The potential of Earth Observation in modelling nutrient loading and water quality in lakes of southern Québec, Canada

Abstract

Phosphorus and nitrogen are key nutrients that affect abundance and growth of aquatic primary producers but cannot be directly remotely sensed as their dissolved or organic forms do not interact with the remote sensing signal. In addition, other lake water quality variables such as chlorophyll a and Secchi disk depth, have been previously successfully estimated with remote sensing, but the retrieval algorithms are site-, season-, and/or scene-specific. Such algorithms do not take into account lake typological features, which can affect the sensitivity of lake to change, or catchment characteristics, for example, land cover that is a major driver of lake water quality change. Here we propose a novel approach that utilises remotely sensed land cover information in the catchment to estimate phosphorus, nitrogen and chlorophyll a concentrations in lake waters. We use land cover type-specific nutrient export coefficients and the NASA MODIS (Moderate Resolution Imaging Spectroradiometer) Land Cover Type product showing that nutrient loading based on remote sensing can explain up to 75% of variability in lake nutrient concentrations and 58% of variability in lake chlorophyll a concentrations. In addition, we show that land cover information, supplemented by satellite measurements and lake morphometry data are good predictors of chlorophyll a (R2 = 0.77) and Secchi disk depth (R2 = 0.87) in lakes with different trophic statuses and in different months and years.