Soe Win Myint

Earth observation (EO) is entering a petabyte-scale era in which multi-sensor satellite archives capture environmental processes across broad spatial, spectral, and temporal dimensions. EO data-processing workflows, however, still rely heavily on task-specific supervised models that require large amounts of labeled data and may generalize poorly when applied to new regions, time periods, sensors, or environmental conditions. This review provides a structured synthesis of the rapidly evolving literature on EO foundation models (EO-FMs), which seek to address these limitations through large-scale pretraining and efficient downstream adaptation. We organize the field of EO-FMs through a data-objective-architecture-adaptation-evidence lens that links training data and harmonization, pretraining objectives, backbone design, adaptation strategies, and empirical evidence across downstream applications. We examine representative model families including Prithvi-EO-2.0, Copernicus-FM, TerraFM, DOFA, SatCLIP, EarthPT, and Aurora, and compare their capabilities across tasks, regions, time periods, and sensing modalities. Across the literature, EO-FMs show strong potential for improving label efficiency, supporting multimodal learning, and enabling the reuse of pretrained representations across applications such as land-cover mapping, hazard response, ecosystem monitoring, urban climate analysis, and Earth-system forecasting. At the same time, empirical evidence on downstream performance, label efficiency, and generalization remains heterogeneous across datasets, sensing modalities, metrics, and benchmark protocols, which limits direct model ranking and underscores the need for task-aware interpretation. We argue that the next phase of EO-FM research should move beyond proof-of-concept gains toward more rigorous and capability-aware benchmarking, explicit assessment of predictive uncertainty and calibration, transparent preprocessing and adaptation procedures, and geographically and temporally robust evaluation. Meeting these requirements would strengthen the reliability, comparability, and scientific utility of EO-FMs for scalable and decision-relevant Earth observation analysis.

Urban heat exposure has been demonstrated to be distributed disproportionately across socio-demographic groups. Most existing studies assess heat exposure based on static residential locations. However, human mobility can significantly reshape the heat exposure patterns that people actually experience. Although recent studies have incorporated mobility, they rarely examine how mobility interacts with neighborhood socio-demographic structure and the mechanisms through which it exacerbates heat inequality. This study addresses these gaps by integrating origin-destination mobility data with remotely sensed Land Surface Temperature (LST) to develop a Mobility-based Heat Burden (MHB) metric. We apply K-means clustering to classify neighborhoods into four socio-demographic types in Phoenix, AZ, and examine how mobility interacts with neighborhood structure to reshape heat exposure. The results reveal significant disparities in residence-based heat exposure across socio-demographic clusters. When mobility is incorporated, MHB is consistently lower than residence-based LST across all clusters, and inter-cluster disparities become narrower, indicating that daily mobility partially mitigates average heat exposure at the city scale. However, this overall reduction does not eliminate inequality. Disadvantaged neighborhoods continue to experience the highest mobility-adjusted heat burden, while origin-destination analysis reveals strong high-to-high mobility patterns in which residents from hotter neighborhoods disproportionately travel to similarly hot destinations. These patterns constitute localized ”heat traps” that sustain recurrent exposure across residential and activity spaces. These findings suggest that urban heat mitigation should move beyond residence-based planning to incorporate mobility, activity spaces, and equity-oriented adaptation strategies. In particular, interventions should prioritize mobility heat traps and high-use destinations within vulnerable populations’ daily movement networks.

Wildfire burn severity is influenced by the interplay between short-term hydroclimatic extremes and long-term ecosystem degradation. However, most models emphasize short-term pre-fire conditions, neglecting multi-decadal landscape change that preconditions areas for wildfire. To address this gap, we analyzed January 2025 Palisades, Eaton, and Hughes fires in Los Angeles County by integrating short-term and long-term pre-fire predictors. Burn severity was quantified using differenced Normalized Burn Ratio (dNBR) from Sentinel-2 images. Short-term predictors capture immediate pre-fire hydroclimatic stress in 2024, including ECOSTRESS Evapotranspiration (ET), Evaporative Stress Index (ESI), Land Surface Temperature (LST), and Landsat-derived Enhanced Vegetation Index (EVI) and Moisture Stress Index (MSI). Long-term predictors captured persistent changes in vegetation conditions and hydroclimatic stress from 2000 to 2023 using Sen’s slope from Landsat composites. We evaluated predictor importance and model performance using Random Forest (RF), Support Vector Regression (SVR), and Artificial Neural Networks (ANN) under short-term, long-term, and coupled configurations. The short-term model reveals that the most intense burning occurred in areas with relatively cooler temperatures, high fuel availability, and acute moisture stress before ignition. High-severity burn areas showed persistent declines in ET and ESI and weak or negative EVI trends, consistent with multi-decadal drying, drought intensification, and progressive vegetation degradation. Coupled models consistently outperformed single-timescale models, with RF performing best (R² = 0.478). Temperature- and moisture-related indicators, together with long-term ET and ESI trends, were dominant predictors. These results show that wildfire severity is better explained by cross-timescale interactions, improving time-aware and actionable wildfire risk assessment under rapid climate change.

Desert cities in the U.S. Southwest experience the nation’s highest aerosol burdens, but how air pollution co-evolves with hydroclimatic–LULC metrics and racial/ethnic exposure dynamics remains unclear. We developed 30m Landsat-derived summer aerosol optical depth (AOD) (2000-2020) for Phoenix, used explainable machine learning to attribute AOD trends to land-surface changes, and linked them to racial/ethnic disparities. We computed pixel-wise Sen’s slope for AOD, land surface temperature (LST), actual evapotranspiration (ETa), and NDVI, and derived LULC transitions. These Sen’s slope metrics were aggregated to census block groups to explore racial/ethnic exposure disparities. Random Forest with SHapley Additive exPlanations (SHAP) quantified contributions of hydroclimatic-LULC changes to AOD trends. Results show citywide AOD rose, shifting from very clean to moderate/elevated levels with concentrations in the urban core and rapidly urbanizing corridors, opposite national declines. NDVI slope decrease (greenness loss) was the main driver of AOD increases, followed by ETa decline (surface moisture loss). LST effects were context-dependent, and LULC transitions further amplify AOD increases. SHAP dependence curves revealed threshold-and-saturation dynamics, where near-zero NDVI/ETa slopes, small positive shifts rapidly reduce AOD slope, but the benefits flattened at higher ETa gains. Racial/ethnic minority communities are disproportionately located in low NDVI/ETa, high LST regimes and experience higher AOD and steeper increases, yet also occupy high-leverage zones where improving vegetation cover and surface moisture can produce disproportionate reductions in AOD. This integrated climate-land surface-society framework supports equity-centered, threshold-saturation-aware actions that prioritize limited resources in high-leverage zones to achieve co-benefits for air quality and environmental justice.

Southeast Asia’s (SEA) rapid urbanization has intensified urban heat effects. While previous studies address urban warming at specific time points, the dynamic relationship between land cover and land use (LCLU) changes and land surface temperature (LST) along urban-rural continuums remains understudied. This study examines LST, LCLU, and NDVI changes across 19 SEA cities of varying urbanization levels from 2000 to 2022 using Landsat images. Results show that LST change rates differ by urbanization levels, with built-up lands driving urban warming. Larger cities exhibited more intense and widespread LST increases extending into rural areas. The megacities experienced significant LST growth within 5 km of city centers despite limited built-up expansion. However, some areas in many cities experience decreasing trends of heat. NDVI significantly influenced LST changes through its cooling effect and the warming impact due to vegetation loss caused by urban expansion. Findings support urban planning strategies that balance growth with enhancing green spaces, redirecting growth to smaller cities, and shifting development from city centers to urban outskirts.

Extreme weather is associated with a variety of water quality issues that can pose harm to humans and aquatic ecosystems. Under dry conditions, contaminants become concentrated in streams with a greater potential for harmful algal blooms, while wet conditions can cause flooding and broadcast pollution. Developing interventions to improve water quality in a changing climate requires a better understanding of how hydroclimatic variability affects watershed processes, and which places are most vulnerable. We developed a Soil and Water Assessment Tool model of the Cape Fear River Basin (CFRB) in North Carolina, USA, representing contemporary land use, point and non‐point sources, and weather conditions from 1979 to 2019. The CFRB is a large, complex river basin undergoing urbanization and agricultural intensification, with a history of droughts and floods. To identify intervention priorities, we developed a Water Quality Risk Index (WQRI) using the average and variability of contaminant loads across dry, normal and wet conditions. The landscape generally contributed the majority of pollutants (e.g., via erosion, fertilizer and manure applications), including 90.1% of sediment, 83.2% of total nitrogen, and 52.4% of total phosphorus at the City of Wilmington's drinking water intake, yet point sources were influential during dry periods. Approximately 16% of the watershed contributed most of the pollutants across conditions—these represent priority locations for interventions such as restoration, urban, or agricultural best management practices. The WQRI approach considering risks to water quality across conditions can help identify locations where interventions are more likely to improve water quality under climate change.

Cities in warmer climate zones have experienced extreme heat and unbearable drought conditions due to climate change. Developing a practical and integrated framework to tackle the trade‐off between land surface temperature (LST) reduction and water conservation is essential for heat mitigation and resilience planning. We developed a multi‐objective framework of spatial optimization for priority areas that considers environmental justice. We identified the priority areas (i.e., residential districts, socio‐economically disadvantaged neighborhoods, hotspot regions, and opportunity areas) using ECOSTRESS‐based LST, actual evapotranspiration (ETa, as a proxy to outdoor water use [OWU]), Landsat‐based LST and ETa changes (2000–2020), and the evaporative stress index. The configuration of vegetation was spatially optimized to achieve the highest co‐benefits by equally balancing both objectives. We found that LST decreased and OWU increased as the vegetation percentage rose from 25% to 45%. The trade‐off between heat reduction and water conservation indicates that LST is not significantly reduced, as OWU reaches above the mid‐level if we focus mainly on LST reduction. The associated green space configurations suggest that grass coverage gradually increases as tree coverage decreases slightly in unmanaged soil. Overall, hotspot regions have a higher potential to lower LST by consuming less water than other areas with the same percentage of new vegetation. Based on this perspective, we suggest that urban planners and city managers prioritize hotspot regions by considering environmental justice and co‐benefits to ensure highly efficient LST amelioration and OWU conservation.

Background: Transitional economies in Southeast Asia—a distinct group of developing countries—have experienced rapid urbanization in the past several decades due to the economic transition that fundamentally changed the function of their economies, societies and the environment. Myanmar, one of the least developed transitional economies in Southeast Asia, increased urbanization substantially from 25% in 1990 to 31% in 2019. However, major knowledge gaps exist in understanding the changes in urban land use and land cover and environment and their drivers in its cities. Methods: We studied Yangon, the largest city in Myanmar, for the urbanization, environmental changes, and the underlying driving forces in a radically transitioned economy in the developing world. Based on satellite imagery and historic land use maps, we quantified the expansion of urban built-up land and constructed the land conversion matrix from 1990 through 2020. We also used three air pollutants to illustrate the changes in environmental conditions. We analyzed the coupled dynamics among urbanization, economic development, and environmental changes. Through conducting a workshop with 20 local experts, we further analyzed the influence of human systems and natural systems on Yangon’s urbanization and sustainability. Results: The city of Yangon expanded urban built-up land rapidly from 1990 to 2000, slowed down from 2000 to 2010, but gained momentum again from 2010 to 2020, with most newly added urban built-up land appearing to be converted from farmland and green land in both 1990–2000 and 2010–2020. Furthermore, the air pollutant concentration of CO decreased, but that of NO₂ and PM_2.5 increased in recent years. A positive correlation exists between population and economic development and the concentration of PM_2.5 is highly associated with population, the economy, and the number of vehicles. Finally, the expert panel also identified other potential drivers for urbanization, including the extreme climate event of Cyclone Nargis, capital relocation, and globalization. Conclusions: Our research highlights the dramatic expansion of urban land and degradation of urban environment measured by air pollutants and interdependent changes between urbanization, economic development, and environmental changes.

Forest ecosystems play an indispensable role in addressing various pressing sustainability and social-ecological challenges such as climate change and biodiversity loss. However, global forest loss has been, and still is today, an important issue. Here, based on spatially explicit data, we show that over the past 60 years (1960–2019), the global forest area has declined by 81.7 million ha (i.e. 10% more than the size of the entire Borneo island), with forest loss (437.3 million ha) outweighing forest gain (355.6 million ha). With this forest decline and the population increase (4.68 billion) over the period, the global forest per capita has decreased by over 60%, from 1.4 ha in 1960 to 0.5 ha in 2019. The spatiotemporal pattern of forest change supports the forest transition theory, with forest losses occurring primarily in the lower income countries in the tropics and forest gains in the higher income countries in the extratropics. Furthermore, economic growth has a stronger association with net forest gain than with net forest loss. Our results highlight the need to strengthen the support given to lower income countries, especially in the tropics, to help improve their capacity to minimize or end their forest losses. To help address the displacement of forest losses to the lower income countries in the tropics, higher income nations need to reduce their dependence on imported tropical forest products.