Groundwater components
Eight PCs, four ICs, and their spatial weights across 300 observation wells.
Browse component files ↗Why this dataset
Ground observations reveal the human signal.
GRACE captures regional water-storage change from space, but reconstruction is difficult where intensive groundwater withdrawal dominates the signal. This study incorporates multi-site groundwater levels to represent that missing human influence, improving case-study RMSE from 6.51 to 3.86 cm and NSE from 0.56 to 0.82.
Interactive explorer
Monthly reconstructed TWSA
Drag the timeline, press play, or select a map cell to inspect its full record.
North China Plain
Loading data…
Spatial mean
—
Water storage deficit
Water storage surplus
Temporal signal
NCP spatial mean
Click anywhere on the chart to change the displayed month.
Groundwater structure
From 300 wells to a compact regional signal.
Principal component analysis and independent component analysis summarize highly correlated groundwater-level observations. Explore the published component series below.
Principal component
PC 1
Published assets
Small files, explicit structure.
Every browser view is generated from the same versioned NumPy arrays available for direct download.
Kriged groundwater
168 monthly interpolated groundwater-level fields on the 24 × 26 domain grid.
Download .npy ↗Reconstructed TWSA
166 monthly case-5 fields with a stable 270-cell North China Plain mask.
Download .npy ↗Method overview
A transparent path from observations to reconstruction.
- 1
Represent groundwater
PCA and ICA extract regional features from multi-site groundwater levels.
- 2
Align space and time
Ordinary Kriging maps point observations; STL separates trend and detrended signals.
- 3
Learn complementary patterns
Linear regression models trend while random forest models nonlinear variability.
- 4
Reconstruct and evaluate
The components are combined and assessed against withheld GRACE observations.
Cite this work
Li, P., Zha, Y., & Tso, C.-H. M. (2023). Reconstructing GRACE-derived terrestrial water storage anomalies with in-situ groundwater level measurements and meteorological forcing data. Journal of Hydrology: Regional Studies, 50, 101528.