This study investigates the complex interaction of biophysical and meteorological factors that drive evapotranspiration (ET) in saline environments. Leveraging a total of 182 cloud-free Landsat 5/8 time-series data from 1988 to 2019, we employed the Surface Energy Balance System (SEBS) model to quantify ET and investigate its relationships with soil salinity, vegetation cover, groundwater depth, and landscape metrics. We validated the predicted ET at two experimental sites using ET observation calculated by a water balance model. The result shows an R² of 0.78 and RMSE of 0.91 mm for the SEBS predicted ET, indicating high accuracy of the ET estimation. We detected abandoned saline farmland patches across Hetao and extracted the normalized difference vegetation index (NDVI), salinization index (SI), and the predicted ET for analysis. The results indicate that ET is negatively correlated with SI with a Pearson correlation coefficient (r) up to −0.7, while ET is positively correlated with NDVI (r = 0.4). In addition, we designed a control-variable experiment in the Yichang subdistrict to investigate the effects of groundwater depth, land aggregation index, soil salinity index, and the area of abandoned saline farmland patches on ET. The results indicate that increased NDVI could significantly enhance ET, while smaller saline farmland patches exhibited greater sensitivity to groundwater recharge, with higher averaged ET than larger patches. Moreover, we analyzed factor importance using Lasso regression and Random Forest (RF) regression. The result shows that the ranking of the importance of the features is consistent for both methods and for all the features, with NDVI being the most important (with an RF importance score of 0.4), followed by groundwater table depth (GWTD), and the influence of the surface area of abandoned saline farmland being the weakest. We found that smaller patches of abandoned saline farmland were more sensitive to changes in groundwater levels induced by nearby irrigation, affecting their averaged ET more dynamically than larger patches. Decreasing patch size over time indicates ongoing changes in land management and ecological conditions. This study, through a multifactor analysis of ET in abandoned saline farmland and its intrinsic factors, provides a reference for evaluating the dry drainage efficiency of abandoned saline farmland in a dry drainage system.

Data-driven upscaling of biogenic fluxes from eddy covariance (EC) sites to the global scale is a powerful complementary approach to process-based models for the derivation of global flux estimates. Nevertheless, significant uncertainties arise due to specific methodological choices such as data availability, data source differences, machine learning model differences, and feature selection. Accurately quantifying these uncertainties from diverse sources is essential for providing error estimates of the simulated fluxes. These uncertainties not only improve our general understanding of carbon cycle processes but also directly inform atmospheric inversions, which can use the upscaled net ecosystem exchange (NEE) as a prior. However, most existing data-driven global carbon flux products focus solely on flux estimates or provide incomplete uncertainty assessments limited to a few sources.In this study, we introduce a comprehensive framework for quantifying the uncertainties associated with carbon flux upscaling across potential sources. The framework involves three key steps: (1) pre-ensemble generation, (2) screening, and (3) uncertainty attribution. First, we construct ensemble members by training machine learning models with varying configurations, which include climate forcing datasets, feature combinations, subsets of EC sites, machine learning algorithms, and their hyperparameters. The experiments are supported by the recently developed data-driven modeling framework FLUXCOM-X, which enables a wide range of experiments with diverse methodological choices. We crafted a feature set that includes about 300 features to capture both current and historical state information. To capture the site representativeness uncertainty, we sample subsets from global EC sites based on geolocation and feature space. Additionally, we will also investigate different machine learning models and the variation of hyperparameters to generate the ensemble. Second, ensemble members that have a low contribution to the ensemble variance will be eliminated while we retain the most representative ones. We employ a feature selection algorithm, HybridGA, to screen important subfeature sets from near-infinite combinations. Moreover, we screen other ensemble members by assessing the distribution and spread of members. Finally, we will attribute uncertainties to various categories from the perspectives of machine learning and process-based modeling, and potential strategies to reduce these uncertainties are discussed. The framework is initially used to evaluate spatiotemporal NEE uncertain patterns in Europe, and will subsequently expand globally. Additionally, the estimated biogenic carbon flux uncertainty will be assessed with independent products. This work not only advances our understanding of the sources and patterns of upscaled flux uncertainties but also enhances the robustness of posterior estimates in atmospheric inversion models.