Quantifying diurnal variability of urban surface temperatures over the Los Angeles Basin using two years of composited ECOSTRESS thermal imagery.

Abstract:

The NASA ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) mission has a precessing orbital geometry that allows for frequent temporal sampling (~2 days) at times distributed over the full diurnal cycle with relatively fine spatial resolution (70 m). These data have the potential to vastly expand knowledge of spatiotemporal patterns of urban heat, as compared to measurements from polar-orbiting thermal imagers (which cannot resolve the diurnal cycle), or from ground-based radiometers (which cannot quantify spatial heterogeneity beyond a specific field site). In this study, we used used two years of ECOSTRESS imagery (n = 269) to construct composite diurnal time series of urban land surface temperature (LST) over urbanized Los Angeles County, USA. We used this time series to map the diurnal development of spatial contrasts in urban heat at sub-neighborhood scales. Spatial variability in LST over the course of a day was large, with the smallest diurnal range of LST (ΔLST = LSTday – LSTnight) observed in highly vegetated coastal regions (ΔLST = <15°C) and the largest diurnal range in impervious interior regions (ΔLST = 30-35°C). In addition, we used a ~0.2 m land cover classification and a ~0.7 m digital surface model to investigate spatial differences in the diurnal time course of LST that were driven by heterogeneity in topography, surface structure, and land cover. We found that the relative importance of these drivers of urban heat differed depending on time of day. Increased distance from the coast was positively correlated with warmer daytime LST (R = 0.43), but weakly correlated with cooler nighttime LST (R = -0.11). Increasing sub-pixel green cover fraction was correlated with cooler LST during the day (R = -0.41) and night (R = -0.30). This effect was stronger for tree (R = -0.32) than for grass covered areas (R = -0.26). Analysis of the diurnal patterns of cooling from sub-pixel green cover indicates peak cooling between 11:00 and 13:00, with rates of 0.73°C per 10% increase in green cover in spring-summer compared to 0.47°C per 10% increase in green cover in fall-winter.