1College of Agriculture, Jawaharlal Nehru Krishi Vishwa Vidyalaya, Ganj Basoda, Vidisha-464221, Madhya Pradesh
2College of Agriculture Engineering, Jawaharlal Nehru Krishi Vishwa Vidyalaya, Jabalpur-482004, Madhya Pradesh
3College of Agriculture, Rajmata Vijayaraje Scindia Krishi Vishwa Vidyalaya, Indore -452001, Madhya Pradesh
Online published on 21 August, 2015.
Artificial neural network is a key tool for soil loss estimation and it is required for proper watershed management and other developmental work in situ. The study area, Kalidevi watershed where this study was carried out, comes under the Bagh block of the Kukshi tehsil of district Dhar of Madhya Pradesh and has been chosen for developing artificial neural network model for estimation of soil loss. Data on daily rainfall, API and days since 1st June and runoff were taken as input variables of period 2003–2005 during rainy seasons have been used for the analysis and development of ANN model for soil loss estimation. It is found that for soil loss modeling, an ANN model with four input variables with 60 neurons single hidden layer and one output and learned with Levenverg-Marquardt (LM) algorithm performed better for the estimation of soil loss.
Antecedent precipitation index (API), Artificial neural network (ANN), Levenverg-Marquardt algorithm (LM), Random access memory (RAM), Soil loss modeling
