| Abstrakt: |
The launch of the Landsat 8 in February 2013 extended the life of the Landsat program to over 40 years, increasing the value of using Landsat to monitor long-term changes in the water quality of small lakes and reservoirs, particularly in poorly monitored freshwater systems. Landsat-based water quality hindcasting often incorporate several Landsat sensors in an effort to increase the temporal range of observations; yet the transferability of water quality algorithms across sensors remains poorly examined. In this study, several empirical algorithms were developed to quantify chlorophyll-a, total suspended matter (TSM), and Secchi disk depth (SDD) from surface reflectance measured by Landsat 7 ETM+ and Landsat 8 OLI sensors. Sensor-specific multiple linear regression models were developed by correlating in situ water quality measurements collected from a semi-arid eutrophic reservoir with band ratios from Landsat ETM+ and OLI sensors, along with ancillary data (water temperature and seasonality) representing ecological patterns in algae growth. Overall, ETM+-based models outperformed (adjusted R2 chlorophyll-a = 0.70, TSM = 0.81, SDD = 0.81) their OLI counterparts (adjusted R2 chlorophyll-a = 0.50, TSM = 0.58, SDD = 0.63). Inter-sensor differences were most apparent for algorithms utilizing the Blue spectral band. The inclusion of water temperature and seasonality improved the power of TSM and SDD models. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink