Hapke-Hsr + Marmir-2: An Improved Soil Reflectance Model

Abstract

Soil reflectance models play a crucial role in agriculture, ecology, hydrology and climate systems. Despite the development of several soil reflectance models in the last decade, further improvement is necessary for better accuracy, particularly in characterizing the dependence of the soil moisture content (SMC). In this study, we first proposed a method to enhance the latest version of the Hapke model (i.e., the Hapke-HSR model) using dry soil spectral reflectance. Then, we coupled the improved Hapke-HSR model with the multilayer radiative transfer model of soil reflectance (MARMIT-2) (i.e., the Hapke-HSR + MARMIT-2 model) to characterize the soil spectral reflectance. Finally, we verified the performance of the Hapke-HSR + MARMIT-2 model using eight different soil databases. The results indicate that all the Hapke-HSR, MARMIT-2, and Hapke-HSR + MARMIR-2 models perform well (NRMSE < 9%) for various soil databases, while the Hapke-HSR + MARMIT-2 model shows better accuracy (R2 = 0.993 and RMSE = 0.007) than the Hapke-HSR (R2 = 0.909, RMSE = 0.028) and MARMIT-2 models (R2 = 0.983, RMSE = 0.012), especially at SMC ≥ 30%. The improved soil reflectance model complements the advantages of the Hapke-HSR and MARMIT-2 models and has potential applications in various fields.

Keywords: Soil, Hyperspectral reflectance, BRDF, Hapke model, MARMIT-2 model, Hapke-HSR + MARMIT-2 model, Soil moisture content