Step-by-Step: Mapping Lake Water Clarity
The following section briefly describes the steps for extracting water clarity information from satellite imagery. More complete descriptions can be found in Chipman et al. (2009) and Olmanson, et al. (2001)
Step 1: Collect ground samples
To extract useful information from satellite images about land features, we generally need to gather some information on the ground, commonly referred to as ground samples, reference data, or field data. These data or samples provide information to verify what the satellite sensors are detecting. They should be collected close in time to when the satellite images are acquired.
Several ways can be used to gather ground samples depending on the water quality aspect being studied. For lake water clarity, we use Secchi depth readings of sample lakes. In Minnesota the Pollution Control Agency and its Citizen Lake Monitoring Program regularly measures Secchi depth on about 900 lakes in Minnesota; collecting ground samples is simply a matter of obtaining the data from the MPCA.
Step 2: Acquire satellite imagery
Currently, several satellites orbit the Earth with sensors capturing images. The satellites and sensors vary in their spatial, spectral, radiometric, and temporal resolutions. Spatial resolution can vary from 0.6 meters (e.g., QuickBird) to 1,000 meters (MODIS). The Landsat Thematic Mapper, the sensor used for most of our lake water clarity work, has medium resolution (30 meters) but covers relatively large areas with a swath width of 115 miles. MODIS, the sensor on the Terra and Aqua satellites, covers a much larger area with 1,400 mile swath width, but at coarse resolution (250-1000 m). Obtaining clear imagery is critical since haze, as well as clouds, can affect the results. If you would like to learn more about the science and technology behind satellite remote sensing, visit the Classroom section.
Step 3: Process satellite imagery
Once the imagery is acquired, analysts go through a series of steps to prepare the imagery for analysis.
1. Following geometric rectification to match other maps, satellite imagery may need to be pre-processed to remove cloud, haze and sun effects.
2. When mapping water features, non-water areas, such as agricultural land, urban land, and forests, are masked out of images.
3. After the imagery has been pre-processed, the relationships between lake clarity and their spectral-radiometric responses (in the simplest sense - colors) are determined by regression modeling for a representative sample of each class. We have found strong relationships between lake water clarity and the responses in the blue and red spectral bands.
4. The regression model then is applied to all lakes in the image, providing a census of lake clarity.
Step 4: Create a map
Once the mathematical relationship between the satellite data and the field data has been developed and applied to all pixels, a map of water clarity can be produced, and the pixels can be classified into discrete classes of clarity level visually displayed such as in the Lake Browser or put into a Geographic Information System (GIS) for further analysis.
Chipman, J.W., L.G.Olmanson and A.A.Gitelson. 2009. Remote Sensing Methods for Lake Management: A guide for resource managers and decision-makers. Developed by the North American Lake Management Society in collaboration with Dartmouth College, University of Minnesota, and University of Nebraska for the United States Environmental Protection Agency. 126 pp.
Olmanson, L.G., Kloiber, S.M., Bauer, M.E., and Brezonik, P.L. 2001. Image processing protocol for regional assessments of lake water quality. Water Resources Center, Public Report Series #14, University of Minnesota, St. Paul, MN, 55108. 13 pp.