Compatibility of non‐synchronous in‐situ water quality data and remotely‐sensed spectral information for assessing lake turbidity levels in complex and inaccessible terrain

Abstract:

Multiple regression modeling was employed to estimate lake turbidity levels in Glacier National Park, Montana, USA using Landsat Thematic Mapper digital data and non‐synchronous in‐situ turbidity data. Variations in model form and components were investigated to exploit spectral/biophysical relationships described in the literature and to accommodate local site and situation conditions. Limitations inherent in using non‐synchronous in‐situ and remote sensing data resulted in the derived models explaining lower levels of turbidity variance as compared to models developed for other studies in other environments. Unique local conditions contributed to the model performance. Non‐parametric tests of correlation and significance were used because of these inherent limitations. Lake turbidity rankings secured through various models were relatively insensitive to changes in model form. The spectral sampling of lakes for comparison to the in‐situ turbidity measures were best achieved by positioning a 3 by 3 pixel window in the deep central portion of the lake. This was necessitated due to the imprecise description of the location of in‐situ data collected for the study lakes. © 1991 Taylor and Francis Group, Ltd.

Año de publicación:

1991

Keywords:

Fuente:

scopus