Tze Ying Fong0000-0001-7287-3394Charlene KOO Chai-Hoon0000-0001-6565-947XYuk Feng Huang0000-0001-5422-5123Chin Ren Jie2025-10-272025-10-2720259783031888496978303188850210.1007/978-3-031-88850-2_3https://dspace-cris.utar.edu.my/handle/123456789/11592Reference evapotranspiration, ETo plays a vital role in the management of agricultural water resources and irrigation schedule. Accurate quantification and estimation of ETo are necessary to ensure the sustainability of water supply for well-informed decisions. However, limited availability and low quality of meteorological data create significant challenges in accurately estimating ETo. Remote sensing become the potential alternative data source to address the limitations posed by ground measurement. In this study, the random forest (RF), support vector machine (SVM) and gated recurrent unit (GRU) were used to estimate ETo based on remote sensing variables at Pulau Langkawi and Kota Bharu located on the West Coast and East Coast of Peninsular Malaysia, respectively. According to the obtained results, the GRU model using downward shortwave radiation (DSR) and daytime land surface temperature (LST) achieved the highest accuracy with RMSE and R2 values as 0.392 mm/day and 0.714 for Pulau Langkawi and 0.407 mm/day, 0.644 for Kota Bharu. DSR proved to be a crucial parameter and had a greater impact on the estimation of ETo compared to daytime LST. However, the importance of daytime LST cannot be neglected, as incorporating DSR and daytime LST as input variables improves prediction accuracy. The findings from this study indicate that utilising remote sensing data for ETo estimation is a viable alternative, particularly when the meteorological data are unavailable. Nonetheless, it is worth mentioning that the significance of remote sensing variables in ETo estimation could be affected by factors such as weather conditions, seasonality and cloud cover at a particular region, which should be further investigated. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.enDeep learningGRULand surface temperatureMachine learningRadiationRFSVMAgricultural machineryAtmospheric temperatureDecision treesEvapotranspirationIrrigationLearning systemsMeteorologyRemote sensingSupport vector machinesSurface measurementSurface propertiesGated recurrent unitMachine-learningRandom forestsReference evapotranspirationRemote sensing dataRemote-sensingShort-wave radiationSupport vectors machinetext::conference output::conference proceedings::conference paper