The Impact of Temporal Aggregation of Land Surface Temperature Data for Urban Heat Island Monitoring

Temporally composited remote sensing products are widely used in monitoring the urban heat island (UHI). In order to quantify the impact of temporal aggregation for assessing the UHI, we examined MODIS land surface temperature (LST) products for 11 years focusing on Houston, Texas and its surroundings. By using the daily LST from 2000 to 2010, the urban and rural daily LST were presented for the 8-day period and annual comparisons for both day and night. Statistics based on the rural-urban LST differences show that the 8-day composite mean UHI effects are generally more intensive than that calculated by daily UHI images. Moreover, the seasonal pattern shows that the summer daytime UHI has the largest magnitude and variation while nighttime UHI magnitudes are much smaller and less variable. Regression analyses enhance the results showing an apparently higher UHI derived from 8-day composite dataset. The summer mean UHI maps were compared, indicating a land cover related pattern. We introduced yearly MODIS land cover type product to explore the spatial differences caused by temporal aggression of LST product. The mean bias caused by land cover types are calculated about 0.5 ~ 0.7K during the daytime, and less than 0.1K at night. The potential causes of the higher UHI are discussed. The analysis shows that the land-atmosphere interactions, which result in the regional cloud formation, are the primary reason.