Comparison of Spatial distribution of tomato fruit borer, Helicoverpa armigera using geostatistics and fuzzy-neural network methods

Despite the use of robust statistical methods and fuzzy-neural networks, models that predict the distribution of organisms have seen rapid development in the field of ecology. However, due to the challenges associated with sampling, these studies often lack sufficient samples. In this research, we compared geostatistics and fuzzy-neural networks to estimate the distribution of the tomato fruit worm in a tomato farm in Kermanshah city. For this purpose, the length and width coordinates of the sampling points at the field level were identified and used as inputs for both methods. The output of each method was the count of this pest at those locations. In the geostatistics approach, we employed the normal Kriging method, while in the fuzzy-artificial neural network approach, we used the sigmoid activation function. A comparison of the results from geostatistics and the fuzzy-neural network demonstrated the superior performance of the fuzzy-neural network. The coefficient of determination for the fuzzy-neural network and geostatistics was 0.9 and 0.6, respectively. In conclusion, the fuzzy-neural network method, by integrating latitude and longitude factors, was able to predict the density of the tomato fruit worm with high accuracy.