A MODIS-based estimation of chlorophyll a concentration using ANN model and in-situ measurements in the southern Caspian Sea

Abstract

Chlorophyll-a data of the MODIS sensor with in-situ chlorophyll measurements from the southern Caspian Sea (SCS) is compared in the present study. Analysis showed an overestimation of chlorophyll-a concentration by MODIS in the area. Results also indicated a root mean square (RMS) log error of 39.4%, for 53 coincident data points. An artificial neural network (ANN) with radial basis function was applied to the in-situ measurements and satellite imagery. It included physical-chemical properties of water as ancillary independent variables in the ANN procedure that enhanced the predictive capability of the model. Evaluation of the predictive capability of ANN approach was satisfying (RMS log error 18.9%). Results showed retrieving chlorophyll-a concentration in the SCS from satellite is possible and will be improved through application of ANN and explanatory environmental parameters.