Downloading Data
Downloading a CSV
With a region or a cell selected, click Download CSV in the upper right of the chart panel. The file contains the monthly values of all
five layers for that region or cell, whichever ones are plotted. It is named grace_gwsa_data.csv for a region (the middle part follows the
displayed layer) or grace_cell_<lat>_<lon>_data.csv for a single cell.
| Column | Contents |
|---|---|
Date |
First day of the month, YYYY-MM-DD |
GWSa, TWSa, SMa, SWEa, CANa |
Area-weighted mean anomaly for the month, cm of liquid water equivalent |
<layer>_upper, <layer>_lower |
The mean plus and minus 1σ |
GWSa_filled, TWSa_filled |
The same series with its gaps filled by the seasonal model; equal to the observed value in every other month |
GWSa_is_filled, TWSa_is_filled |
1 in months the seasonal model filled, 0 otherwise |
For example, the start of the gap between missions for the Northern Midwest Aquifer System:
Date,GWSa,GWSa_upper,GWSa_lower,GWSa_filled,GWSa_is_filled,TWSa,...
2017-06-01,9.529,14.913,4.146,9.529,0,12.536,...
2017-07-01,,,,11.146,1,,...
2017-08-01,,,,12.694,1,,...
The file has one row for every month from April 2002 to the latest release. In months with no GRACE data, the GWSa and TWSa cells are blank,
while the GLDAS layers (SMa, SWEa, CANa) still have values, because the land surface models run every month. The _filled columns
supply a value for those months (see Gaps in the Record). They are written whatever the Gap filling setting, so
the file is the same however the chart is drawn.
Excel, Google Sheets, Python and R all read the file directly, and the ISO dates need no conversion (in pandas, use
pd.read_csv(path, parse_dates=["Date"])). To get
the uncertainty σ back from the bounds, take (upper − lower) / 2.
Converting to volume
The anomalies are depths of water averaged over the region. Multiply by the region's area to get a volume:
ΔV (km³) = GWSa (cm) × Area (km²) / 100,000
For example, a GWSa of −12 cm over a 400,000 km² aquifer is 12 × 400,000 / 100,000 = 4.8 km³ (4,800 million m³) less water than the 2004–2009 average. The change between two dates is the difference of their anomalies times the area, and a trend in cm/yr times the area gives a rate of loss or gain in km³/yr.
Use the area of the region as you defined it. For an uploaded region, a GIS can report the polygon's area; compute it in an equal-area projection, not in degrees.
What to do next
The CSV is the starting point for analyses outside the app. To estimate recharge, use the app's Recharge Analysis,
which runs the water table fluctuation method on the filled series and has its own CSV download. The Seasonal Model describes how the GWSa_filled and TWSa_filled columns are estimated.