A simple yet robust framework to estimate accurate daily mean land surface temperature from thermal observations of tandem polar orbiters

Remotely sensed and accurate daily mean land surface temperature (T-dm) is valuable for various applications such as air temperature estimation and climate change monitoring. However, most traditional methods employed by the remote sensing community estimate T-dm by averaging the - usually few - observed cloud-free land surface temperatures (LSTs). Such estimates can have large sampling bias, especially for tandem polar orbiters, due to their sparse sampling of diurnal LST dynamics and the unavailability of under-cloud LSTs. To estimate accurate T-dm based on thermal observations from tandem polar orbiters, here we propose a simple yet robust framework that combines the annual temperature cycle (ATC) and the diurnal temperature cycle (DTC) models (termed the ADTC-based framework). The ATC model is used to reconstruct daily instantaneous undercloud LSTs, based on which the DTC model is employed to establish diurnally continuous LST dynamics for estimating T-dm. The proposed framework is validated with geostationary LST observations and in-situ thermal measurements under both cloud-free and overcast conditions. The validations show that, under cloud-free conditions, the ADTC-based framework is able to reduce the positive sampling bias obtained with simple averaging (> 2.0 K) and yields a mean absolute error (MAE) of approximately 0.5 K. Under overcast conditions, the ADTC-based framework yields MAEs of 1.0 K and 0.5 K at the daily and monthly scales, respectively. Furthermore, a contribution analysis indicates that the ATC model reduces the MAE from around 4.2 K to 2.0 K while the DTC model reduces the MAE from around 2.0 K to 1.0 K. Based on our validation results and tests performed with MODIS data, the presented simple yet robust ADTC-based framework is able to accurately estimate large-scale spatiotemporally continuous T-dm from thermal observations of tandem polar orbiters. Therefore, the ADTC-based framework is a potentially valuable tool for many related applications.