Remote Sensing of Global Lake Gross Primary Production

Lakes contribute to local and regional climate conditions, cycle nutrients, and are indicators of climate change due to their sensitivity to disturbances in their air and watersheds. Spaceborne remote sensing (RS) techniques have promise for studying lake dynamics by allowing for consistent spatial and temporal observations and estimates of lake functions without in situ measurements. Recent advances in modeling lake metabolism use high frequency sensor data, but there are few existing algorithms that relate RS products to in-lake estimates of metabolic rates. I use satellite surface temperature observations from MODIS product MYD11A2 and published in-lake gross primary productivity (GPP) estimates for ten globally distributed lakes, with areas greater than 1 km2 , varying trophic states and surrounding land cover to produce a univariate quadratic equation model. Statistical analyses reveal a significant positive relationship (p<.00001) between MODIS temperature data and in-lake GPP for the global model. I performed preliminary validation on the global model using a lake reserved from the data set (Lake Acton) resulting in a strong correlation (R2=0.76) between MODIS-derived GPP and previously modeled values. Lake-specific algorithms such as those for Rotorua (NZ) and Kentucky (USA) had stronger relationships than the global model derived from all ten lakes, pointing to the influence of regional biological and physical characteristics of the lakes and their watersheds. Analyses of land cover type within lake watersheds and in-lake GPP revealed a positive correlation with forested land cover and GPP (R=0.67, p=0.03).Land cover type was incorporated into a separate model that was not statistically significant. These data suggest that it may be possible to predict GPP for lakes across a wide geographic region.