Estimation Of Surface Air Temperature From MODIS Land Surface Temperature And Its Application To The Study Of Urban Heat Island In The East China Metropolitan Area

With the rapid urbanization around the globe, the urban heat island (UHI) effects have received considerable attention by scientists. In this paper, high resolution MODIS daytime land surface temperature (LST), MODIS land cover types, DMSP/OLS nighttime light imagery and meteorological stations data were used to estimate the regional surface air temperature. Spatial patterns and variability of UHI over Yangtze River Delta area and Beijing-Tianjin-Tangshan metropolitan area were analized.(1) Based on the remote sensing data from MODIS and ground stations data, a simple statistical model was developed. The accuracy of the model was analyzed. Results show that the linear relationships between MODIS land surface temperature and stations daily maximum air temperature are significant. The2008-2009data was used for the training, and2001-2003data was used for accuracy assessment of the model. It is found that the error (mean absolute error) of the estimated surface air temperature is2℃-3℃. The estimated value has slightly lower error over Yangtze River Delta area than that over Beijing-Tianjin-Tangshan Metropolitan Area. For error analysis, we improve the MODIS land surface temperature data accuracy by using best quality values only, results show that it is successeful to increase the accuracy of the model to about2℃. However, this study has used LST data at all quality levels in develping the statistical model as using best quality data will reduce dramatically the valid data points.(2) By analyzing the seasonal variation of UHI in2009over Yangtze River Delta area and in2001over Beijing-Tianjin-Tangshan Metropolitan Area, results show that the UHI effect observed from the daytime Ts is most significant in the spring and summer; weaker in the autumn, and is not significant in the winter.(3) Analyzing the2001-2010summer MODIS LST and estimaed surface air temperature over Yangtze River Delta area and Beijing-Tianjin-Tangshan Metropolitan Area, we found that during last decade, the heat island areas over the Yangtze River Delta region are increased rapidly, leading to significant decreases in the size of temperature transition region. The strong heat island areas increase fastest, especially in the Suzhou-Wuxi-Changzhou city cluster. The number and size of UHI over the Yangtze River Delta region are grown very fast, connecting each other gradually and forming a giant heat island. At the same time, the heat island areas over Beijing-Tianjin-Tangshan Metropolitan Area region are also increased rapidly, especially in the Beijing-Langfang-Tianjin city cluster.(4) By analyzing the relationship between the change of temperature and the different urbanization area over the study areas, the change of mean temperature is highly correlated to the change of the nighttime lights, the spatial correlation coefficient between mean temperature and nightlight values was high, indicating that the human activity (urbanization) is highly responsible for the local temperature change.