A Study on the Prediction of Evapotranspiration Using Freely Available Meteorological Data

Vaz, Pedro J.; Schütz, G.; Guerrero, Carlos; Cardoso, Pedro J. S.

doi:10.1007/978-3-031-08760-8_37

Cited by 4 publications

(9 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…So, in this section, the solar radiation measured by the WS was used as target. Initially, the features used in the solar radiation models were the same as the ones proposed by Vaz et al [7], namely: M onth, Day, maximum and minimum temperature (T empM ax and T empM in), average humidity (HumididtyAvg), average wind speed (W indAvg), dewpoint, and the polynomial feature M onth 2 × Day. However, it was found that the model metrics were improved by dropping dewpoint and the polynomial feature M onth 2 × Day and, instead, adding the sunset hour angle (ω s ) and daylight hours (N ).…”

Section: ) Solar Radiation Estimation Using Ann Methods With a Limite...mentioning

confidence: 99%

“…Using experimental data collected from a WS located in Vale do Lobo, south Portugal, approach (ii.2) achieved the best result with a coefficient of determination (R 2 ) of 0.977, improving results previously achieved in [7]. The quality and generalization of the proposed models was then tested using 11 weather stations of the Colorado Agricultural Meteorological Network (CoAgMET), that is composed of automatic weather stations distributed across the state of Colorado, USA.…”

Section: Introductionmentioning

confidence: 93%

“…Vaz et al [7] used data from a Vale do Lobo WS, in south Portugal, and explored the use of machine learning for ET o estimation. They concluded that instead of directly esti-mating ET o , the best result was obtained by using machine learning for SR estimation, having as input a limited set of meteorological features, and then use that result together with temperature, humidity and wind speed as input to the FAO56-PM equation, achieving an R 2 of 0.975, a MAE of 0.18 mm/day, and a MAPE of 5.51 %.…”

Section: Reference Evapotranspiration Problematic and Known Solutionsmentioning

confidence: 99%

“…The WS is composed of sensors from Davis Instruments, where the following weather parameters are measured periodically throughout the day and stored with a daily resolution: temperature (minimum, maximum, and average), dew point (minimum, maximum, and average), relative humidity (minimum, maximum, and average), solar radiation (maximum and average), wind speed (minimum, maximum, and average), wind direction, atmospheric pressure (minimum, maximum, and average), rain intensity, and precipitation. This is the same dataset previously used by Vaz et al [7], where the use of machine learning algorithms was explored to create ET o and SR estimation models. A ratio of 75 % to 25 % of train and test data was used, respectively, resulting in train data starting from February 1st, 2019 up to February 3rd, 2021, and test data from February 4th, 2021 up to September 30th, 2021.…”

Section: Proposed Models a Experimental Setup And Datasetmentioning

confidence: 99%

“…Furthermore, in the work presented by Vaz et al [7], for the same problem and dataset in study here (using limited weather parameters as input features), several machine learning regression models were compared in their performance, namely: Ordinary Least Squares, Ridge, Lasso, k-Nearest Neighbors, Support Vector Machine, Decision Tree, and Random Forest (RF). Since Random Forest [48], [49] gave the best results for ET o and SR estimation models, RF will also be evaluated, so that the proposed neural network based models can be directly compared with other top performing methods.…”

Section: B Neural Network Models' Architecturementioning

confidence: 99%

See 4 more Smart Citations

Hybrid Neural Network Based Models for Evapotranspiration Prediction Over Limited Weather Parameters

et al. 2023

Self Cite

View full text Add to dashboard Cite

Evapotranspiration can be used to estimate the amount of water required by agriculture projects and green spaces, playing a key role in water management policies that combat the hydrological drought, which assumes a structural character in many countries. In this context, this work presents a study on reference evapotranspiration (ET o ) estimation models, having as input limited meteorological parameters, namely: temperature, humidity, and wind. Since solar radiation (SR) is an important parameter in the determination of ET o , SR estimation models are also developed. These ET o and SR estimation models compare the use of Artificial Neural Networks (ANN), Long Short Term Memory (LSTM), Gated Recurrent Unit (GRU), Recurrent Neural Network (RNN), and hybrid neural network models such as LSTM-ANN, RNN-ANN and GRU-ANN. Two main approaches were taken for ET o estimation: (i) directly use those algorithms to estimate ET o , and (ii) estimate solar radiation first and then use that estimation together with other meteorological parameters in a method that predicts ET o . For the later case, two variants were implemented: the use of the estimated solar radiation as (ii.1) a feature of the neural network regressors, and (ii.2) the use of FAO-56PM method to compute ET o , which has solar radiation as one of the input parameters. Using experimental data collected from a weather station (WS) located in Vale do Lobo (Portugal), the later approach achieved the best result with a coefficient of determination (R 2 ) of 0.977. The developed model were then applied to data from eleven stations located in Colorado (USA), with very distinct climatic conditions, showing similar (R 2 > 0.95) results to the ones for which the models were initially designed, proving a good generalization. As a final notice, the reduced-set features were carefully selected so that they are compatible with free online weather forecast services.

show abstract

Section: ) Solar Radiation Estimation Using Ann Methods With a Limite...mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 93%

Section: Reference Evapotranspiration Problematic and Known Solutionsmentioning

confidence: 99%

Section: Proposed Models a Experimental Setup And Datasetmentioning

confidence: 99%