Modeling Daily Reference Evapotranspiration from Climate Variables: Assessment of Bagging and Boosting Regression Approaches

Abstract

The increasing frequency of droughts and floods due to climate change has severely affected water resources across the globe in recent years. An optimal design for the scheduling and management of irrigation is thus urgently needed to adapt agricultural activities to the changing climate. The accurate estimation of reference crop evapotranspiration (ET0), a vital hydrological component of the water balance and crop water need, is a tiresome task if all the relevant climatic variables are unavailable. This study investigates the potential of four ensemble techniques for estimating precise values of the daily ET0 at representative stations in 10 agro-climatic zones in the state of Karnataka, India, from 1979 to 2014. The performance of these models was evaluated by using several combinations of climatic variables as inputs by using tenfold cross-validation. The outcomes indicated that predictions of ET0 by all four ensemble models based on all climatic variables were the most accurate in comparison with other input combinations. The random forest regressor was found to deliver the best performance among the four models on all measures considered (Nash–Sutcliffe efficiency, 1.0, root-mean-squared error, 0.016 mm/day, and mean absolute error, 0.011 mm/day). However, it incurred the highest computational cost, whereas the computational cost of the bagging model for linear regression was the lowest. The extreme gradient-boosting model delivered the most stable performance with a modified training dataset. The work here shows that these models can be recommended for daily ET0 estimation based on the users’ interests.

Reference evapotranspiration prediction using machine learning models: An empirical study from minimal climate data

“…The PM model was used as the reference model in this study, and various ML models based on LR, RF, NN, SVM, etc., were developed to estimate ET 0 with limited climatic variables and their combinations as input parameters. The study findings align with previous research studies, which suggested that high prediction accuracy can be achieved by incorporating all the climatic parameters in the models (Ayaz et al., 2021; Elbeltagi et al., 2022; Jayashree et al., 2023). However, according to Alves et al.…”

“…Under the limited availability of all the parameters, the effectiveness of the PM method remains compromised in determining ET 0 as in the case of CROPWAT 8.0 software, which provides average wind velocity data, which does not vary in months, and thus software computes ET 0 either under/over in the absence of actual wind speed data for a given location. ML algorithms have the capability of handling real-time massive data to accurately estimate ET 0 (Jayashree et al, 2023;Kim et al, 2022), particularly in locations where data are scarce. As a result, the purpose of this work is to assess ML techniques for estimating ET 0 with limited climatic inputs.…”

Smart irrigation for coriander plant: Saving water with AI and IoT

“…Notably, our model boasts the highest Rsquared values (0.98-0.99), which indicates a very strong correlation between predicted and observed EV T 0 values. While Tr et al [16] achieved a lower RMSE (0.016), their results may not be directly comparable due to potentially limited data scope or specific application. Compared to other studies (Hill et al [9], Hu et al [12], Dong et al [18]), our approach demonstrates significantly lower and more consistent RMSE and MAE values across a wider range, which suggests greater generalizability and accuracy.…”

Assessment of Empirical Methods for Estimating Reference Evapotranspiration in Different Climatic Zones of Bosnia and Herzegovina

“…However, the FAO-PM method is not without limitations. The main disadvantage of the FAO-PM method is the need for numerous input data, which are often not available at meteorological sites [19][20][21], especially in developing countries such as Bosnia and Herzegovina (BiH) [19] or other regions [22]. When the data are available, they are often inaccurate due to poor maintenance of meteorological sites and sensors, especially those for solar radiation, humidity, and windspeed measurement [23].…”

“…There are numerous studies in which different methods were tested against the Penman-Monteith method, and the results were dependent on climatic conditions and the specific site location, the period of observation and the type of data (daily, monthly) used for the ET o estimate, and the site elevation as well as seaside vicinity [19][20][21][22][23][24][25]. The remarkable development of artificial intelligence in recent years has enabled researchers to acquire large data sets with non-linear relationships between various climatic variables to accurately predict ET o [21]. Models such as ANNs (Artificial Neural Networks), ELMs (Extreme Learning Machines), SVMs (Support Vector Machines), ANFISs (Adaptive Neuro-Fuzzy Inference Systems), etc.…”