Application of dynamic programming and Markov Chain in soil fertility management

Date

4-1993

Degree

Bachelor of Science in Agriculture

College

College of Arts and Sciences (CAS)

Adviser/Committee Chair

Myrna C. Belarmino

Abstract

Policies on fertilizer application during dry and wet season were evaluated using rainfall as a decision variable.

A first-order Markov chain was fitted to seasonal rainfall distribution in Maligaya Rice Research Experimental Station in Nueva Ecija. Two transition probability matrices were produced using Pegram model from the fifteen-year historical rainfall data. The model is on the form P = pI+(1-p)p1. Each transition matrix for dry and wet season described three states namely: very dry, dry and wet.

Return and reward matrices were also produced using the average yield response of IR 6 to manage fertilizer inputs. Each reward structure corresponded to each classified state.

Results using the policy iteration method for finding the optimal policy showed that the rate 140-60-0 and 70-60-0 for dry season and wet season were optimal, respectively.

The method used should be applied to other data to fully illustrate its power

Language

English

Location

UPLB Main Library Special Collections Section (USCS)

Call Number

Thesis

Document Type

Thesis

This document is currently not available here.

Share

COinS