Validation of the TIMELINE AVHRR Land Surface Temperature (LST) Product with MODIS and in situ LST

Abstract

Land Surface Temperature (LST) is an important quantity describing the interface between long- and shortwave radiation fluxes on one side and turbulent heat fluxes and the ground heat flux on the other side. Therefore, it is a useful parameter for tracing the impact of changing climatic conditions on our environment. Satellite derived LST are an indispensable tool for monitoring these changes consistently over large areas and for detecting long term trends and patterns. Thermal Infrared (TIR) data from the AVHRR (Advanced Very High Resolution Radiometer) sensors on NOAA-satellites are operationally available since the early 1980s at 1 km spatial resolution. Several algorithms exist to derive LST from top of atmosphere brightness temperatures, which requires accounting for the attenuation due to of water vapor and for effects from land surface emissivity and sensor degradation. In the TIMELINE project (www.timeline.dlr.de), one aim is to retrieve accurate and consistent LST for the entire operating period of AVHRR sensors over Europe using the approach created by Frey et al. (2017) based on the Qin et al. (2001) and Becker & Li (1990) algorithms. In this study we present the validation results for the TIMELINE AVHRR LST product as well as preliminary analyses of patterns and trends in this multi-decadal data set. The validation approach consists of the following steps: a) an assessment of the consistency of the AVHRR LST time series between the different NOAA missions; b) an inter-comparison between AVHRR LST and in-situ LST obtained at 11 stations located in Europe (Evora, Donana), North America (7 surface radiation network (SURFRAD) stations) and Southern Africa (Gobabeb, Heimat); and c) a comparison of the AVHRR LST product to concurrent MODIS (Moderate Resolution Imaging Spectroradiometer) LST (MOD21 product). The comparisons were conducted on a daily basis for five different years (2000, 2010 - 2013). Only pixels satisfying a certain homogeneity criteria were selected. Furthermore, a maximum time difference of 10 minutes between AVHRR and MODIS acquisitions and in situ LST was permitted according to the suggestions from the CEOS LST Product Validation Best Practice Protocol. In total, 4000 in situ measurements and 1000 MODIS scenes were compared to the AVHRR LST product values. First results indicate a successful derivation of stable LST time series from multi- decadal AVHRR data. The comparison between the TIMELINE LST product and the selected data sets show diurnal and annual patterns as well as differences for individual land cover types. For 2010, on average, an absolute deviation of the AVHRR LST by 1.5 K from in situ LST as well as a difference of in average 1.5 K from the MODIS product was observed.