Methods and tools for assessing impacts of saharan dust deposition in Lake Garda from remote sensing data.

Mineral dust aerosol is generated by Aeolian erosion of soil in desert areas. Due to its origin, it is rich in nutrients and trace metals (e.g. Fe, Si, Al, Mg) useful for phytoplankton growth. Several studies show dust deposition in oligotrophic stratified waters, could induce algal blooms. They have been conducted on oceans and Mediterranean Sea waters, and remote lakes, exploiting Remote Sensing data and techniques. These techniques allow in fact a frequent collection of data with a synoptic point of view on the study area, at relative low costs. In the activities of LTER CNR Sirmione station and SINOPIAE project framework, the effect of dust deposition was inspected on deep clear meso-oligotrophic Lake Garda. For this analysis, different optical satellites images (e.g. MERIS, MODIS) were exploited to retrieve chlorophyll-a concentration (chl-a), proxy of phytoplankton abundance: a processing chain for the retrieval of chl-a values was calibrated and validated. Chl-a is in fact retrieved through empirical or bio-optical models exploiting water reflectance. This physical quantity is not directly measured by sensors, but it can be retrieved from their products through the removal of the atmospheric contribution to the signal and other disturbing factors (e.g. specular reflection and adjacency effect). For MERIS imagery, this operation has been conducted through BEAM-C2R Neural Network, which allows the retrieval of both reflectance and chl-a values, while for MODIS and LANDSAT imagery, 6SV and bio-optical model BOMBER were used. Aerosol parameters retrieved by AERONET sunphotometers in Northern Italy (the last one installed in 2014 in Sirmione) were exploited for both image atmospheric correction and dust deposition events individuation. Aerosol can in fact be characterized by some optical and physical properties (e.g. AOD, refractive index, particle size distribution) which allow to infer their origin. Once the events had been individuated, chl-a concentration in the time-window comprising each deposition event, was analyzed. To assess the actual contribution of deposition to possible chl-a increase, other influencing factors were analyzed (i.e. temperature, to which phytoplankton growth is positive correlated, precipitation, which could induce both wet deposition and the run-off of nutrients previously deposited along the coast). For this purpose an user friendly interface was realized to allow easy identifying anomalies in chl-a from time series of chl-a maps. In particular, the tool allows classifying waters on the basis of some chl-a threshold based on WFD criteria, identifying anomalies in chl-a long time series and evaluating the area where dust deposition could have mainly affected chl-a anomalies, considering both temperature and precipitation. Further progressions of this study include the validation of a new processing chain for the new-generation sensors, Sentinel-2 and Sentinel-3, from which some preliminary results are thus presented