Estimation of potential evapotranspiration using multiple linear regression and particle swarm optimization
Evapotranspiration (ET) is an essential element in the hydrological cycle which at the same time plays an important role in ensuring the effectiveness of the water balance system. Generally, there are two methods in obtaining ET; direct measurement and empirical model. The main objective of this stu...
Saved in:
Main Author: | |
---|---|
Format: | Thesis |
Language: | English |
Published: |
2022
|
Subjects: | |
Online Access: | http://eprints.utm.my/id/eprint/101535/1/NorFarahAtiqahAhmadPSKA2022.pdf |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | Evapotranspiration (ET) is an essential element in the hydrological cycle which at the same time plays an important role in ensuring the effectiveness of the water balance system. Generally, there are two methods in obtaining ET; direct measurement and empirical model. The main objective of this study was to formulate new empirical models to estimate ET for Peninsular Malaysia. The models were developed and validated based on the sixteen years of historical meteorological data obtained from Malaysia Meteorological Department. Using the Multiple Linear Regression (MLR) algorithm, simpler potential evapotranspiration (ETp) estimation models were developed for selected meteorological stations in Peninsular Malaysia. Furthermore, Particle Swarm Optimisation (PSO) was used to optimise the performance of developed models and these models were compared against ETp estimated by the Food and Agricultural Organization Penman-Monteith model. Four different measures of performance indicators were used to highlight its accuracy. The sensitivity analysis showed that air temperature, T and solar radiation, Rs were observed to significantly influenced ET under humid tropical climates such as Peninsular Malaysia. The proposed MLR models at all stations were able to maintain reliable and stable results with an average of 83%, 0.3%, 0.8% and 0.34 for the coefficient of determination (R2), mean bias error (MBE), percentage error (PE) and root mean square error (RMSE) respectively. The accuracy of optimized MLR-PSO models was improved with a minimum of 85%. The minimum accuracy of 85% indicates that the proposed MLR models are already reliable for ETp estimation. The proposed empirical models in this study have successfully contributed to the development of ETp estimation models that is more site-specific and tailored to the Peninsular Malaysia climate due to the easy to obtain and availability of meteorological data such as air temperature and solar radiation. |
---|