UPM Institutional Repository

Reference evapotranspiration estimation using adaptive neuro-fuzzy inference system


Citation

Karimaldini, Fatemeh (2011) Reference evapotranspiration estimation using adaptive neuro-fuzzy inference system. Masters thesis, Universiti Putra Malaysia.

Abstract

Evapotranspiration is an integral part of the hydrologic cycle and an important component in water resource development and management, especially in the arid and semi-arid conditions such as those found in Iran, where water resources are limited. The standard FAO-56 Penman Monteith (PM) equation requires several meteorological parameters for estimating reference evapotranspiration (ETO) that are not usually available in most stations. In addition, traditional methods that require limited climatic parameters for ETO estimation are not applicable to all climatic conditions. As an alternative to traditional techniques, soft computing techniques can be used for modelling this nonlinear phenomenon. This research investigates the potential of the Adaptive Neuro-Fuzzy Inference System (ANFIS) technique for daily reference evapotranspiration modeling under arid, semi arid and humid conditions of Iran. The Gamma Test (GT) technique is employed to find the best input combination and number of sufficient data points for the model calibration. The training and testing data sets are chosen based on the K-fold method of cross validation. The estimates of ANFIS models are compared with FAO-56 reduced-set PM ETO approaches and conventional empirical ETO equations (Hargreaves, priestly Tailor, Makkink, Blaney-Criddle and Turc) that are calibrated with the Genetic Algorithm technique. The FAO-56 full-set PM method is adopted as the reference ETO equation, and it is applied to calibrate other ETO equations and ANFIS models. The accuracy of all models are measured based on MAE, MSE and STD values. The comparison results indicate that when similar meteorological inputs are considered, the ANFIS models performed better than all the methods pursued under arid, semi arid and humid conditions of Iran. In addition, the general ANFIS model (whose inputs are temperature, solar radiation, relative humidity, wind speed and aridity coefficient) performed well under arid, semi arid and humid conditions of Iran. It is concluded that after temperature, the wind speed data is the most important parameter that should be considered in the combination of inputs for ETO estimation under arid and semi-arid conditions, however the effect of wind speed in the combination of inputs for ETO estimation under humid conditions is found to be not significant. It has been found that the minimum required meteorological parameters that gives the low error rate under arid and semi-arid conditions are temperature and wind speed data, and under humid conditions are temperature data and estimated relative humidity data from temperature data (by FAO-56 reduced set approach). Therefore, using the ANFIS technique is strongly suggested as an alternative to the traditional methods that suffer from several stringent meteorological requirements or invalidity under various climatic conditions.


Download File

[img]
Preview
PDF
FK 2011 83R.pdf

Download (962kB) | Preview

Additional Metadata

Item Type: Thesis (Masters)
Subject: Evapotranspiration
Subject: Fuzzy systems
Subject: Neural networks (Computer science)
Call Number: FK 2011 83
Chairman Supervisor: Professor Lee Teang Shui, PhD
Divisions: Faculty of Engineering
Depositing User: Haridan Mohd Jais
Date Deposited: 14 Mar 2016 07:07
Last Modified: 14 Mar 2016 07:07
URI: http://psasir.upm.edu.my/id/eprint/42280
Statistic Details: View Download Statistic

Actions (login required)

View Item View Item