Quantifying the ability of bidirectional reflectance distribution function (BRDF) model to Respond to soil moisture and the normalized difference vegetation index (NDVI)

• The proposed algorithm uses dynamic land surface parameters for BRDF characterization, achieving optimal accuracy-efficiency balance. • Transfer Function Analysis identifies land surface parameters as the main annual drivers of surface reflectance, while kernel geometries dominate short-term variations, consistent with theoretical and validation results. • An operational solution for ill-posed BRDF inversions supplements to mainstream algorithms, improving inversion efficiency for MODIS products in cloudy regions like South Asia. A so-called Ross-Thick-Li-Sparse reciprocal (RTLSR) model, regarded as the prototype in this research, is adjusted with adopting dynamically multi-land surface parameters (RTLSR_MP) including the normalized difference vegetation index (NDVI) and soil moisture (SM) for characterizing land surface bidirectional reflectance precisely. Through comprehensive evaluation of RTLSR_MP and the classic BRDF retrieval algorithms (e.g., MCD and VJBBF) across global and ESAsia experiments, the distinct performance characteristics are revealed: (1) Treating high-quality observed MODIS SR as the reference truth, the optimal RTLSR_MP variant (e.g., Pro2), MCD43, and VJBBF all demonstrate reliable precision in both Global and ESAsia retrievals. Specifically, Pro2 achieves an average RMSE% of 4.62%±4.5% (Global, 76-day inversion period) and 7.81%±1.30% (ESAsia, 268-day inversion period) across MODIS Bands 1–7—comparable to MCD43′s 8.32%±6.57%/7.0%±17.2% and VJBBF’s 7.95%±1.65% (ESAsia) (note: RMSE% is calculated as follows: daily RMSE% per band for pixels with valid results from all methods, averaged across the full inversion period, then aggregated across Bands 1–7). (2) Pro2 successfully remedies VJBBF and MCD43C1 full inversion’s deficiencies in retrieval efficiency and accuracy across mid-low latitudes; the no-cloud MODIS band 1 observation on 1st / 197th day of 2012increased from 20%/17% of ESAsia area to of 38.9%/44.8%, 37.3%/35.4% and 36.6%/ 47.9% while filled by MCD full inversion, VJBBF and Pro2 orderly, with the maximum RMSE% being less than 12%. (3) Then, a Transfer Function Analysis (TFA) tries to quantify the high-quality reconstructed MODIS band 2 SR’s response to NDVI, SM and solar-viewing geometry in a new perspective, which also is consistent with the validating reliability and the theoretical soundness of RTLSR_MP, MCD and VJBBF’s frameworks from the perspective of cross spectral analysis. In general, RTLSR_MP scheme is an exploration of remedying for the ill-posed nature inherent in current BRDF model retrievals.