What’s happening with groundwater in my region?

What you can answer

Total water storage anomaly (TWS) trend from GRACE+GRACE-FO (2002–present)
Surface + root-zone soil moisture trend from SMAP (2015+)
Groundwater storage anomaly = TWS minus (surface + root-zone + snow + surface water)
Drought severity as deviation from 20-year mean
Irrigation-driven depletion in major aquifer systems (Central Valley CA, Indo-Gangetic, North China Plain, MENA)

What you can NOT answer with these alone

Point-scale well-water levels — GRACE’s 300 km native resolution can’t see your well
Aquifer-specific volumes without ancillary geology/well-log data
Future projections without a hydrological model (LIS or NASA NLDAS)
Sub-regional contrasts finer than 300 km (downscaling adds model assumptions)

Code template

import earthaccess
import xarray as xr
import numpy as np

earthaccess.login(strategy="netrc")

# Region: California Central Valley
aoi = (-122, 35, -119, 39)
window = ("2002-04-01", "2025-12-31")

# 1. GRACE+GRACE-FO mascon LWE
grace = earthaccess.search_data(short_name="GRACEFO_L3_JPL_RL06.X_M",
                                bounding_box=aoi, temporal=window)
# Earlier GRACE: "TELLUS_GRAC_L3_*"
grac_old = earthaccess.search_data(short_name="TELLUS_GRAC_L3_JPL_RL06_LND_v04",
                                   bounding_box=aoi, temporal=window)

# Open + concat (mind the 2017-2018 gap)
# Spatial mean over AOI → monthly TWS anomaly time series

# 2. SMAP L4 root-zone moisture (0-1m)
smap = earthaccess.search_data(short_name="SPL4SMAU", bounding_box=aoi,
                                temporal=("2015-04-01", "2025-12-31"))
# Convert volumetric soil moisture × depth → mm water-equivalent

# 3. Precipitation context
imerg = earthaccess.search_data(short_name="GPM_3IMERGM", bounding_box=aoi,
                                 temporal=window)

# 4. Compute groundwater storage:
#    GWSA ≈ TWS - root_zone_SM_anomaly - snow_anomaly - surface_water_anomaly
#    (assumes snow + SW small in this region; check assumption)

# 5. Plot TWS, RZSM, derived GWSA, precipitation — common x-axis

Expected output

Time-series chart: TWS anomaly · SMAP root-zone anomaly · derived GWSA anomaly (the key product) · monthly precipitation
Anomaly bar: each year’s GWSA deviation from 2004-2018 baseline
Map: GWSA trend (mm/yr) over the region 2002-2024

Caveats

300 km resolution is real. Don’t claim point-scale aquifer signals from GRACE.
The GWSA decomposition has known weaknesses in snow-dominated regions (LIS or NLDAS snowpack uncertainty propagates).
GRACE → GRACE-FO 2017-2018 gap needs careful handling — many studies use ad-hoc bridging.
Mascon vs spherical-harmonic solutions give slightly different answers; pick one and stick with it for trend studies.
Anthropogenic vs natural decomposition requires comparison to climate forcing — pure satellite data shows the integrated signal.