The possibility of using the Landsat image archive for monitoring long time trends in coloured dissolved organic matter concentration in lake waters

Recent studies indicate that lakes are regulators of carbon cycling and climate. Therefore, it is important to know how the lake carbon content has changed over the last decades. In situ long time data series about the amount of dissolved organic carbon (DOC) in lakes are rare. The only potential way to study retrospectively the changes in lake carbon over the last decades is by means of remote sensing data provided there are sensors that can provide data about coloured dissolved organic matter (CDOM) in lakes over long periods. Landsat data archive contains images from 1984 to nowadays and covers the whole Earth. Although the sensors were not designed for remote sensing of aquatic environments it is still tempting to utilise the long data series. Landsat 4, Landsat 5 and Landsat 7 imagery available in free Landsat image archive was compared with time series of CDOM in situ data from 19 sampling stations available in the Swedish University of Natural Sciences lake monitoring database. There was no correlation between the image and in situ data when all the above mentioned data were used. Low radiometric resolution of the sensor, small size of many lakes (= large adjacency effects) and high concentration of CDOM (negligible water leaving radiation) could be the reasons. The results were more promising (R2 = 0.62) when Lake Mنlaren stations were analysed separately. The lake is sufficiently large and with variable, but not extremely high, CDOM content. The Lake Mنlaren in situ data showed very different trends in CDOM concentrations in different basins of the lake over the last 45 years. Although the correlation between the image and in situ data was a bit low for accurate daily estimation of CDOM concentrations from Landsat data it could allow detecting general trends in lake CDOM content. Unfortunately, there is currently a gap in Landsat archive (for our study sites) between 1988 and 1998 which makes calculations of long time trends unreasonable for the time being.