2022 ESA Annual Meeting (August 14 - 19)

PS 37-196 Satellite remote sensing based detection of coldwater stream habitats over larger geographical extents in the upper Midwest USA: Potential, challenges and the way forward

5:00 PM-6:30 PM
ESA Exhibit Hall
Niti Mishra, University of Wisconsin-La Crosse;Michael Siepker,Iowa Department of Natural Resources;
Background/Question/Methods

Coldwater streams are crucial habitats for many types of biota including Salmonidae and Cottidae species that are unable to tolerate warmer temperature. Climate change is projected to alter the prevalent thermal characteristics of coldwater streams. To manage coldwater streams, we must first know where they exist on the landscape. During winter, coldwater streams, with cooler groundwater input, remain free from ice cover whereas warm water streams remain ice covered. We analyzed winter imagery for two selected watersheds in the driftless region in Iowa (where in-situ data was available) to examine (i) the potential of high-resolution winter imagery for detecting coldwater streams and (ii) accessed the accuracy of existing maps of coldwater reaches and understand the physical factors that contribute to inaccuracies in detecting coldwater habitat? Selected multi-temporal winter imagery were visually interpreted over the entire watershed to assign coldwater designation at two confidence levels. At level 1, streams represented clear, open water with little to no obstacles impeding the view of the stream and at level 2 streams were semi-clear with obstacles impeding the stream path (image angle, clouds, tree cover). Detected stream reaches were compared against the existing coldwater stream reaches and spatial agreement and disagreements were examined.

Results/Conclusions

Results indicate that existing Iowa-DNR maps missed significant stretches of coldwater streams. The mismatch ranged from areas that were clearly visible as open water (without canopy cover and significant stream width) to areas where the stream width was much narrower requiring closer interpretation. Comparison also showed areas of disagreement where the DNR’s interpretation of coldwater stream was not found in our interpretation. This was attributed to shadows from stream banks or other extreme topography hindering our interpretation due to low image acquisition angle. Recent feedback from coldwater stream managers in the DNR revealed certain incorrect assignments in our level 2 interpretation. Ongoing and future research is focusing on removing these errors by agreement of multiple interpreters and cross-validating results against temperature logger data. These results and gained understanding provide the basis for developing a semi-automated methodology for stream classification over a larger geographical extent.