Satellite remote sensing and geographical information systems: Tools for regional surface water assessment

A procedure was developed for regional assessment of Secchi disk transparency (SDT) of lakes using Landsat imagery. Analysis of a large monitoring database showed strong seasonality in SDT and that a sine-function model could adjust for seasonal effects in satellite SDT estimates, making them more comparable from year-to-year. Results indicate that a single late-summer SDT measurement yields reasonable estimates of growing-season means. Inspection of the Landsat archive showed that 75% of the years have at least one cloud-free image during this period, thus supporting an annual assessment frequency.; The procedure was applied to approximately 500 lakes in the metropolitan area of Minneapolis and St. Paul, MN, using 13 Landsat images over the period 1973–1998. Satellite brightness values from lakes were calibrated against historical SDT measured within ±1 day of image acquisition. Regression equations for late summer images had a range of r2 of 0.72–0.93 for ten TM images and 0.60–0.79 for three MSS images. A subset of this data was analyzed along with lake and landscape variables to identify important relationships. Of the 44 variables examined, 16 had significant correlations with water clarity. Nine land cover variables (mostly related to agricultural activity) were significantly correlated, but these correlations were weak (r2 < 0.05). The strongest correlations with water clarity were with depth, watershed to lake area ratio, and water residence time. Regional analysis of long-term SDT trends shows that in spite of large land-use changes within the region, only 49 (∼10%) of assessed lakes showed significant temporal trends; more lakes had increasing SDT (34) than decreasing SDT (15).; A watershed model was developed incorporating simple algorithms for nonpoint source pollutant loading and in-lake nutrient dynamics. The watershed-lake network model was implemented in a GIS framework and was applied to three watersheds having a range of land uses and containing 33 lakes. Output was calibrated to a trophic state index based on satellite-estimated water clarity. The model explains 50% of the regional variability in trophic state. Using only basic lake and watershed characteristics, this model can provide estimates of trophic state that may be suitable for regional analyses.