ANALYSIS AND PREDICTION OF GREENHOUSE GAS EMISSION USING FEEDFORWARD NEURAL NETWORK

Main Article Content

Pranali K. Kosamkar , Dr. Vrushali Y. Kulkarni

Abstract

Greenhouse Gas (GHG) emission is caused by decomposition of biomass and dead plant residues, livestock enteric fermentation in ruminants, and burning of crop residues. As the concentration of GHG rises it raises the temperature on the globe causing the Global Warming.  Alterations in agriculture management practices may reduce the GHG emission.  Therefore, there is a need to analyze and forecast the GHG emission from Agriculture. We have built the Feedforward Neural Network using sequential neural network in keras for predicting CO2 and CH4 emission for Onion crop from open farm and poly house. We selected Onion for our study because Onion is one of the second most important commercial crops of the India. The GHG emission may vary in open farm and poly house for onion crop because the environment is controlled in poly house as compared to open farm. For this study we collected the field data of soil attributes, climatic attributes and CO2, CH4 greenhouse gases from the experiment field. We hyper tune the model with 3, 4 and 5 layers with different epoch. We have used Root mean squared error (RMSE), Mean squared error (MSE) and R-square as a coefficient of correlation for model prediction accuracy. Model predicted that Nitrogen, Moisture, Pressure, Humidity and Temperature are major affecting factors for emission of GHG for onion crop from open farm and poly house. The model indicates good prediction response for GHG emission with major influencing attribute for onion crop.

Article Details

Section
Articles