Surface soil moisture (SM) is the keystone variable of terrestrial ecohydrology. Yet, the rapid diversification and development of deep learning architectures for satellite SM estimation have outpaced practitioners’ capacity to select among them. This review synthesizes 37 deep learning studies from the SMAP era (2015–2026) across five architecture families (MLP and physics-informed neural networks [MLP/PINN], long short-term memory [LSTM] and gated recurrent unit [GRU] networks, convolutional neural networks [CNN], convolutional LSTM and graph neural networks [GNN], and Transformer-based models) to establish an architecture–task-matching framework that links each family to its dominant estimation niche. The analysis reveals consistent specializations: MLP/PINN models achieve competitive surface SM retrieval from satellite inputs; recurrent networks extend SMAP temporally (RMSE ≤ 0.035 ${\mathrm{m}}^3\hspace{0.17em}{\mathrm{m}}^{-3}$); CNN disaggregates SMAP to 1 km (reported unbiased root-mean-square error (ubRMSE) approaching 0.04 ${\mathrm{m}}^3\hspace{0.17em}{\mathrm{m}}^{-3}$); ConvLSTM and GNN address spatiotemporal gap-filling (low reported ubRMSE 0.022 ${\mathrm{m}}^3\hspace{0.17em}{\mathrm{m}}^{-3}$); and Transformers enable global multi-source fusion and decadal climate-scenario projection. Across all families, four physics-DL integration modes (hard architectural constraints, soft loss-function penalties, physics-as-input feature engineering, and physics-ML hybrid output fusion) consistently yield RMSE reductions of 8–50% relative to data-driven baselines. These findings provide a practitioner-oriented framework that is applicable to ecohydrological monitoring of plant water stress, agricultural drought, early flood warnings, and land–atmosphere coupling.

Coupling coordination degree (CCD) between persistent late-night radiance and population density may vary nonlinearly with urban functional density, yet linear and single-city analyses cannot distinguish shared from city-specific patterns. A method-controlled dual-city XGBoost-SHAP framework was applied to 14,215 H3 cells in Shanghai and 31,067 in Beijing using six point-of-interest density factors. XGBoost outperformed OLS, with random hold-out R² values of 0.899 and 0.918 vs. 0.617 and 0.626. Public service (X2) ranked first in both cities, accounting for 39.77% and 58.73% of total mean absolute SHAP magnitude. The secondary hierarchy diverged as follows: commercial finance (X3) ranked second in Shanghai at 27.91% and formed the strongest interaction with X2, whereas transport infrastructure (X6) ranked second in Beijing at 18.63% and formed the strongest interaction with X2. Nonlinear analysis identified reproducible negative-to-positive crossings for X2, X3, and X6, peak-type responses for X4 and X5, and no stable second saturation threshold. Spatial OOF, grid, rank, LOWESS, and residual checks supported the leading-factor contrast while showing scale sensitivity and residual spatial dependence. The results identify a common leading attribution alongside city-specific secondary, nonlinear, and spatial patterns within the two observed megacities.