A Four‐Parameter Model for Estimating Diurnal Temperature Cycle From MODIS Land Surface Temperature Product

Abstract

AbstractLand surface diurnal temperature cycle (LSDTC) is an important element in climate system. Diurnal temperature cycle (DTC) models are potential for deriving diurnal variation of land surface temperature (LST) over extended regions from satellite observations. To estimate the LSDTC at high spatial resolution from clear‐sky Moderate Resolution Imaging Spectrometer (MODIS) data, this study reduced the parameter number of a five‐parameter DTC model to only four by fitting the time of maximum temperature as a function of normalized differential vegetation index (NDVI), elevation and latitude. Using the ground‐based LSTs, the proposed four‐parameter DTC (FPD) model was evaluated for 48 and 4 available LST observations per day. The results showed that the root mean square errors (RMSEs) were 2.08 K, 0.88 K, 1.14 K, 1.31 K, 2.27 K, and 1.12 K for grassland, cropland, swamp meadow, desert steppe, desert shrub, and sand desert, respectively, for 48 observations. Comparatively, RMSEs for four observations corresponding to the MODIS overpass times were increased by less than 1 K, except for swamp meadow with an elevation greater than 3,500 m that the error was increased by 1.5 K. Applying the FPD model to MODIS LST product showed that the model could capture the variation of LSTs during a day, and the accuracy of the modeled LSTs depended on the accuracy of MODIS LST product. Generally, the proposed FPD model is simple and easy to be applied to MODIS four LSTs per day under clear skies to acquire LSDTC over extended regions with high spatial resolution.