Exploring combinations of multi-temporal and multi-frequency radar backscatter observations to estimate above-ground biomass of tropical forest

Abstract

Approaches exploiting the complementary information on forest above-ground biomass (AGB) contained in multi-temporal and multi-frequency radar backscatter have hardly been explored in the tropics. Having available a multi-seasonal stack of air- and spaceborne X-, C-, L-, and P-band imagery for forest sites in Lope and Mondah, Gabon, we analyzed the sensitivity of backscatter at different frequencies to AGB under varying environmental imaging conditions, and the performance of an AGB retrieval when combining multi-temporal and multi-frequency backscatter observations. For the tropical forest sites in Gabon, which differ significantly with respect to forest composition and climatic conditions, we find that i) P-band allows for estimating AGB with the highest accuracy, ii) multi-temporal L-band backscatter may achieve accuracies close to what can be achieved with few P-band observations, iii) the use of multi-temporal observations is beneficial/crucial at all frequencies, in particular when the imaging conditions are persistently moist and the sensitivity of individual images to AGB reduced, and iv) combining P- and L-band backscatter allows for improving the AGB retrieval primarily when few P-band observations are complemented by a larger number of L-band observations. While the results of this study reemphasize that P-band is the most suitable frequency for mapping AGB in the tropics, they also advise consideration of the growing archives of spaceborne backscatter observations acquired at higher frequencies when mapping AGB in the tropics.