| Abstract: | Lake studies play an important role in water management, ecology and other environmental issues. Monitoring lake stages is typically the first step on the lake studies. However, for the Prairie Pothole Region (PPR) of North America with millions of small lakes and potholes, onsite measurements for lake stages are almost impossible with the conventional gage stations due to the limited accessibility to lakes and the requirement of economical and human resources. To overcome this limitation, we employed the remote‐sensing approach to extract lake stages in PPR for bulk processing. To estimate lake stages, several image‐processing techniques were used with the Shuttle Radar Topography Mission (SRTM) data in January, 2000, a one‐time snapshot useful in historic lake level reconstruction. In this research, image processing techniques, for example, averaging, Fast Fourier Transform (FFT), Lee‐Sigma filtering were applied and masking with Canny Edge Detector (CED) and semi‐automated contouring were developed for lake stage estimation. In terms of simplicity, averaging might be useful if 1·5 m of the local bias for the study area in stage estimates. However, the FFT and Lee‐Sigma methods were slightly better than averaging. Masking with CED and semi‐automated contouring provided outstanding accuracy in the estimation. The masking and contouring methods provided ± 0·5 m of stage differences for relatively large lakes greater than 10 km2 when compared with stage reading in spite of their calculation complexities and was shown as the best approaches among implemented methods in this study. Copyright © 2010 John Wiley & Sons, Ltd. |
