摘要:Lake water clarity as measured by Secchi disk transparency (SDT) is a cost-effective measure of water quality. However, in regions where there are thousands of lakes, sampling even a small proportion of those lakes for SDT year after year is cost prohibitive. Remote sensing has the potential to be a powerful tool for assessing lake clarity over large spatial scales. The overall objective of our study was to examine whether Landsat-7 ETM could be used to measure water clarity across a large range of lakes. Our specific objectives were to: 1) develop a regression model to estimate SDT from Landsat data calibrated using 93 lakes in Michigan, U.S.A., and to 2) examine how the distribution of SDT across the 93 calibration lakes influenced the model. Our calibration dataset included a large number of lakes with a wide range of SDT values that captured the summer statewide distribution of SDT values in Michigan. Our regression model had a much lower r2 value than previously published studies conducted on smaller datasets. To examine the importance of the distribution of calibration data, we simulated a calibration dataset with a different SDT distribution by sub-sampling the original dataset to match the distribution of previous studies. The sub-sampled dataset had a much higher percentage of lakes with shallow water clarity, and the resulting regression model had a much higher r2 value than our original model. Our study shows that the use of Landsat to measure water clarity is sensitive to the distribution of water clarity used in the calibration set.