Abstarct: Sustainable management of water resources in the agriculture sector – especially in arid and semi-arid regions – relies on accurate knowledge of water consumption. Thus, estimating accurate actual evapotranspiration (AET), as a main component of water balance, is vitally essential. In the present study, the ability of the Least Square Support Vector Machine (LSSVM) model to estimate daily AET across the Neishaboor watershed; April to October 2019; was evaluated. To this end, meteorological parameters (relative humidity, wind speed, and air temperature) and remotely sensed information derived from MODIS satellite images (reflectance of bands 31 and 32, and land surface temperature (LST)) were implemented as the model inputs in the form of three scenarios. The first scenario (No.1) consisted of the meteorological data and band 31 reflectance, the second scenario (No.2) included the meteorological data and reflectance from both bands 31 and 32, and the third scenario (No.3) incorporated all inputs, including LST. The SEBAL-baxsed AET values were used as a benchmark. In all months, the results showed that scenario No.3 outperformed the other two scenarios, with the lowest and highest error reduction being 6.1% (April) and 37.6% (August), respectively; which highlighted the significant role of LST in estimating AET. Additionally, scenario No.3 led to the lowest AET error estimation in August (RRMSE=0.068, RMSE=0.295, NSE=0.964) and June (RRMSE=0.096, RMSE=0.354, NSE=0.959) and applying the scenario No.2 yielded the following results in August (RRMSE=0.091, RMSE=0.391, NSE=0.937) and June (RRMSE=0.114, RMSE=0.432, NSE=0.941). The results indicated that the AET could be estimated as a function of meteorological data and reflectance from both bands 31 and 32 using a LSSVM model. Therefore, the proposed method – through control of water extraction – can play a significant role in the long-term management of watersheds.