Techniques for water storage estimates in Central Minnesota

by John Nieber

How much water is there in Minnesota at any given time? How much of that water is available to human use and use by nature? Information about the amount of water stored in the surface waters, soil, and groundwater systems on earth is valuable to water resource managers. Decisions on water allocations rest on assumptions or assessments of the availability of water in these different storage zones. The forecasting of the near-future states of the hydrologic system is dependent on estimates of water storage amounts. For instance, the occurrence of floods results when the storage capacity is insufficient to store imminent rainfall or snowmelt, and hydrologic droughts occur when the water in storage is insufficient to maintain stream baseflow or lake/wetland water levels. Unfortunately, real-time information about water storage is not readily available, and in the past water managers have had to rely upon estimates being made from water balance calculations. These calculations can be in error due to inaccuracies in quantifying precipitation, evapotranspiration, and streamflow. Approaches for measuring water storage directly and in real time would be very helpful.

In the past few decades satellite platforms have become available for monitoring water storage on the earth. Examples of these include the GRACE satellite, the SMOS/SMAP satellites, and the Landsat satellite. The GRACE satellite measures the gravitational field of the earth and since this is affected by the mass of water stored at any given location it is possible to quantify the changes in total terrestrial water storage (water in lakes, wetlands, soil moisture, groundwater) over time. The SMOS/SMAP satellites measure the amount of water stored in the shallow surface layer of the soil profile, and from this one can estimate the amount of water stored in the soil profile. The Landsat satellite provides images of the earth’s surface and so one can detect and quantify the amount of open water for lakes and large wetlands.

figure 1

Figure 1. The study region. It contains 17 HUC-8 watersheds, all of which are delineated. Each of the HUC-8 watersheds has a streamflow gaging station.

Combining ground-based measurements (monitoring wells, lake levels, soil moisture, meteorological) with the satellite observations offers the opportunity to track the changes in water storage on a spatial and temporal scale relevant to the needs of water resource management.

The project “Techniques for water storage estimates in Central Minnesota” funded by the LCCMR (M.L. 2017, Chp. 96, Sec. 2, Subd. 04h) involved the use of ground-based measurements, satellite measurements, and hydrologic modeling to quantify the water stored within 17 HUC-8 watersheds in Central Minnesota. The project study area stretched from the Twin Cities to Moorhead (Figure 1). The study region contained 151 monitoring wells, 816 surveyed lakes, and over 17 streamflow monitoring sites. The bathymetric data for the surveyed lakes was used to develop a regression equation to calculate the volume of water in other lakes of the region. Water levels measured in lakes and water-table wells, along with groundwater recharge estimates from a statewide groundwater recharge model were used to map annual changes in the water table depth.  The water table position was then used with geologic information to calculate groundwater storage.  The groundwater water storage estimate is for the Quaternary aquifer, and does not include groundwater stored in deeper bedrock aquifers.  The streamflow measurements were applied in hydrologic modeling and quantification of groundwater discharge to streams. These data were applied to derive a reference water storage estimate for lake waters, soil moisture, and Quaternary aquifer within the study region. Satellite data, meteorological data, and streamflow data for the period 2002 to 2015 were then used in water balance calculations to quantify the variation in total terrestrial water storage in the study region.

figure 2

Figure 2. Summary of water storage estimates for different storage components in the Central Minnesota region. The estimate for the total volume is 1,569 km3.


The estimates for the reference water storage in the study region is summarized in Figure 2 which shows the breakdown of the water stored in lakes, wetlands, soil moisture and Quaternary aquifer groundwater. As might be expected, the vast majority of the water stored is in the groundwater aquifer. While Minnesota is known as the land of 10,000 lakes, within the study region the lake number is about 40,054 lakes, which includes water bodies greater than 1 acre in surface area.

To illustrate the information acquired from satellites we look at the seasonal change in total terrestrial water storage in the study area based on the GRACE satellite for the period 2002 to 2015. This change in water storage is illustrated in Figure 3. The mean seasonal monthly change in water storage is expressed in units of mm of water. Also plotted is the result of a simple hydrologic model which used precipitation (available from the State Climatology Office), evapotranspiration available from NOAA, and streamflow available from MnDNR/USGS. The change in water storage shows decreases in storage during the growing season, due to evapotranspiration, and increases in the non-growing season. The comparison between the modelled water storage changes and the satellite yields and R of 0.71 and a Nash-Suttciffe Efficiency equal to 0.18. Reasons for the differences between the modelled result and the satellite data are still under investigation.

figure 3

Figure 3. Seasonal average change in water storage (mm/month) from the GRACE satellite and from a water balance model. P is precipitation, ET is evapotranspiration, and Q is stream discharge.


Although the LCCMR project has been completed, we are continuing to work with the data and information acquired during the project. One current application is to test the Reager Flood Potential Index for flood forecasting in the study region. The RFPI is based is based on quantifying the water storage deficit in the landscape, and the original development was applied using GRACE data on 10,000 km2 tiles. Our effort is the development of downscaled water storage data derived using GRACE satellite data, ground-based data, and the HSPF model, and testing the RFPI based on the downscale storage information.