U of M wordmark
website banner
A collaboration of the University of Minnesota's Remote Sensing and Geospatial Analysis Laboratory and Water Resources Center

navigation edge

 
website logo
 

 

 
   

Mapping Stream Quality

Use of Airborne Remote Sensing Imagery for Water Quality Assessment of Minnesota's Rivers

Minnesota has around 92,000 miles of rivers and streams. It has been estimated that 40 percent may be impaired. To date less than 10 percent of Minnesota river and stream miles have been assessed. We are exploring the use of airborne remote sensing as a cost-effective way to gather the information needed for river assessments. We previously have had great success assessing lake water clarity using reflectance information from Landsat imagery and have found similar relationships for large rivers. However, compared with lakes, rivers and streams pose a challenging set of problems for application of remote sensing techniques to water quality assessment because:

  1. They are temporally more dynamic.
  2. The resolution of Landsat (30 m) is too coarse for small rivers and streams.
  3. And, if we want more than clarity, we really need a better set of spectral bands than the Landsat bands.

Our solution has been to use airborne high-resolution hyperspectral imagery obtained from a small aircraft flying over stretches of rivers. For calibration purposes, water samples were collected concurrently with the fly-overs, and to provide a range of conditions for calibrations, we focused our initial measurements around the confluences of river systems in Minnesota that have different water quality characteristics.

Methods

On August 19, 2004 an aircraft fitted with the AISA-Classic sensor collected high resolution 1-3 m hyperspectral (35 well selected bands) imagery over six river segments (identified by red boxes on the map). In a collaborative effort sampling crews from the Minnesota Pollution Control Agency, Metropolitan Council, Minnesota Department of Natural Resources, Minnesota Department of Agriculture and the University of Minnesota were dispatched to collect 39 water samples. These samples were analyzed by the Metropolitan Council and Minnesota Department of Health laboratories for a number of water quality characterization variables.

On August 15, 2005 an aircraft fitted with the AISA-Eagle (AE) Hyperspectral Imager (VNIR) collected high resolution 2 m hyperspectral (97 contiguous bands ~2.5 nm from 435-724 and ~10 nm from 724-950 nm) imagery over a fairly large area along (36 mile stretch) the Mississippi River from Spring Lake to Lake Pepin (identified by purple boxes on the map). At the same time sampling crews from the Minnesota Pollution Control Agency and the Metropolitan Council collected 22 water samples. The in-situ water quality data and remotely sensed data are currently being analyzed to determine the best model for each variable. Preliminary single band, band ratio and multiple band regression analysis models were used to create the maps of each water quality variable for each river segment.

Results

Preliminary results are promising with strong relationships for a number of important water quality variables (see water quality table). With additional statistical analysis we anticipate developing improved models. Using the best fit models from our preliminary assessment, we were able to map important water quality variables for river segments throughout each image. The maps show the complex interactions of sediment and different types of algae in these important river segments.

In the future we anticipate that remote sensing will be an important tool in assessing water and land resources including thousands of miles of rivers. This should enable us to see more detailed water quality patterns than we could ever sample with volunteers or more advanced field diagnostic methods. Remote sensing allows us to see the big picture of land and water resources as well as being able to zoom in and get a more detailed view. This “complete view” can be used to detect problem areas and help allocate limited field monitoring resources to areas that need additional attention.

Acknowledgements

This research has been conducted by the faculty and staff of the University of Minnesota, Department of Civil Engineering and College of Natural Resources -- Remote Sensing and Geospatial Analysis Laboratory and Water Resources Center, with support from the Legislative Commission on Minnesota Resources, Minnesota Pollution Control Agency and Metropolitan Council.

Back to the top
Back to the top