Modeling of wheat yield and soil fertility characteristics using learning machines

Dahmardeh, A. R. (2019) Modeling of wheat yield and soil fertility characteristics using learning machines. Masters thesis, university of zabol.

[img]
Preview
Text
Modeling of wheat yield and soil.pdf

Download (297kB) | Preview

Abstract

Wheat is a strategic agricultural crop for each country, so it is important the predicting wheat yield and studying the factors affecting it by using modern modeling methods. The present study was done to compare the ability of predicting wheat yield by regression, artificial neural network and random forest methods at the research field of University of Zabol. For this purpose, the topsoil (0-30 cm) and plant were sampled from 100 plots completely random. Soil particle size distribution, soil reaction (pH), electrical conductivity, adsorbed phosphorus, potassium, organic matter, nitrogen and soil apparent electrical conductivity were measured and grain weight, 1000 grain weight and total yield as yield components were measured. Then, wheat yield was modeled using regression, MLP type of artificial neural network and random forest methods. It is noteworthy that in addition to the soil parameters mentioned above, the remote sensing data were used from Sentinel 2 satellite for estimating environmental variables in random forest method. The results showed that the MLP neural network model with 16 neurons in the hidden layer had the best estimation of wheat yield with R2 and RMSE of 0.76 and 0.46 respectively, versus the regression (R2 = 0.4 and RMSE = 1.62) and random forest methods (R2 = 0.095 and RMSE = 921.48). Also, sensitivity analysis results showed that nitrogen, clay, organic matter and silt were the most important factors affecting wheat yield, respectively, and soil apparent electrical conductivity as a factor measured by soil proximal sensing techniques (EM38-MK2) is the fifth most important factor in predicting wheat yield in the study area.

Item Type: Thesis (Masters)
Uncontrolled Keywords: Proximal soil sensing, Soil fertility, Apparent electrical conductivity, Wheat yield, Sistan
Subjects: S Agriculture > S Agriculture (General)
Depositing User: Mrs najmeh khajeh
Date Deposited: 16 Nov 2019 05:51
Last Modified: 16 Nov 2019 05:51
URI: http://eprints.uoz.ac.ir/id/eprint/2580

Actions (login required)

View Item View Item