Regional-scale algorithm to estimate the particulate organic carbon (POC) in inland waters using Landsat-5/TM images

Abstract

A regional-scale algorithm was developed in order to test if the Landsat-5/TM can be used to estimate the particulate organic carbon (POC) in oligotrophic-to-mesotrophic inland water. To develop the POC algorithm two fieldworks were conducted, the first in May and the second in September 2009. The algorithm was calibrated using the dataset from September and validated using the dataset from May. The results showed that the best calibration was obtained using a polynomial fitting function (R2 of 0.80, p < 0.0001). This model was validated with a normalized root mean square error (NRMSE) of 6.21%. The algorithm was then applied in two Landsat-5/TM images from April and July 2009. The spatial distribution of POC obtained from the satellite images reveals a strong dependence of POC concentrations with the weather conditions. These results allowed us to conclude that there is a great potential to study the temporal dynamics of POC in inland waters using Landsat-5/TM images.