DOF_SM: DINEOF+ Derived Observation-Faithful Daily 1 km Surface Soil Moisture over the Contiguous United States (2015-2025)

Le résumé fourni par la source

This dataset provides gap-free daily surface soil moisture (0–5 cm, m³/m³) at 1 km spatial resolution over the Contiguous United States (CONUS) from April 2015 through December 2025. It is named DOF_SM (DINEOF+ derived Observation-Faithful Soil Moisture) to reflect its core design principle: gap-filling via spectral reconstruction without injecting model output, preserving the observational fidelity of the underlying SMAP retrievals. Algorithm. DOF_SM is generated using the DINEOF+ (Data Interpolating Empirical Orthogonal Functions Plus) algorithm applied to SMAP THySM 1 km retrievals (SPL2SMAP_S, NSIDC). DINEOF+ reconstructs missing values via iterative truncated Singular Value Decomposition (SVD) on space × time data matrices. Before reconstruction, frozen soil periods are pre-masked using the SMAP L3 SPL3SMP_E freeze/thaw flag to ensure physically invalid retrievals are treated as structural gaps. A connectivity masking step removes isolated reconstructed pixels lacking spatial or temporal neighbors in the original observations. The raw THySM product has 41–99% daily spatial gaps over CONUS; DOF_SM produces year-round seamless coverage. File format. Eleven annual NetCDF4 files (CF-1.8 compliant), named dof_sm_YYYY.nc (2015–2025). Each file contains: soil_moisture [time, lat, lon] — float32, m³/m³, fill value −9999.0 reconstruction_confidence [time, lat, lon] — uint8 flag: 0 = masked/invalid, 1 = LOW (<25% neighbors observed), 2 = MEDIUM (25–50%), 3 = HIGH (≥50%) Compressed with zlib level 4. Total size ~130 GB. Grid: 2,637 × 5,640 (CONUS), EPSG:4326. Validation. Validated against 805 in-situ stations from 11 ISMN networks (SCAN, USCRN, SNOTEL, PBO_H2O, TxSON, COSMOS, ARM, CW3E, FLUXNET-AMERIFLUX, iRON, SOILSCAPE) over 2015–2025. Overall Pearson r = 0.541, RMSE = 0.101 m³/m³, Bias = +0.011 m³/m³, KGE = 0.537 (n = 1,175,945 station-days). Frozen soil periods excluded from all statistics. Recommended uses. Agricultural drought monitoring, flash drought detection, crop yield forecasting, precision irrigation scheduling, hydrological model inputs and evaluation, climate trend analysis over CONUS. Not recommended for. Sub-daily applications, frozen soil/cold-season periods, dense forest canopy. Related publication. Zhao et al. (in preparation), A gap-free daily 1 km surface soil moisture dataset over the contiguous United States (2015–2025) derived from SMAP THySM using DINEOF+, Earth System Science Data. Algorithm reference. Zhao et al. (2025), Validating Data Interpolation Empirical Orthogonal Functions (DINEOF+) Interpolated Soil Moisture Data in the Contiguous United States, Agriculture. Zhao et al. (2025), Reconstructing SMAP-Derived 1-km Soil Moisture Dataset Using DINEOF+ Algorithm. Source data. SMAP THySM, CropCASMA.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.