SCALING FIELD SAMPLES OF LAKE WATER QUALITY CHARACTERISTICS TO REGIONAL AND NATIONAL SCALES IN GOOGLE EARTH ENGINE

Abstract:

Inland waters are an active component of the landscape, influencing multiple ecological processes and impacting phenomena at various scales including carbon cycling and regional nutrient cycles. Furthermore, lakes are under increasing human pressure, with climate change alone already impacting both the availability and quality of freshwater. Canada has the most lakes of any nation on Earth, and they occur in complex spatial patterns with connections among them that are not well understood. Despite their importance in Canada, a national-scale accounting of lake water parameters including carbon concentration has, until very recently, been impossible. With the large field effort of the NSERC Lake Pulse Strategic Network, hundreds of lakes have been sampled and analysed as the basis for a nationwide lake assessment. Yet with hundreds of thousands of other lakes to consider, satellite data provides the only viable means for extensive monitoring of these bodies of water. With the recent launches of new satellites and the Google Earth Engine computing platform, prospects for national-scale lake monitoring have greatly improved. Landsat-8 and Sentinel-2 based empirical algorithms are used to predict the concentration of coloured dissolved organic matter (CDOM) and Secchi disk depth, and we will also assess the suitability of using TensorFlow to develop these algorithms. We evaluate the error of the satellite derived water quality parameters resulting from two atmospheric correction methods with in-situ data; the use of traditional surface reflectance products and products derived from the Modified Atmospheric Correction for INland waters (MAIN). It will provide a baseline estimate for carbon content in the lakes of Canada, an important goal that is of high practical importance. Creating this pipeline of analysis and estimation will advance lake science and, as more and more satellites are launched, provide an estimation framework that improves our understanding of Canada’s lakes.