Agriculture is one of the economic sectors that has both a direct and indirect effect on climate change, contributing to greenhouse gas (GG) emissions. This emission is created by the putrification of biomass and dry plants deposits and burning of crop leftovers. The cropped soils generate more than half of the earth’s anthropogenetic N_2 O fluxes, stock and discharge the organic carbon as 〖CO〗_2, and can produce as well as consume〖CH〗_4. The agricultural soils can be both emitters and sinks of GG. The sources and sinks of each of these GG varies, and each is affected differently by agronomic management. The Kyoto Protocol's architecture accounts for the interplay that occurs between the stratospheric ozone layer, solar ultraviolet (UV) rays, and climate change. In this research, the high temporal and spatial variability of farming emit the GG from the soil that need to analyse and forecast the climatic changes by using deep convolutional neural network (CNN) with Random Forest Classifier (RFC) named as CNN-RFC. The prediction is performed using humidity of air, soil texture and temperature of the atmosphere. The performance metrics is obtained from the values of root-mean-squared-error (RMSE):2.104, 13.231 and 4.128, mean-absolute-error (MAE):1.053, 10.032 and 3.653, Pearson: 0.325, 0.138 and 0.357,〖 R〗^2 coefficient:0.432, 0.473 and 0.432 and standard deviation (SD):5.463, 10.587 and 3.463 for 〖CO〗_2,〖CH〗_4 and N_2 O respectively.