Erosion of Natural Cooling Infrastructure in Southeast Asian Cities: Comparative Land Surface Temperature Effects of Wetland and Paddy Loss in Phnom Penh and Can Tho

Abstract Rapid urbanisation across Southeast Asia is displacing two distinct forms of natural cooling infrastructure: urban water bodies and peri-urban paddy fields. The land surface temperature (LST) consequences of each have rarely been quantified using consistent, satellite-based methods, and have never been compared directly within a single study design. This study compares urban wetland loss in Phnom Penh, Cambodia, and paddy conversion in Can Tho, Vietnam, across six anchor epochs (2000–2024), using Landsat-derived land-cover classification and a difference-in-differences design to isolate conversion-attributable LST anomalies from background warming. In Phnom Penh, water-body loss produced a DiD-corrected anomaly of 7.13°C, concentrated in a small number of large, documented infilling events. In Can Tho, paddy loss produced a smaller anomaly of 0.55°C, diffused across thousands of small, independent conversions. The comparative synthesis does not resolve to a single answer: measured as per-unit-area intensity, Phnom Penh's loss is roughly seven times more severe; measured as total city-wide thermal impact, Can Tho's far larger converted extent gives it the greater aggregate impact instead. This asymmetry reflects a deeper structural contrast in how the two cities lose cooling infrastructure, Phnom Penh through concentrated, large-scale events and Can Tho through diffuse, many small conversions, with direct implications for how mitigation should be designed in each context. The findings show that the thermal cost of cooling-infrastructure loss cannot be captured by a single metric common to both resource types: it depends on which resource is lost, how much is lost, and whether severity or scale is the question being asked. Keywords: Urban heat island; Land surface temperature; Wetland loss; Paddy field conversion; Southeast Asia; Remote sensing Article highlights: Phnom Penh water loss drives a 7.13°C DiD-corrected LST anomaly Can Tho paddy loss drives a smaller, diffuse 0.55°C LST anomaly Phnom Penh's per-area cooling-loss intensity is ~7x Can Tho's Can Tho's total city-wide thermal impact is ~3x Phnom Penh's instead Reproducible open-data workflow transfers to other SEA secondary cities