Estimation and analysis of plant grains based on satellite data and artificial neural network

Over the last few years, green areas are decreasing rapidly due to urbanization; especially in the western regions of Turkey. NDVI (Normalized Difference Vegetation Index) remote sensing method has been used to obtain the vegetation index for the study area. MODIS satellites were used to estimate the vegetation coverage for the Kandilli region and the neighboring villages based on the daily NDVI values observed for ten years. ANFIS (Adaptive Neuro-Fuzzy Inference System) model, which is one of the ANN (Artificial Neural Network) varieties, was used for vegetation prediction. In this study, the NDVI values between September 2005 and October 2016 were used for the Kandilli region located at a latitude value of 41 ° 4 ' and a longitude value of 29 ° 3'. The TERRA satellite which is a MODIS (Moderate Resolution Imaging Spectroradiometer) variety was used to measure the values. Measurements were made on an area of approximately 6.25 square kilometers. Estimation of vegetation was done using ANN (Artificial Neural Network) with the NDVI values obtained from the TERRA satellite. ANFIS, which is defined as a Neuro-Fuzzy Designer in MATLAB environment, is considered as a combination of classical set theory and fuzzy logic theory. In this method, the main goal is to determine whether a value belongs to that cluster or not, using membership functions. The error tolerance values of the membership functions in ANFIS are compared with each other. Estimation was performed by using the most successful membership function.