The conditional nature of error estimates in multi-source satellite SIF cross-evaluation: How product combinations, statistical models, spatial scales, and temporal treatments shape the conclusions

Cross-evaluation results of multi-source satellite solar-induced chlorophyll fluorescence (SIF) products depend critically on the analytical choices made during evaluation, including product combinations, statistical models, and spatiotemporal preprocessing. This study constructs a diagnostic framework to systematically assess the methodological sensitivity of cross-evaluation. Using four representative SIF products from 2018 to 2021, we isolated the independent effects of product combination, error model (Three-Cornered Hat, TCH, vs. Quadruple Collocation Analysis, QCA), spatial aggregation scale, and seasonal-cycle treatment through controlled experiments. Direct comparison showed the closest agreement between the two TROPOMI products, whereas GOME-2 SIF showed the largest discrepancies. A key finding is that when a product combination includes two products that share observations from the same sensor, TCH produces a structural bias due to violation of the error-independence assumption: it systematically underestimates the errors of the shared-source products while overestimating the errors of the independent product. Allowing a prescribed non-zero error covariance between the shared-source pair, QCA mitigated this bias. Under QCA, more than 97% of valid vegetated grid cells for OCO-2 SIF, TROPOSIF Caltech , and TROPOSIF ESA had RMSE below 0.2 mW·m −2 ·sr −1 ·nm −1 , whereas more than 77% of GOME-2 grid cells fell within 0.2–0.4 mW·m −2 ·sr −1 ·nm −1 . Coarser-resolution products were more sensitive to spatial aggregation, while seasonal-cycle removal reduced error estimates for all products, with the largest reduction (~52%) for OCO-2 SIF. These results show that cross-evaluation error estimates reflect both product-specific error characteristics and the analytical framework under which they are derived. The framework is transferable to uncertainty analyses of other satellite-retrieved geophysical variables.