamic threshold model for daily drought monitoring in the Golden Maize Belt using Near-Infrared and Short-Wave Infrared bands, resulting in high accuracy when validated against historical drought records [20]. Li et al. developed a high-resolution Standardized Precipitation Evapotranspiration Index (SPEI) dataset to improve spatial drought detection, while Mu and Tian addressed spatiotemporal limitations in NDVI data by fusing MODIS and Landsat imagery to produce daily 30 m NDVI for cropland monitoring [21,22].

At the national level, Thailand has progressively adopted and contextualized these global and regional approaches. Homdee et al. compared three climatic drought indices in the Chi River Basin and concluded that including evapotranspiration data through indices such as the SPEI and the SPAEI enhances drought detection in tropical monsoon climates [23]. Khadka et al. expanded this understanding with a multi-timescale analysis of meteorological drought in the Mun River Basin, confirming that rainfall variability, rather than annual precipitation, drives drought risk [24]. In an urban context, Srisuwan demonstrated how spatial analysis and stakeholder participation can support drought adaptation by developing a Water-Sensitive Land Use Map for Lamphun Municipality [25]. Raksapatcharawong et al. combined remote sensing data with the SIMRIW-RS crop growth model to assess drought impacts on rice yield, achieving high prediction accuracy and demonstrating the practical benefits of satellite-assisted yield forecasting [26].

1.4. Global and Regional Perspectives on Drought Monitoring Frameworks

Over the past two decades, many regions around the world have developed integrated drought monitoring frameworks that combine meteorological, hydrological, and remote sensing data to strengthen early warning systems and climate risk management [13]. For example, in the United States, the U.S. Drought Monitor (USDM) adopts a convergence-of-evidence approach that synthesizes multiple drought indices, expert input, and local impact reports to generate weekly drought severity maps with high spatial resolution [27]. In Europe, the Combined Drought Indicator (CDI) under the European Drought Observatory integrates precipitation anomalies, soil moisture models, and satellite-based vegetation indicators to detect agricultural drought hotspots across national borders [7].

Beyond these widely cited examples, countries in Asia and Africa have advanced region-specific monitoring systems that reflect diverse climatic and socio-economic contexts. In China, multi-scalar drought indices and seasonal climate forecasts are integrated to monitor regional drought dynamics and support large-scale water resource allocation [13]. Similarly, India employs regional drought early warning systems that combine rainfall anomalies, temperature trends, and remote sensing-derived vegetation indices to assess agricultural drought risk in monsoon-dependent cropping zones [28]. In Africa, indicator-based frameworks have been developed to combine remote sensing data with socio-economic vulnerability factors, enabling early warning tools that address both biophysical and human dimensions of drought impacts [12].

These examples highlight several shared principles relevant to drought-prone agricultural landscapes such as Thung Kula Ronghai. First, multi-source data integration reduces the limitations of single-index methods by capturing compound drought drivers and local impacts. Second, multi-scale monitoring frameworks enable both real-time alerts and retrospective analysis, supporting proactive risk management. Third, regionally calibrated thresholds and stakeholder inputs ensure that drought classifications are meaningful within local agro-ecological contexts. By drawing on such global and regional precedents, this study’s integrated multi-index framework aims to advance current drought monitoring practice for Thailand’s rainfed rice systems and to inform scalable adaptation strategies under increasing climate variability.

1.5. Identified Gaps in Existing Drought Monitoring Systems

Despite advances in regional and global drought monitoring frameworks, significant gaps remain that limit their practical application in heterogeneous agricultural zones such as Thung Kula Ronghai. Many operational systems in Southeast Asia still rely heavily on short-term meteorological data with limited spatial resolution, which constrains the detection of localized drought impacts in fragmented rainfed landscapes [29]. In particular, single-index approaches that depend solely on rainfall anomalies or basic vegetation indicators often fail to capture compound drought conditions, especially when atmospheric drivers and vegetation responses diverge due to complex soil, irrigation, or land management factors [13,28].

Furthermore, institutional fragmentation frequently results in inconsistent monitoring methodologies and delayed reporting, which weakens the effectiveness of early warning and coordinated response [29]. Another critical gap is the lack of region-specific calibration for drought severity thresholds. Many indices rely on standardized global baselines without adjusting for local crop phenology, soil constraints, or cultural cropping practices, which can lead to misclassification of drought onset and severity [12]. Without such localized refinement, drought monitoring remains reactive rather than proactive, leaving smallholder farmers particularly exposed to climate shocks.

To address these limitations, there is a clear need for an integrated, multi-index, and regionally calibrated framework that combines meteorological data, vegetation indices, and contextual agro-ecological factors. Such a framework would enhance the capacity to detect both the physical drivers and ecological impacts of drought conditions, thereby supporting more timely and targeted adaptation measures in climate-sensitive rice production systems like Thung Kula Ronghai.

1.6. Study Objectives and Significance

Building upon the identified gaps in conventional drought monitoring systems, this study aims to develop and apply an integrated multi-index framework for assessing spatiotemporal drought dynamics in Thung Kula Ronghai, one of Thailand’s most critical rainfed rice-producing regions. By combining meteorological indicators such as the SPI, SPEI, RDI, and AI with vegetation-based NDVI anomaly, the research seeks to overcome the limitations of single-index approaches and capture the compound climatic and ecological dimensions of drought. Specifically, the study pursues three interrelated objectives. First, it analyzes historical drought patterns and trends from 2001 to 2023 using multi-source remote sensing and climatic datasets. Second, it quantifies the spatial heterogeneity and seasonal variability of drought conditions and evaluates the correlation between drought indices and vegetation responses under different climatic regimes. Third, it identifies priority sub-regions within Thung Kula Ronghai that are most vulnerable to recurrent and severe drought events, thereby providing a scientific basis for targeted water management and adaptive agricultural strategies.

The findings of this research are expected to contribute to the refinement of regional drought monitoring frameworks by demonstrating the value of a multi-index design that integrates physical drivers and biophysical impacts. Moreover, the study offers empirical evidence that can support local and national policymakers in formulating site-specific adaptation measures to enhance the resilience of Thailand’s high-value Hom Mali rice production system under intensifying climate risks.

2. Materials and Methods

Figure 1 illustrates the overall research workflow employed in this study. The process begins with the acquisition of multi-source input data, including GPM precipitation, MODIS-based PET and NDVI products, and administrative shapefiles. These datasets undergo systematic pre-processing and resampling in Google Earth Engine (GEE) version v1.5.24rc0 to ensure spatial and temporal consistency. Subsequent stages involve the computation of five drought-related indices (the SPI, SPEI, AI, RDI, and NDVI anomaly) and the generation of spatiotemporal drought frequency and severity maps. Finally, correlation analyses are conducted using ArcGIS Pro version 3.1 and Python version 9.3 to examine the relationships between drought indices and vegetation response, providing an integrated framework for drought monitoring in rainfed rice areas.

2.1. Study Area

Thung Kula Ronghai is a prominent rainfed rice production zone located in Northeastern Thailand, covering approximately 3800 square kilometers across five provinces: Roi Et, Surin, Maha Sarakham, Si Sa Ket, and Yasothon. The region is topographically characterized by low-lying plains with sandy and saline soils that pose challenges for water retention and crop productivity. The area relies primarily on seasonal monsoon rainfall to sustain extensive paddy fields, which makes it particularly sensitive to rainfall variability and recurring drought conditions. Figure 2 illustrates the spatial extent of the study area.

2.1.1. Geological and Soil Characteristics

The geological and soil characteristics of the Thung Kula Ronghai region, located in the Korat Basin, present significant edaphic constraints for agriculture. The area’s soil salinity is largely influenced by the underlying Mahasarakham Formation, a major source of salt [30]. The dominant topsoil is sandy loam, characteristically low in organic matter, resulting in poor fertility and limited water-holding capacity [30]. These soil-based constraints, coupled with the region’s vulnerability to drought, pose significant challenges to sustainable rice cultivation and contribute to low crop yields [5, 30]. Therefore, adapting local management strategies is essential, focusing on both improving soil conditions [30] and optimizing water use to mitigate drought impacts [5].

2.1.2. Climate and Hydrology

Thung Kula Ronghai has a Tropical Savannah Climate (Aw in the Köppen classification) characterized by a distinct monsoon pattern with clearly defined wet and dry seasons. Annual rainfall averages between 1200 and 1400 mm, with nearly 85 percent of total precipitation falling during the wet season from May to October [1]. The remaining dry season, which extends from November to April, is marked by limited rainfall and elevated evapotranspiration rates, leading to recurrent soil moisture deficits.

The region’s west-to-east sloping topography accelerates surface runoff during intense rainfall events, often resulting in rapid water loss and flash flooding while limiting infiltration into the soil profile [5]. These hydrological conditions, combined with the area’s naturally low soil water retention, contribute to frequent and prolonged drought conditions during the dry season. Although major rivers such as the Mun and Chi intersect parts of the region, their regulated flows and limited storage capacity are insufficient to meet irrigation demands throughout the year, underscoring the area’s vulnerability to seasonal water scarcity.

2.1.3. Drought Susceptibility and Agricultural Impact

The combination of sandy and saline soils, pronounced rainfall seasonality, and limited natural water retention capacity renders Thung Kula Ronghai highly susceptible to drought-related stress. These conditions restrict the availability of soil moisture needed to sustain rainfed rice cultivation and heighten the risk of yield losses during extended dry periods. In this study, the spatiotemporal dynamics of drought across the region from 2001 to 2023 are examined using integrated remote sensing and climatic datasets to quantify both the frequency and severity of drought events. The analysis aims to delineate drought-prone sub-regions within rice cultivation zones and provide evidence to inform water resource management and adaptive agricultural strategies to strengthen the resilience of one of Thailand’s most significant rainfed rice landscapes.

2.2. Data Collection and Sources

This study integrated multiple environmental and geospatial datasets to conduct a comprehensive spatiotemporal drought assessment for Thung Kula Ronghai covering the period from 2001 to 2023. Drought-related indices and vegetation health metrics were derived primarily from satellite-based observations and processed using remote sensing and geographic information system (GIS) platforms.

2.2.1. Satellite-Derived Environmental Data

Monthly precipitation data were acquired from the Global Precipitation Measurement (GPM) Mission, providing consistent coverage for computing the SPI, SPEI, AI, and RDI across the study area. Potential evapotranspiration (PET) data were sourced from the MODIS MOD16A2 product, while NDVI values for vegetation anomaly analysis were extracted from the MODIS MOD13A1 dataset. Each index was calculated for two main seasonal intervals: the wet season (May to October) and the dry season (November to April), using 6-month cumulative precipitation and PET values.

All datasets were imported into the Google Earth Engine (GEE) to enable large-scale pre-processing and temporal alignment. The GPM and MODIS layers were resampled to a common spatial resolution to ensure consistency across indices. Seasonal NDVI anomalies were computed by standardizing seasonal NDVI values against the long-term mean and standard deviation for each seasonal window.

2.2.2. Geospatial and Administrative Data

Paddy rice cultivation zones within Thung Kula Ronghai were delineated using vector shapefiles provided by the Land Development Department and Rice Department of Thailand. These vector boundaries were used to mask the drought index raster layers, ensuring that the analysis was spatially constrained to active rice-growing areas only.

2.2.3. Analytical Tools and Scripting

All drought indices were computed using custom Python scripts that applied appropriate probability distributions, including the Gamma distribution for the SPI and the log-logistic distribution for the SPEI. NDVI anomaly was calculated using standardized z-scores. The use of automated Python workflows ensured consistency and reproducibility across the full 23-year study period.

Spatial correlation between drought indices and vegetation anomalies was evaluated by systematically sampling 30-point locations within rice cultivation areas. Raster index values at these sample points were extracted using the Extract Values to Points function in ArcGIS Spatial Analyst for Desktop (License: ESU420743008). The extracted values were analyzed in Python to compute Pearson correlation coefficients, which were visualized in a correlation matrix to interpret the strength and direction of relationships among the indices.

2.3. Computation of Drought Indices

This study employed five drought-related indices: the SPI, SPEI, AI, RDI, and NDVI anomaly to characterize the spatiotemporal dynamics of drought across Thung Kula Ronghai from 2001 to 2023. Each index was calculated separately for the wet season (May to October) and the dry season (November to April) to enable detailed seasonal comparisons.

2.3.1. Standardized Precipitation Index (SPI)

The SPI was computed using 6-month cumulative precipitation values derived from the GPM dataset. For each seasonal interval, cumulative rainfall data were fitted to a Gamma probability distribution and then transformed into a standard normal distribution:

$$\mathrm{S}\mathrm{P}\mathrm{I}={\displaystyle \frac{{\mathrm{P}}_{\mathrm{i}}-\overline{\mathrm{P}}}{{\mathsf{\sigma }}_{\mathrm{P}}}},$$

(1)

where ${\mathrm{P}}_{\mathrm{i}}$ is the cumulative precipitation for the given period, $\overline{\mathrm{P}}$ is the long-term mean, and ${\mathsf{\sigma }}_{\mathrm{P}}$ is the standard deviation [10]. The process was fully automated using a Python script to ensure consistency across the study period.

2.3.2. Standardized Precipitation Evapotranspiration Index (SPEI)

The SPEI was computed by integrating precipitation and PET data. The difference between precipitation and PET was first calculated for each 6-month interval and then fitted to a log-logistic distribution:

$$\mathrm{S}\mathrm{P}\mathrm{E}\mathrm{I}={\displaystyle \frac{{\mathrm{D}}_{\mathrm{i}}-\overline{\mathrm{D}}}{{\mathsf{\sigma }}_{\mathrm{D}}}},{\mathrm{D}}_{\mathrm{i}}={\mathrm{P}}_{\mathrm{i}}-{\mathrm{P}\mathrm{E}\mathrm{T}}_{\mathrm{i}},$$

(2)

where ${\mathrm{P}}_{\mathrm{i}}$ is precipitation and ${\mathrm{P}\mathrm{E}\mathrm{T}}_{\mathrm{i}}$ is potential evapotranspiration for the same period [6].

2.3.3. Aridity Index (AI)

The Aridity Index (AI) was computed as the ratio of total precipitation to total PET for each seasonal window:

AI = P/PET,

(3)

where P is cumulative precipitation and PET is cumulative potential evapotranspiration [11].

2.3.4. Reconnaissance Drought Index (RDI)

The RDI was derived in two stages. First, the initial RDI value was computed as:

RDI_initial = P/PET

(4)

Subsequently, the standardized RDI was calculated through the following transformation:

RDI = (RDI_initial − μ)/σ,

(5)

where μ is the mean of RDI_initial and σ is the corresponding standard deviation over the 23-year period.

2.3.5. NDVI Anomaly

The NDVI anomaly was calculated from MODIS MOD13A1 imagery. The NDVI values for each season were extracted and standardized using the formula:

NDVI_anomaly = (NDVI_current − NDVI_mean)/NDVI_std,

(6)

where NDVI_current and NDVI_std are the long-term mean and standard deviation of the NDVI for the corresponding seasonal window, respectively. This anomaly captures vegetation stress as a function of deviation from the expected seasonal vegetation state.

All computations were automated using Python version 9.3 to ensure methodological consistency and reproducibility across the full spatiotemporal domain.

2.4. Spatial Correlation Analysis of Drought Indices and NDVI Anomaly

To assess the relationship between drought severity and vegetation stress, a spatial correlation analysis was conducted using the five selected indicators: the SPI, SPEI, AI, RDI, and NDVI anomaly. A total of 30 sample points were systematically selected within paddy rice cultivation zones across Thung Kula Ronghai to ensure spatial representativeness across different agro-ecological conditions. For each year from 2001 to 2023, the raster values of the five indices were extracted at each sample point using the Extract Values to Points function in ArcGIS Spatial Analyst for Desktop (License: ESU420743008).

Seasonal index values were averaged to represent annual conditions and compiled into a correlation matrix. Pearson correlation coefficients were then computed using Python to evaluate the strength and direction of the relationships between meteorological drought indices and NDVI anomaly. The resulting matrix provided quantitative evidence to interpret how well each drought index reflected the observed vegetation response over the full study period.

2.5. Spatiotemporal Drought Frequency and Trend Analysis

2.5.1. Methodology for Drought Frequency Analysis

This study assessed the long-term frequency of drought events across Thung Kula Ronghai using the five selected drought indices. The analysis was conducted for three overlapping temporal windows: the recent 5 years (2019–2023), 10 years (2014–2023), and 20 years (2004–2023). Annual and seasonal maps were produced to identify recurrent drought zones and to detect intensifying trends over time.

For each index, drought conditions were classified using established threshold values: the SPI, SPEI, and RDI were assigned drought status when values were below −0.99; the AI indicated drought when it fell below 0.5; and the NDVI anomaly signaled vegetation stress when the value was less than −0.1 [3,6,8,11]. Each pixel was assigned a binary value, with “1” indicating drought and “0” indicating non-drought status.

A location was considered to be under drought stress if at least three out of the five indices simultaneously signaled drought conditions during a given seasonal or annual step. These classified layers were then aggregated for each of the three temporal windows to generate composite raster maps representing drought occurrence frequency at annual, wet-season, and dry-season scales.

2.5.2. Methodology forDrought Severity Analysis

A majority classification approach was used to determine the dominant drought severity level at each pixel location based on the five drought indices. For each index, drought severity was categorized into four classes: no drought, moderate drought, severe drought, and very severe drought, using standardized thresholds established in the literature [6,8,10]. At each pixel for every seasonal and annual step, the most frequently occurring severity class among the five indices was recorded as the dominant condition. This composite classification produced a single raster layer representing the prevailing drought severity for each of the three temporal windows: the past 5 years (2019–2023), past 10 years (2014–2023), and past 20 years (2004–2023).

3. Results

3.1. Standardized Precipitation Index (SPI) Analysis (2001–2023)

Figure 3 presents the annual SPI maps for Thung Kula Ronghai from 2001 to 2023, separated into wet (May–October) and dry (November–April) seasons. These maps depict spatiotemporal patterns in precipitation anomalies, with blue indicating wetter-than-normal conditions, red for drier-than-normal, and yellow to white for near-average precipitation.

3.1.1. Trends in SPI During the Wet Season (May–October) of SPI

During 2001–2005, SPI values were predominantly positive across the region, suggesting above-average rainfall conditions. However, isolated red zones indicated localized drought pockets in 2001, 2005, and 2007. From 2016 to 2022, a marked shift toward negative SPI values was observed, with 2017 standing out due to widespread and intense dry conditions.

3.1.2. Trends in SPI During the Dry Season (November–April)

The dry season exhibited greater temporal fluctuation. While near-normal precipitation occurred many years before 2010, a notable shift emerged after 2016, with SPI values increasingly falling below zero. The year 2017 again emerged as a period of extreme dryness, and persistent precipitation deficits were evident from 2019 to 2022, pointing to mounting dry-season water stress.

3.2. Standardized Precipitation-Evapotranspiration Index (SPEI) Analysis (2001–2023)

Figure 4 presents the time series of the Standardized Precipitation-Evapotranspiration Index (SPEI) across Thung Kula Ronghai from 2001 to 2023, segmented into the wet season (May–October) and dry season (November–April). These maps depict the spatiotemporal variability in the regional water balance, derived from the difference between precipitation and potential evapotranspiration. Positive SPEI values (blue) indicate wetter-than-normal conditions, negative values (red) signify drier-than-normal conditions, and yellow to white areas represent near-average climatic conditions.

3.2.1. Trends in SPEI During the Wet Season (May–October)

During the early 2000s, years such as 2002, 2006, and 2008 showed predominantly positive SPEI values across the region, indicating adequate precipitation relative to evapotranspiration. In contrast, red-colored zones representing localized drought conditions emerged in 2001, 2003, 2005, 2007, 2013, 2014, and 2015. After 2014, a clear spatial expansion of negative SPEI values was observed, especially in western subregions, suggesting an increased prevalence of wet-season droughts.

3.2.2. Trends in SPEI During the Dry Season (November–April)

Dry season SPEI anomalies exhibited greater interannual variability. Prior to 2008, alternating positive and negative SPEI values were evident, reflecting shifts in the regional hydro-meteorological balance. After 2008, negative SPEI anomalies became more dominant, with pronounced droughts in 2017 and 2022. These years were marked by widespread negative values, highlighting substantial dry-season water deficits across extensive areas of the study region. The dry season maps also reveal more pronounced spatial heterogeneity in SPEI patterns, with localized patches of wetter-than-normal conditions appearing intermittently within broader drought-dominated zones. This emphasizes the uneven distribution of climatic water balance across the region during the dry months.

3.3. Aridity Index (AI) Analysis (2001–2023)

Figure 5 shows the annual Aridity Index (AI) maps for Thung Kula Ronghai from 2001 to 2023, divided into two seasonal periods: May–October (wet season; left column) and November–April (dry season; right column). The AI illustrates the balance between precipitation and potential evapotranspiration, serving as an indicator of regional water availability.

3.3.1. Trends in AI During the Wet Season (May–October)

Throughout the wet season, AI maps consistently display blue-shaded areas across most of the region, indicating higher AI values and sufficient water availability. These values suggest that monsoon rainfall reliably replenished soil moisture each year, with no significant signs of widespread aridity detected during this season.

3.3.2. Trends in AI During the Dry Season (November–April)

In contrast, the dry season AI maps are dominated by yellow to red shades, signifying persistently lower AI values. These indicate widespread arid conditions and notable water stress across the region during the dry months. While certain years such as 2008 and 2021 showed slightly less dryness, markedly arid periods were evident in 2011–2015, 2018–2019, and 2022–2023, reflecting extended phases of limited water availability.

3.4. Reconnaissance Drought Index (RDI) Analysis (2001–2023)

Figure 6 presents the annual RDI maps for Thung Kula Ronghai from 2001 to 2023, covering both the wet season (May–October; left column) and dry season (November–April; right column). The RDI indicates meteorological drought severity based on the ratio of precipitation to potential evapotranspiration, where positive values represent wetter-than-normal conditions and negative values indicate drought stress.

3.4.1. Trends in RDI During the Wet Season (May–October)

During the wet season, most years across the region exhibited positive RDI values (blue shades), indicating sufficient rainfall to offset evaporative demand. However, localized negative anomalies (red shades) were observed in years such as 2003, 2004, 2005, 2006, 2009, 2012, 2015, 2018, and 2019. These areas suggest seasonal deficits where precipitation was insufficient to fully meet potential evapotranspiration needs.

3.4.2. Trends in RDI During the Dry Season (November–April)

The dry season revealed more pronounced variability in RDI patterns. Between 2001 and 2008, values fluctuated between positive and negative phases, reflecting alternating wet and dry conditions. Severe drought conditions were evident in 2004, 2007, 2009, 2010, 2014, 2015, 2018, 2019, and 2023, as indicated by extensive red-shaded areas. A sustained negative trend has been apparent since 2009, with the years 2019 and 2023 displaying especially widespread drought signals across the region.

3.5. NDVI Anomaly Analysis (2001–2023)

Figure 7 illustrates the spatiotemporal variations in the Normalized Difference Vegetation Index Anomaly (NDVI anomaly) across Thung Kula Ronghai from 2001 to 2023, highlighting seasonal and annual deviations in vegetation health. NDVI anomaly values are presented for two seasonal divisions: May to October (wet season; left column) and November to April (dry season; right column). Blue shades represent positive anomalies (denser vegetation), red shades indicate negative anomalies (vegetation stress), and yellow to white tones correspond to near-normal conditions.

3.5.1. Trends in NDVI Anomaly During the Wet Season (May–October)

In most years, NDVI anomaly values during the wet season remained close to the long-term climatological average (yellow areas), suggesting relatively stable vegetation conditions. Notable negative NDVI anomalies were recorded during 2001–2002, indicating localized vegetation stress. In contrast, 2009 marked the first widespread occurrence of positive anomalies (blue shades). Other years such as 2010, 2012–2013, 2015, 2017, and 2018 exhibited intermittent red zones, likely reflecting temporary reductions in vegetation productivity.

3.5.2. Trends in NDVI Anomaly During the Dry Season (November–April)

The dry season revealed more pronounced fluctuations in NDVI anomalies. Prior to 2003, positive anomaly values were dominant, indicating generally vigorous vegetation growth. From 2003 to 2006, NDVI values declined sharply, pointing to severe dry conditions that suppressed vegetation greenness. A period of partial recovery began in 2008, as positive anomalies reappeared in several zones. Negative NDVI anomalies re-emerged in 2018–2019, while strong positive anomalies in 2021–2022 suggested a recent improvement in vegetation conditions during the dry months.

3.6. Correlation Analysis of Drought Indices (the SPI, SPEI, RDI, AI) and NDVI Anomaly

The correlation matrix in Figure 8 summarizes the statistical relationships among the five key drought indices, averaged across both wet and dry seasons from 2001 to 2023. The analysis shows that SPI exhibits a weak negative correlation with NDVI anomaly (r = –0.23), indicating that increases in precipitation alone do not consistently translate to improved vegetation conditions. The SPEI demonstrates a very weak correlation with NDVI anomaly (r = −0.05), reflecting the limited influence of combined precipitation and evapotranspiration on short-term vegetation response in this region. RDI presents a slightly stronger but still weak positive correlation (r = 0.22), suggesting that the precipitation-to-evapotranspiration ratio may capture vegetation moisture availability more directly than the SPI or SPEI alone. In contrast, the Aridity Index shows an almost negligible correlation with NDVI anomaly (r = 0.03), implying that long-term aridity patterns alone have a minimal immediate effect on seasonal vegetation variation. Among the drought indices themselves, the SPI and SPEI are strongly correlated (r = 0.65), confirming their close dependence on precipitation-based inputs.

3.7. Drought Frequency Analysis

To evaluate recurring drought conditions within paddy cultivation areas of the Thung Kula Ronghai region, a frequency-based analysis was conducted using five drought indices: the SPI, SPEI, RDI, AI, and NDVI anomaly. A pixel was classified as experiencing drought if at least three out of the five indices indicated drought simultaneously. Spatial drought frequency was assessed over three time periods: 5 years (2019–2023), 10 years (2014–2023), and 20 years (2004–2023).

The combined drought frequency maps (Figure 9) illustrate a progressive increase in drought occurrences over time and across seasons. The left column shows the total drought frequency for each multi-year period. The central column displays drought frequency during the wet season (May–October), while the right column presents drought frequency during the dry season (November–April).

For the overall frequency (left column), during the 5-year period, most of the study area remained in the lowest drought category (deep purple), although light blue patches emerged in eastern Surin Province, specifically in Phon Khrok, Thung Kula, and Phrom Thep subdistricts, with up to six drought events recorded out of ten seasonal intervals.

In the 10-year analysis, drought-affected areas expanded, with maximum frequencies reaching eleven events out of twenty intervals. Transition zones from green to yellow emerged, particularly along lower elevation areas and riverbanks in Surin Province.

The 20-year assessment revealed the most extensive drought coverage, with up to fifteen drought events documented out of forty intervals. High-frequency drought zones (red) appeared in central Thung Kula Ronghai and along the Mun River corridor, while moderate-frequency areas (six to nine events; blue/green) became widespread in peripheral districts.

A seasonal comparison shows that wet-season droughts (central column) were generally minimal in extent and severity across all three periods. In contrast, dry-season droughts (right column) were more frequent and spatially dominant, significantly contributing to the region’s total annual drought burden.

The numerical breakdown of these patterns is summarized in

Table 1. Percentage of unit area under different drought frequencies across wet and dry seasons for the 20-year, 10-year, and 5-year periods.

Unit Area: %	20-Year Period		10-Year Period		5-Year Period
Drought Frequency (Occurrences)	Wet Season	Dry Season	Wet Season	Dry Season	Wet Season	Dry Season
0	78.93	1.05	79.71	8.19	86.47	38.99
1	19.47	0.64	18.79	21.62	13.53	29.74
2	1.57	8.55	1.49	23.48	-	22.87
3	0.03	13.89	0.01	19.3	-	7.60
4	-	17.03	-	14.15	-	0.70
5	-	14.72	-	8.13	-	0.10
6	-	13.73	-	3.76	-	-
7	-	11.88	-	1.06	-	-
8	-	9.01	-	0.27	-	-
9	-	5.55	-	0.05	-	-
10	-	2.51	-	-	-	-
11	-	0.97	-	-	-	-
12	-	0.35	-	-	-	-
13	-	0.1	-	-	-	-
14	-	0.03	-	-	-	-

There is one wet season and one dry season annually.

, which presents the percentage of the total study area falling under different drought frequency classes, distinguished by wet and dry seasons across the three time periods.

3.8. Drought Severity Analysis

To evaluate the spatial and temporal dynamics of drought severity across Thung Kula Ronghai, a majority classification approach was applied to five drought indices (the SPI, SPEI, RDI, AI, and NDVI anomaly) over three-time intervals: 5 years (2019–2023), 10 years (2014–2023), and 20 years (2004–2023). Drought severity was categorized into four levels: no drought, moderate drought, severe drought, and very severe drought, using standard classification thresholds. The spatial distribution of drought severity for each period and season is presented in Figure 10.

During the 5-year period, most of the region experienced moderate drought conditions, with the central area showing the highest concentration. Some eastern zones recorded severe drought, while small patches remained without drought stress. Over the 10-year interval, drought severity intensified as moderate zones expanded into severe drought, especially in the central and northeastern parts of the region. Areas without drought declined significantly, reflecting an overall increase in severity.

In the 20-year analysis, widespread moderate and severe drought conditions were observed throughout the region. Severe drought zones extended further into Maha Sarakham Province, while only a small area in Mueang Khaen Subdistrict, Si Sa Ket Province, remained largely unaffected. Seasonal analysis confirms that wet seasons generally maintained moderate drought conditions, although some local areas shifted to severe drought, particularly in the recent 5-year period. The 20-year wet season map shows a slight reduction in severe drought coverage compared to shorter periods.

In contrast, the dry season consistently exhibited higher drought severity across all time intervals. The 5-year dry season results indicate extensive severe drought, especially in Surin Province, with only limited areas classified as moderate. Over the 10-year span, severe drought zones expanded further east, while parts of Maha Sarakham showed minor improvement. The 20-year dry season assessment revealed the highest concentration of persistent severe drought, particularly in western Maha Sarakham and lower Surin, with severe drought covering approximately 2.49 million hectares and moderate drought accounting for about 1.25 million hectares.

These patterns, as illustrated in Figure 10, underscore the structural vulnerability of rainfed agricultural areas in Thung Kula Ronghai and highlight the need for locally adapted water storage, irrigation expansion, and drought-resilient cropping to reduce long-term risks (

Table 2. Percentage of unit area classified by drought severity levels across wet and dry seasons for the 20-year, 10-year, and 5-year periods.

Unit Area: %	20-Year Period		10-Year Period		5-Year Period
Drought Severity	Wet Season	Dry Season	Wet Season	Dry Season	Wet Season	Dry Season
No drought	-	0.06	0.30	0.22	19.82	0.54
Moderate	34.54	33.41	55.39	38.52	49.42	49.10
Severe	65.23	66.53	44.09	61.26	30.76	50.36
Very severe	0.23	-	0.22	-	-	-

).

4. Discussion

4.1. SPI Trends and Drought Implications

The SPI analysis reveals an evident long-term trend of increasing meteorological drought conditions in the Thung Kula Ronghai region, with pronounced intensification observed particularly after 2016. This persistent drying signal is consistent across both the wet and dry seasons, confirming that precipitation anomalies have become more widespread and recurrent over time [10]. The notable convergence of severe SPI anomalies in 2017 underscores the magnitude of that year’s drought and its substantial potential to disrupt rainfed rice production cycles. Furthermore, the expanding spatial extent of negative SPI anomalies after 2016 suggests that drought events in this region have transitioned from being localized to affecting larger contiguous tracts of agricultural land, a trend similarly noted in other rice-dependent regions of mainland Southeast Asia [1]. This evolving drought pattern has direct implications for smallholder farming communities dependent on Hom Mali rice cultivation, which remains highly sensitive to seasonal rainfall variability. The demonstrated responsiveness of SPI to seasonal precipitation shifts reinforces its utility as a core indicator for early warning systems and adaptive water resource management [6]. By capturing both the frequency and severity of meteorological droughts, SPI-based monitoring provides a practical basis for identifying priority sub-regions within Thung Kula Ronghai that require targeted drought preparedness and resilient cropping strategies.

4.2. SPEI Trends and Drought Dynamics

The negative shift in SPEI values, especially after 2014, indicates a clear intensification of drought severity in Thung Kula Ronghai. Unlike the SPI, which captures only precipitation anomalies, the SPEI reflects both declining rainfall and increasing evaporative demand, offering a more comprehensive representation of the region’s seasonal water balance [6]. The observed post-2014 decline in SPEI value highlights that drought conditions have emerged not only during the dry season but increasingly during the wet season as well. This challenges traditional assumptions that monsoonal precipitation consistently mitigates drought risk in rainfed rice systems [1]. The westward concentration of negative SPEI anomalies and the spatial expansion of severe drought zones imply that local hydrometeorological dynamics and land use patterns may be amplifying evapotranspiration stress. These findings underscore the need for region-specific water resource planning that incorporates both seasonal precipitation and potential evapotranspiration trends. By accounting for compound drought drivers, the SPEI provides a valuable indicator for assessing climatic water balance anomalies and guiding adaptive irrigation strategies, particularly as warming scenarios are projected to increase evaporative demand in similar semi-arid agricultural zones [6,13].

4.3. Aridity Patterns and Agricultural Implications

The Aridity Index (AI) results reveal a distinct seasonal contrast in the long-term water availability across the Thung Kula Ronghai region. High AI values during the wet season reaffirm the significance of monsoonal rainfall in replenishing soil moisture and sustaining rice cultivation cycles. Conversely, the persistently low AI values during the dry season confirm the region’s structural exposure to extended periods of water scarcity [11]. This pattern underscores the underlying climatic aridity that shapes baseline drought risk, regardless of short-term precipitation anomalies. It is important to acknowledge that the AI is primarily designed as a broad-scale climatic classification tool rather than an event-specific drought index [12]. Unlike the SPI or SPEI, the AI does not capture interannual fluctuations or abrupt seasonal shifts but instead contextualizes the prevailing semi-arid nature of the region’s agroecological conditions. When integrated with other indices, the AI serves to identify areas with chronic arid tendencies that may require sustained investment in irrigation infrastructure, soil moisture conservation practices, and drought-resilient cropping strategies. The alignment of consistently low AI values with zones of high drought frequency further highlights the need for long-term land and water resource planning in Thung Kula Ronghai. As climate projections indicate intensifying dry-season aridity under future warming scenarios, the AI remains a vital reference for designing region-specific adaptation frameworks that reinforce agricultural resilience.

4.4. RDI-Based Drought Dynamics and Agricultural Implications

The temporal evolution of RDI patterns highlights a marked intensification of drought severity in Thung Kula Ronghai over the past two decades, with the trend particularly evident during the dry season. While the wet season generally maintains sufficient rainfall to support rice cropping, the recurrence of negative RDI anomalies in specific years points to increasing rainfall variability and growing risk of seasonal water stress [8]. The persistent post-2012 shift towards lower RDI values aligns with the broader trend of declining climatic water balance reported in other tropical agricultural zones [13]. Unlike the SPI and SPEI, which separately capture precipitation anomalies and climatic water balance, respectively, the RDI integrates both variables to depict a more holistic measure of drought intensity over seasonal scales. This makes it particularly useful for characterizing the agro-hydrological conditions that directly affect rice yield reliability in semi-arid regions like Thung Kula Ronghai [3]. The identified recurrence of dry-season water stress in years such as 2019 and 2023 signals the increasing fragility of rainfed production systems under intensifying climatic extremes. These findings reinforce the importance of adopting adaptive water management measures, including the expansion of irrigation capacity and the promotion of drought-resilient crop varieties. By providing a nuanced understanding of both short-term and seasonal drought dynamics, the RDI serves as a practical decision-support tool for sub-regional drought preparedness and targeted interventions aimed at sustaining the Hom Mali rice system under variable climatic regimes.

4.5. Vegetation Response to Climatic Variability

The long-term pattern of NDVI anomalies reveals the dynamic response of vegetation to fluctuating hydro-meteorological conditions in Thung Kula Ronghai. Negative NDVI anomalies observed during 2001–2002 and again from 2003 to 2006 reflect pronounced periods of vegetation stress likely induced by insufficient soil moisture and rainfall deficits [3]. The subsequent recovery in 2008 and the occurrence of positive NDVI anomalies in 2021–2022 suggest the capacity of local cropping systems to rebound under more favorable climatic conditions or improved land management interventions. Importantly, NDVI responses do not always coincide directly with short-term drought signals captured by meteorological indices due to inherent time lags associated with crop phenology and soil moisture buffering capacity [13]. This lagged relationship highlights the role of vegetation indices as complementary indicators for monitoring the biophysical impacts of drought rather than as stand-alone drought predictors. Spatial clustering of positive NDVI anomalies near water bodies further underscores the importance of localized irrigation networks and micro-hydrological factors in moderating vegetation stress. These findings reaffirm that integrating vegetation-based monitoring with meteorological and hydrological indices enhances the detection of drought-induced impacts on crop health. For rainfed rice systems such as Hom Mali rice cultivation in Thung Kula Ronghai, this integrated approach supports more nuanced drought risk assessments and informs targeted resilience strategies at the sub-regional level.

4.6. Implications of Drought–Vegetation Correlations

The correlation analysis conducted in this study indicates that the statistical relationships between meteorological drought indices and vegetation response, as measured by the NDVI anomaly, are relatively weak. The negative correlation between the SPI and NDVI anomaly (r = −0.23) suggests that higher precipitation levels do not consistently result in improved vegetation conditions, which may be explained by localized factors such as soil saturation, drainage inefficiency, or temporary waterlogging effects [3]. Similarly, the near-zero correlation observed for the SPEI (r = −0.05) highlights the limited predictive capacity of combined precipitation–evapotranspiration balance alone in capturing vegetation stress under heterogeneous field conditions [13]. The RDI shows a marginally stronger positive correlation with NDVI anomaly (r = 0.22), implying that its integrated approach may partially reflect moisture availability relevant to crop growth cycles. However, this relationship remains modest and reinforces the understanding that vegetation stress is shaped by multiple interacting drivers beyond climatic water balance alone, including soil texture, irrigation practices, and crop management. Therefore, while the RDI demonstrates some potential for linking climatic drought signals to vegetation response, it should not be regarded as a stand-alone predictor. Instead, these correlation patterns underscore the necessity of integrating meteorological indices with vegetation-based monitoring and local agronomic information to build more robust early warning systems. For regions like Thung Kula Ronghai, where rainfed rice farming remains highly climate-sensitive, this integrated framework provides a more comprehensive basis for targeted adaptation planning and site-specific drought mitigation.

4.7. Implications of Drought Frequency Patterns

The spatial and temporal assessment of drought frequency in Thung Kula Ronghai reveals a clear pattern of intensifying drought occurrences, particularly over the past two decades. The results indicate that both the recurrence and geographic extent of drought events have expanded, with the highest frequencies recorded in the central basin and subdistricts such as Phon Khrok and low-lying areas along the Mun River. These findings are consistent with broader regional trends in Southeast Asia, where the combined effects of seasonal rainfall deficits and limited water storage capacity exacerbate the vulnerability of rainfed agricultural zones [12]. The comparison between short-term (5-year) and long-term (20-year) drought frequency maps demonstrates that while isolated extreme events can occur in the short term, persistent multi-year drought clusters pose a more significant threat to agricultural sustainability. The quantitative breakdown in Table 1 highlights that dry-season drought events are notably more frequent than wet-season occurrences, underscoring the structural reliance of the region on monsoonal rainfall to offset prolonged dry spells. Identifying sub-regional hotspots with repeated drought exposure is therefore critical for designing risk-sensitive water management strategies. This includes prioritizing off-season water storage investments, enhancing irrigation network coverage, and promoting drought-resilient rice varieties tailored to local microclimatic conditions. By integrating these frequency-based insights into practical planning, stakeholders can strengthen the adaptive capacity of Hom Mali rice systems to withstand recurring hydro-meteorological stressors.

4.8. Interpretation of Drought Severity Trends

The multi-decadal assessment of drought severity underscores a clear intensification of hydro-meteorological stress in Thung Kula Ronghai over the past 20 years. The results show a distinct shift from moderate drought to severe drought conditions, with the most affected zones expanding into traditionally less drought-prone areas such as Maha Sarakham Province. This trend aligns with long-term regional analyses which emphasize the compounding effects of rainfall deficits and increasing evapotranspiration in semi-arid agroecosystems [12]. A comparison across the 5-year, 10-year, and 20-year periods indicates that while short-term variations exist, the persistence and geographic spread of severe drought conditions have grown substantially. The slight reduction in severity observed in parts of eastern Surin during the most recent 5-year interval may suggest the initial benefits of improved irrigation coverage and local adaptation measures. However, the dry season remains the dominant period driving overall drought severity, as shown by the higher proportion of severe drought classifications relative to the wet season. These findings highlight the urgent need for proactive investment in off-season water storage infrastructure, targeted irrigation development, and adaptive cropping strategies that can reduce the exposure of vulnerable sub-regions to intensifying drought risk. By anchoring severity-based insights in practical land and water management plans, local stakeholders and policymakers can strengthen the resilience of Hom Mali rice production against future climatic variability.

4.9. Limitations and Suggestions

This study provides a comprehensive spatiotemporal assessment of drought conditions in the Thung Kula Ronghai region by integrating multiple satellite-derived indices and vegetation anomaly metrics. While the results offer valuable insights, several limitations must be acknowledged to contextualize the findings and guide future research directions.

First, the exclusive reliance on medium-resolution satellite data, such as MODIS and GPM products, constrains the detection of localized drought impacts, particularly in fragmented smallholder farming areas with heterogeneous land cover [28]. Incorporating higher-resolution imagery, for example, from Sentinel-2 or Landsat missions, would enhance spatial precision and allow for finer-scale mapping of drought impacts at the field level.

Second, the analysis presented here, similar to the work by Chere [16], relies on monthly temporal data, which may not capture short-term drought dynamics. Chere [16] suggests that future research could yield more precise results by investigating drought conditions at a finer temporal resolution, such as weekly, and by incorporating newly emerging algorithms like machine or deep learning to enhance predictive capabilities. Future studies should prioritize the integration of station-based observations and field surveys to validate remotely sensed indices and strengthen the robustness of spatiotemporal drought assessments.

Third, although the study discusses the potential influence of large-scale climate oscillations such as El Niño on regional drought severity, no dedicated statistical analysis of ENSO indices was conducted. As a result, any assumed teleconnections remain speculative and should be addressed in future work through explicit correlation analysis using established ENSO datasets [13].

Finally, while the multi-index framework captures the compound drivers of drought and their biophysical impacts, future research could expand its practical relevance by developing operational decision-support tools for local farmers, policymakers, and water managers. Integrating high-resolution monitoring with climate forecasts and adaptive cropping models would help translate spatiotemporal drought patterns into actionable strategies that strengthen the resilience of Hom Mali rice cultivation under intensifying climatic variability.

5. Conclusions

This study provides a detailed spatiotemporal assessment of drought conditions in the Thung Kula Ronghai region from 2001 to 2023, using multiple satellite-derived indices (the SPI, SPEI, RDI, AI) and NDVI anomalies. The results show that dry season droughts now affect up to 17 percent of the total agricultural area in some years, while severe drought zones cover more than 2.5 million hectares over the 20-year period. In the most recent 5 years, about 50 percent of cultivated land has experienced moderate to severe drought. Among the indices, the RDI showed the strongest correlation with NDVI anomalies (r = 0.22), highlighting its relative value for indicating vegetation stress. The combined index approach enabled the clear identification of high-risk sub-regions, especially in central Thung Kula Ronghai and lower Surin, where drought conditions recur with increasing frequency. These quantitative findings confirm the region’s vulnerability to prolonged dry-season water shortages and reinforce the need for site-specific irrigation planning and adaptive cropping strategies. The framework developed here can support improved drought monitoring, early warning, and decision-making to help protect Thailand’s climate-sensitive Hom Mali rice production in a warming climate.