Publication:
Utilising Machine Learning for Pan Evaporation Prediction - A Case Study in Western Australia

Simple item page
dc.citedby1
dc.contributor.authorAbed M.en_US
dc.contributor.authorImteaz M.A.en_US
dc.contributor.authorAhmed A.N.en_US
dc.contributor.authorid36612762700en_US
dc.contributor.authorid6506146119en_US
dc.contributor.authorid57214837520en_US
dc.date.accessioned2024-10-14T03:20:09Z
dc.date.available2024-10-14T03:20:09Z
dc.date.issued2023
dc.description.abstractEvaporation has a significant impact on the management of water resources, irrigation system designs, and hydrological modelling due to its complex and nonlinear nature. This is because evaporation is a result of the interactions of various climatic factors. In Australia, research suggests that evaporation causes about 40% of the water in open water lakes to be lost each year. Given the potential consequences of climate change, this water loss could become a major issue. This paper presents efficiency of Transformer Neural Network (TNN) approach in predicting monthly pan evaporation (Ep) through a case study in Perth, the capital of Western Australia. Daily meteorological data from a weather station in Perth was deployed for testing and training the model by utilising weather parameters, including maximum temperature, minimum temperature, solar radiation, relative humidity, and wind speed for the period 2009�2022. The Pearson correlation coefficient was used to determine the optimal ML model input parameters. Several models have been developed by combining different input combinations and other model parameters. To evaluate the ML model's performance, it was compared to Stephens and Stewart, a widely used empirical technique. The model's performance was subsequently assessed using standard statistical measures. The results of the performance evaluation criteria suggest that the Transformer model proposed in this study can effectively predict the monthly evaporation rate, benefiting from its self-attention mechanism. The proposed model performed admirably (R2=0.986, RMSE=0.031, MAE=0.025, and NSE=0.987). Additionally, it was demonstrated that the transformer model was more accurate than the empirical method for the same input sets, leading to a notable improvement in the estimation of monthly evaporation rates. � 2023 Newswood Limited. All rights reserved.en_US
dc.description.natureFinalen_US
dc.identifier.epage18
dc.identifier.scopus2-s2.0-85170522450
dc.identifier.spage14
dc.identifier.urihttps://www.scopus.com/inward/record.uri?eid=2-s2.0-85170522450&partnerID=40&md5=d2dbc630ddacc3395ee3669165aaf58d
dc.identifier.urihttps://irepository.uniten.edu.my/handle/123456789/34493
dc.identifier.volume2245
dc.pagecount4
dc.publisherNewswood Limiteden_US
dc.sourceScopus
dc.sourcetitleLecture Notes in Engineering and Computer Science
dc.subjectEvaporation
dc.subjectSelf-Attention
dc.subjectStephens and Stewart Model
dc.subjectTransformer Model
dc.subjectClimate change
dc.subjectCorrelation methods
dc.subjectForecasting
dc.subjectMachine learning
dc.subjectWater management
dc.subjectWind
dc.subjectCase-studies
dc.subjectEvaporation rate
dc.subjectMachine-learning
dc.subjectModeling performance
dc.subjectPan evaporation
dc.subjectSelf-attention
dc.subjectStephen and stewart model
dc.subjectTransformer modeling
dc.subjectWaters resources
dc.subjectWestern Australia
dc.subjectEvaporation
dc.titleUtilising Machine Learning for Pan Evaporation Prediction - A Case Study in Western Australiaen_US
dc.typeConference Paperen_US
dspace.entity.typePublication
Files
Collections
SCOPUS