Landsat and TRMM imagery potentiality for assessing water superficial variation and bottom depth of shallow lake: case study of Lake Poopó.

Abstract:

In flat regions, lake extent fluctuations may be used as an indicator of climate variability and/or pressures on water resources at the basin scale. In this context, an accurate delineation through remote sensing is a valuable tool. Here we focused on Lake Poopó located in the Bolivian Andean plateau at an elevation of about 3686 m. With an extent from 500 to 3000 km² and a mean depth ranging from few cm to 2 m between dry and wet season, respectively, this lake is very sensitive to water use and climatic change.

Field spectral-radiometric measurements were collected on both Lake shore and shallow regions in 2014. A total of 84 measurements matching with a Landsat-OLI 8 overpass were made available for our study. The database was used to test FLASSH atmospheric correction module of ENVI software under different parametrization. It shows the usefulness and necessity of such correction before using Landsat imagery. Then, five commonly used indexes for separating inland water (NDWI, MNDWI, WRI, NDVI and AWEI) were computed from Atmospheric corrected Landsat image and compared with field spectral-radiometric measurement. WRI was found the most suitable indexes to delineate Lake Poopó extent according to spectral-radiometric measurements. Using FLAASH atmospheric correction and WRI index, Landsat imagery was used to estimate Lake Poopó extent for a 17 years period from 1998 to 2015. Fluctuations are compared with rainfall measurement from TRMM TMPA-3B42 v7 and In Situ evaporation to highlights climatic or water use changes during this period.

Finally, 130 water depth measurements collected in 2005 were used to establish a logarithmic correlation between WRI and water depth. The relation was applied for 2014 and computed water depths are in agreement with In Situ measurement with an overall RMSE value of 5 cm.