Spatiotemporal Characterization and Analysis of Drought in Mainland Southeast Asia: Unlocking the Potential of Long Earth Observation Time Series

Drought is a complex and reoccurring hazard event. It ranks among the costliest natural climate-related disasters and manifests in various forms, posing significant challenges in mapping its characteristics and assessing its widespread impacts. Mainland Southeast Asia (MSEA), a major region of tropical agriculture and vegetation ecosystems, has become increasingly susceptible to drought hazards. As global warming continues, the frequency and severity of droughts in this region are projected to intensify, presenting risks to food security, biodiversity, and local economies. Despite the urgency of addressing these challenges, the limited availability of publicly accessible, long-term ground-based climate data in the study region hampers the development of an accurate and consistent region-wide drought monitoring system. This data gap makes tracking, predicting, and mitigating drought impacts more challenging. In this context, satellite-based data offers a great alternative. Satellite-based measurements can provide frequent and consistent records of land surface variables at a high spatial resolution over large and remote areas. Many satellite sensors, such as the Moderate Resolution Imaging Spectrometer (MODIS), enable the detection and monitoring of long historical drought conditions, overcoming the limitations of ground-based systems. This dissertation aims to harmonize long-term satellite observations for monitoring and assessing drought conditions at regional and local scales. Specifically, this thesis first provided a detailed and comprehensive review of satellite-based drought remote sensing. Subsequently, a monitoring framework for mapping vegetation-based drought characteristics was developed, facilitating more automatic and consistent observations. The research then examined the responses of vegetation-based drought to multi-temporal climate and soil drought conditions, uncovering critical relationships between these variables. Finally, this thesis conducted a local scale assessment of agricultural drought vulnerability in a drought-prone region. A detailed review of 102 satellite-based drought studies within the region revealed a significant increase in drought articles over the study period (2000-2021), especially over the last eight years. Vietnam and Thailand had the largest number of drought articles, accounting for nearly 40%. Most studies focused on monitoring drought conditions using a single vegetation-based or precipitation-based drought indicator (~53%). Agricultural and vegetative droughts received the most attention, but long-term drought studies and deep-learning methods were still limited. Notably, no studies have offered spatial details of drought characteristics derived from optical sensors. Also, little research has been conducted to investigate the responses of vegetation to multitemporal climate-soil drought conditions and assess drought vulnerability. Given these research gaps, this analysis generated a gap-free monthly vegetation time series derived from the MODIS data. A developed framework of mapping vegetation-based drought characteristics enabled the automatic identification of the severe drought years and their characteristics from 2000 to 2021. Apart from drought characteristics, the seasonal trends of vegetation-based drought were identified using the Mann-Kendall test and Theil-Sen slope estimator. The findings of this analysis indicated several severe drought years in the study region, namely 2000, 2004-2005, 2010, 2016, and 2019-2020. Higher drought duration and frequency were primarily detected in Central Myanmar, Cambodia, eastern Thailand, and the Lower Mekong Delta. Central Myanmar witnessed the highest drought occurrence rate (~ 60%), while other countries experienced nearly 40%, except for Laos. Thailand and Vietnam experienced nearly 15 prolonged drought events over the last two decades. Recent analyses indicate that the longest drought events have primarily been detected in Cambodia and Laos, where increased dry trends have been observed. Land use types and elevation significantly influence drought conditions derived from vegetation-based data, underscoring the complex interactions between environmental factors and drought severity. Subsequently, vegetation-based drought conditions were examined with multitemporal climate-soil drought indices, representing short-term, median-term, and long-term drought conditions. This analysis revealed a significant positive relationship between vegetation-based drought and climate-soil drought indices. Stronger drought responses were observed in lower elevation zones, especially in rainfed cropland, mixed forests, and deciduous forests. In contrast, evergreen forest and irrigated crop vegetation responded less to drought indices, especially in high-elevation zones. Short-term drought disturbances accounted for nearly 93% of the observed variations in the natural and undisturbed vegetation ecosystems among the twelve climate-soil drought indices. In addition, this study assessed agricultural drought vulnerability in the drought-prone region of Vietnamese Central Highlands by integrating time series vegetation-climate-soil drought indices with socioeconomic data. The analysis followed the vulnerability framework established by the Intergovernmental Panel on Climate Change (IPCC), focusing on three key components: exposure, sensitivity, and adaptive capacity. The findings revealed that exposure, largely driven by soil moisture deficits, accounted for nearly 65% of the agricultural drought vulnerability. The highest levels of vulnerability were concentrated in the central areas of Dak Lak and Gia Lai provinces, where drought exposure and sensitivity were elevated. In contrast, Lam Dong province exhibited lower drought vulnerability despite high exposure, thanks to its lower sensitivity and stronger adaptive capacity, highlighting the critical role of socioeconomic resilience in mitigating drought impacts.