Estimation AOD550 with MODIS AODs Using an Artificial Neural Network Limited Meteorological Data in Iraq/Al Ramadi C

Abstract

Aerosols affect the amount of sunlight that reaches the Earth causing many damages, so it can be measured by the optical depth of the aerosols. We developed an artificial neural network and trained it on the daily climatic data of Ramadi city (temperature, solar radiation and atmospheric pressure) for four consecutive years. climatic used. Activation functions (Gaussian, sigmoid, Hyperbolic Tangent and Hyperbolic Secant), number of hidden layers used (1,2, 3 and 4), adjustments ranging from 10,000 to 50,000 on a scale of 10,000 were used each time for both the output and hidden layers. In order to obtain the best results for the developed ANN models, the statistical criteria were determined based on the correlation coefficient (R), root mean square error (RMSE). According to the statistical criteria it was calculated to evaluate the results of the developed ANN models. It was found that the best ANN models among all the ANN models that were trained and tested was the ANN model (60) where (R = 0.9169) and (RMSE = 0.0866). After obtaining these results, the obtained ANN model can be generalized.