Evapotranspiration Forecasts Research Articles

MODIS Evapotranspiration Forecasting Using ARIMA and ANN Approach at a Water-Stressed Irrigation Scheme in South Africa

The forecasting of evapotranspiration (ET) in some water-stressed regions remains a major challenge due to the lack of reliable and sufficient historical datasets. For efficient water balance, ET remains the major component and its proper forecasting and quantifying is of the utmost importance. This study utilises the 18-year (2001 to 2018) MODIS ET obtained from a drought-affected irrigation scheme in the Eastern Cape Province of South Africa. This study conducts a teleconnection evaluation between the satellite-derived evapotranspiration (ET) time series and other related remotely sensed parameters such as the Normalised Difference Vegetation Index (NDVI), Normalised Difference Water Index (NDWI), Normalised Difference Drought Index (NDDI), and precipitation (P). This comparative analysis was performed by adopting the Mann–Kendall (MK) test, Sequential Mann–Kendall (SQ-MK) test, and Multiple Linear Regression methods. Additionally, the ET detailed time-series analysis with the Keiskamma River streamflow (SF) and monthly volumes of the Sandile Dam, which are water supply sources close to the study area, was performed using the Wavelet Analysis, Breaks for Additive Seasonal and Trend (BFAST), Theil–Sen statistic, and Correlation statistics. The MODIS-obtained ET was then forecasted using the Autoregressive Integrated Moving Average (ARIMA) and Artificial Neural Networks (ANNs) for a period of 5 years and four modelling performance evaluations such as the Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), and the Pearson Correlation Coefficient (R) were used to evaluate the model performances. The results of this study proved that ET could be forecasted using these two time-series modeling tools; however, the ARIMA modelling technique achieved lesser values according to the four statistical modelling techniques employed with the RMSE for the ARIMA = 37.58, over the ANN = 44.18; the MAE for the ARIMA = 32.37, over the ANN = 35.88; the MAPE for the ARIMA = 17.26, over the ANN = 24.26; and for the R ARIMA = 0.94 with the ANN = 0.86. These results are interesting as they give hope to water managers at the irrigation scheme and equally serve as a tool to effectively manage the irrigation scheme.

Innovative Multi-Temporal Evapotranspiration Forecasting Using Empirical Fourier Decomposition and Bidirectional Long Short-Term Memory

Reference evapotranspiration (ETo) is an essential environmental variable that is intimately significant to agriculture. Managing water and crop planning relies heavily on precise forecasting of ETo. This research used a novel time series decomposition technique, Empirical Fourier Decomposition (EFD), to forecast ETo accurately. Four machine learning techniques were used to forecast ETo using decomposed lagged ETo values. The input data source was from Prince Edward Island (PEI) weather stations (Harrington and St Peters Stations). First, autocorrelation analysis was performed to determine effective lags. Then, ETo data were decomposed using EFD, and lagged data was created based on EFD results. The Kbest feature selection algorithm was used to choose effective inputs, reducing the training time. The accuracy of models was evaluated using different statistical metrics such as correlation coefficient (R) and root mean square error (RMSE). The results showed that using EFD decomposition can significantly improve forecast accuracy. The comparison between different machine learning models showed that the deep learning-based model (Bidirectional LSTM (Long Short Term Memory)) (R=0.956, RMSE= 0.451 mm/day for Harrington station and R=0.956, RMSE= 0.451 mm/day for St Peters station) performed better than the Generalized Regression Neural Network (GRNN), K-nearest neighbor (KNN), and Random Forest (RF) models. Finally, the best model (EFD-Bidirectional LSTM) was used to forecast multitemporal ETo at both stations. Results showed that the developed model can forecast ETo for up to 28 days with reasonable accuracy. However, the accuracy of multi-step ahead forecasting decreases when evapotranspiration values are high, as the models tend to underestimate these values. The findings of this study can assist in accurately calculating crop water requirements and help farmers optimize their irrigation schedules.

Hybrid machine learning system based on multivariate data decomposition and feature selection for improved multitemporal evapotranspiration forecasting

Evapotranspiration is an essential component of the hydrological cycle. Forecasting the reference crop evapotranspiration (ETo) using a reliable and generalized framework is crucial for agricultural operations, especially irrigation. This study was aimed at evaluating the performance of a hybrid system including the K-Best selection (KBest), multivariate variational mode decomposition (MVMD), and Machine learning (ML) models for 1-, 3-, 7-, and 10-day-ahead forecasting of the daily ETo in twelve stations of California. The analysis covered a span of 20 years, from 2003 to 2022. Three stand-alone ML models, namely Cascade Forward Neural Network (CFNN), Extreme Learning Machine (ELM), and Bagging Regression Tree (BRT) are used and were integrated with various preprocessing techniques to construct three hybrid models, i.e., MVMD-KBest-CFNN, MVMD-KBest-ELM, and MVMD-KBest-BRT. According to the results obtained in the testing phase, averaged across all stations, all three stand-alone models (CFNN, ELM, and BRT) yielded similar outcomes. In contrast, the hybrid models exhibited significantly enhanced performances compared with the standalone models, and MVMD-KBest-CFNN and MVMD-KBest-ELM models outperformed MVMD-KBest-BRT model. The BRT-based models were vulnerable to overfitting. The performance of the best models is superior compared to similar existing studies. Examining the variations across stations, it was found that the stations located further from the coast and in arid regions could be susceptible to prediction errors and necessitate more attention.

Advanced evapotranspiration forecasting in Central Italy: Stacked MLP-RF algorithm and correlated Nystrom views with feature selection strategies

Evapotranspiration is a key parameter in hydrology, particularly in the field of water resources management. Reference evapotranspiration (ETo) stands as a crucial metric, embodying the influence of climate on water loss from both soil and plant surfaces. The accurate forecasting of future ETo values is paramount for informed decision-making in agricultural practices and water supply planning. The anticipation of evapotranspiration variations supports optimized irrigation, drought assessment, and efficient water allocation. Employing innovative algorithms, specifically the Multilayer Perceptron-Random Forest (MLP-RF) Stacked Model and the Correlated Nystrom Views (XNV), this study focuses on predicting ETo up to 60 days ahead in Agro Pontino, an area in Central Italy known for its flourishing agricultural production in the Mediterranean Europe. A Radial Basis Function (RBF) Neural Network serves as a benchmark. Taking into account firstly all available weather-climatic variables and subsequently adopting two different strategies for reducing the variables, based respectively on Principal Component Analysis and feature selection using Particle Swarm Optimization, three distinct sets of input variables were considered. The models based on the complete set of exogenous climatic variables demonstrated superior accuracy. However, even models relying on only mean temperature, maximum relative humidity, and shortwave solar radiation as inputs produced good results, also for the 60-day forecasting horizon, with Kling-Gupta Efficiency (KGE) and Mean Absolute Percentage Error (MAPE) equal to 0.98 and 8.356 %, respectively, in the case of MLP-RF Stacked Model. The latter consistently outperformed XNV and RBF across various combinations of input variables and forecasting horizons. Notably, the reduction in accuracy with extended forecasting horizons was mild, suggesting the potential for accurate results over significantly more extended horizons. These forecasting models facilitate precise irrigation scheduling, minimizing water wastage and conserving resources. This targeted approach enhances crop yields, quality, and environmental sustainability, rendering agriculture economically viable and adaptable to climate variability.

Spatial analysis of remote sensing and meteorological indices in a drought event in southwestern Spain

The effects of global warming and climate change are being felt through more extreme and prolonged periods of drought. Multiple meteorological indices are used to measure drought, but they require hydrometeorological data; however, other indices measured by remote sensing and used to quantify vegetation vigor can be correlated with the former. This study investigated the correlation between both index types by vegetation type and season. The correlations were also spatially modeled in a drought event in southwestern Spain. In addition, three maps with different levels of detail in terms of vegetation categorization were compared. The results generally showed that grassland was the most well correlated category between the SPEI and the FAPAR, LAI, and NDVI. This correlation was more pronounced in autumn and spring, which is when most changes in vegetation senescence and growth occur. The spatiotemporal analysis indicated a very similar behavior for grasslands grouped in an area indicated by the climate change adaptation maps as having a high evapotranspiration forecast. Finally, in a forest-based forecast analysis, the indices that best explained the performance of the SPEI were again FAPAR, LAI, and NDVI, with a lag of up to 20 days. Therefore, the results showed that remotely sensed indices are good indicators of drought status and can be variably explanatory of traditional drought indicators. Moreover, complementing the study with spatiotemporal analysis made it possible to detect areas particularly vulnerable to climate change.

A review of recent advances and future prospects in calculation of reference evapotranspiration in Bangladesh using soft computing models

Evapotranspiration (ETo) is a complex and non-linear hydrological process with a significant impact on efficient water resource planning and long-term management. The Penman-Monteith (PM) equation method, developed by the Food and Agriculture Organization of the United Nations (FAO), represents an advancement over earlier approaches for estimating ETo. Eto though reliable, faces limitations due to the requirement for climatological data not always available at specific locations. To address this, researchers have explored soft computing (SC) models as alternatives to conventional methods, known for their exceptional accuracy across disciplines. This critical review aims to enhance understanding of cutting-edge SC frameworks for ETo estimation, highlighting advancements in evolutionary models, hybrid and ensemble approaches, and optimization strategies. Recent applications of SC in various climatic zones in Bangladesh are evaluated, with the order of preference being ANFIS > Bi-LSTM > RT > DENFIS > SVR-PSOGWO > PSO–HFS due to their consistently high accuracy (RMSE and R2). This review introduces a benchmark for incorporating evolutionary computation algorithms (EC) into ETo modeling. Each subsection addresses the strengths and weaknesses of known SC models, offering valuable insights. The review serves as a valuable resource for experienced water resource engineers and hydrologists, both domestically and internationally, providing comprehensive SC modeling studies for ETo forecasting. Furthermore, it provides an improved water resources monitoring and management plans.

Statistical and deep learning models for reference evapotranspiration time series forecasting: A comparison of accuracy, complexity, and data efficiency

Reference evapotranspiration (ETo) is an essential variable in agricultural water resources management and irrigation scheduling. An accurate and reliable forecast of ETo facilitates effective decision-making in agriculture. Although numerous studies assessed various methodologies for ETo forecasting, an in-depth multi-dimensional analysis evaluating different aspects of these methodologies is missing. This study systematically evaluates the complexity, computational cost, data efficiency, and accuracy of ten models that have been used or could potentially be used for ETo forecasting. These models range from well-known statistical forecasting models like seasonal autoregressive integrated moving average (SARIMA) to state-of-the-art deep learning (DL) algorithms like temporal fusion transformer (TFT). This study categorizes monthly ETo time series from 107 weather stations across California according to their length to better understand the forecasting models' data efficiency. Moreover, two forecasting strategies (i.e., recursive and multi-input multi-output) are employed for machine learning and DL models, and forecasts are assessed for different multi-step horizons. Our findings show that statistical forecasting models like Holt-Winters' exponential smoothing perform almost as well as complex DL models. Unlike statistical models, DL models generally suffer from low data efficiency and perform well only when enough data is available. Importantly, although the computational costs of most DL models are higher than statistical methods, this is not the case for all. Considering computational cost, data efficiency, and forecasting accuracy, our findings point to the superiority of the neural basis expansion analysis for interpretable time series forecasting (N-BEATS) architecture for univariate ETo time series forecasting. Moreover, our results suggest Holt-Winters and Theta methods outperform SARIMA – the most employed statistical model for ETo forecasting in the literature – in accuracy and efficiency.

Explainable hybrid deep learning and Coronavirus Optimization Algorithm for improving evapotranspiration forecasting

Reference evapotranspiration is a critical hydrological measurement closely associated with agriculture. Accurate forecasting is vital in effective water management and crop planning in sustainable agriculture. In this study, the future values of reference evapotranspiration are forecasted by applying a recurrent long short-term memory neural network optimized using the Coronavirus Optimization Algorithm, a novel bioinspired metaheuristic based on the spread of COVID-19. The input data is sourced from the Sistema Agrometeorológico para a Gestão da Rega no Alentejo, in Portugal, with meteorological data such as air temperature or wind speed. Several baseline models are applied to the same problem to facilitate comparisons, including support vector machines, multi-layer perceptron, Lasso and decision tree. The results demonstrate the successful forecasting performance of the proposed model and its potential in this field. In turn, to gain deeper insights into the model’s inner workings, the SHapley Additive exPlanation tool is applied for explainability. Consequently, the study identifies the most relevant variables for reference evapotranspiration forecasting, including previously measured evapotranspiration values. Additionally, a univariable model is tested using historic evapotranspiration values as input, offering a comparable performance with a considerable reduction of computational time.

New achievements on daily reference evapotranspiration forecasting: Potential assessment of multivariate signal decomposition schemes

Reference evapotranspiration (ETo) is a vital climate parameter affecting plants' water use. ETo can generate large deficits in soil moisture and runoff in different regions and seasons, leading to uncertainties in drought warning systems. A novel multivariate variational mode decomposition integrated with a boosted regression tree (i.e., MVMD-BRT) is constructed to forecast daily ETo. Firstly, the correlation matrix based on cross-correlation was computed to investigate the significant input predictor lags of daily ETo. Secondly, the MVMD technique decomposes the significant input lags into signals called intrinsic mode functions (IMFs). Thirdly, the IMFs were then employed in the BRT to build the MVMD-BRT model for daily ETo forecasting. A comparative assessment of MVMD against multivariate empirical mode decomposition (MEMD) was also performed on the same lines to develop the MEMD-BRT model. The MVMD-BRT model is compared against the random forest (RF) and hybrid MVMD-RF, MEMD-RF, extreme learning machine (ELM), and hybrid MVMD-ELM, MEMD-ELM, and cascaded feedforward neural network (CFNN) along with its hybrid MVMD-CFNN models for two stations in Queensland, Australia using a set of goodness-of-fit metrics. The results prove that the MVMD-BRT provide accurate daily ETo forecasting against the benchmark models. The MVMD-BRT model yielded the highest accuracy in terms of (WIE = 0.9070, NSE = 0.8421, LME = 0.6529, KGE = 0.8792) and (WIE = 0.8966, NSE = 0.8396, LME = 0.6521, KGE = 0.8803) for Brisbane and Gympie stations against the comparing models.

Assessing the Potential of Hybrid-Based Metaheuristic Algorithms Integrated with ANNs for Accurate Reference Evapotranspiration Forecasting

Evapotranspiration (ETo) is one of the most important processes in the hydrologic cycle, with specific application to sustainable water resource management. As such, this study aims to evaluate the predictive ability of a novel method for monthly ETo estimation, using a hybrid model comprising data pre-processing and an artificial neural network (ANN), integrated with the hybrid particle swarm optimisation–grey wolf optimiser algorithm (PSOGWO). Monthly data from Al-Kut City, Iraq, over the period 1990 to 2020, were used for model training, testing, and validation. The predictive accuracy of the proposed model was compared with other cutting-edge algorithms, including the slime mould algorithm (SMA), the marine predators algorithm (MPA), and the constriction coefficient-based particle swarm optimisation and chaotic gravitational search algorithm (CPSOCGSA). A number of graphical methods and statistical criteria were used to evaluate the models, including root mean squared error (RMSE), Nash–Sutcliffe model efficiency (NSE), coefficient of determination (R2), maximum absolute error (MAE), and normalised mean standard error (NMSE). The results revealed that all the models are efficient, with high simulation levels. The PSOGWO–ANN model is slightly better than the other approaches, with an R2 = 0.977, MAE = 0.1445, and RMSE = 0.078. Due to its high predictive accuracy and low error, the proposed hybrid model can be considered a promising technique.

Short-term daily reference evapotranspiration forecasting using temperature-based deep learning models in different climate zones in China

The reference evapotranspiration (ETo) pertains to the evapotranspiration of cold-season grasses with an approximate height of 0.12 m or full-covered alfalfa with a height of 0.50 m. Accurate short-term ETo forecasts are indispensable for informed irrigation decisions by relevant departments and individuals. Four deep learning (DL) models, including Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), Bidirectional LSTM (Bi-LSTM), and Bidirectional GRU (Bi-GRU), as well as two calibrated empirical models (Hargreaves-Samani (HS) and reduced-set Penman–Monteith (RPM)), were used to evaluate the performance of the ETo forecast with a lead time of 1–7 d using temperature forecasts in different climates. The results reveal that the DL models and calibrated HS and RPM models exhibited comparable trends in the ETo forecasts for lead times of 1–7 d. Nonetheless, the DL models consistently outperformed the HS and RPM models across the diverse climatic regions in China. The DL models displayed an average root mean square error (RMSE) and mean absolute error (MAE) of less than 0.887 and 0.633 mm/d, respectively. Moreover, the mean correlation coefficient (R) and accuracy (ACC) exceeded 0.807% and 89.701%, respectively. Among the DL models, the LSTM model demonstrated slightly superior performance in short-term daily ETo forecasts in diverse climates. The LSTM model exhibited RMSE and MAE ranges of 0.563–0.875 mm/d and 0.418–0.626 mm/d, respectively, along with R and ACC ranges of 0.81–0.90 and 89.94–98.11%, respectively. Furthermore, even with an increase in lead time, the DL models continued to exhibit strong predictive capabilities, consistently surpassing the performance of the HS and RPM models. Overall, the trained DL models presented an exceptional ability to forecast the short-term daily ETo in various climatic regions of China. These models require only a few input variables and readily available data, making them highly advantageous for practical applications in ETo forecasting. Such models hold promise for significantly enhancing regional agricultural water-resource planning and management.

IoT and Ensemble Long-Short-Term-Memory-Based Evapotranspiration Forecasting for Riyadh.

Evapotranspiration (ET) is the fundamental component of efficient water resource management. Accurate forecasting of ET is essential for efficient water utilization in agriculture. ET forecasting is a complex process due to the requirements of large meteorological variables. The recommended approach is based on the Internet of Things (IoT) and an ensemble-learning-based approach for meteorological data collection and ET forecasting with limited meteorological conditions. IoT is part of the recommended approach to collect real-time data on meteorological variables. The daily maximum temperature (T), mean humidity (Hm), and maximum wind speed (Ws) are used to forecast evapotranspiration (ET). Long short-term memory (LSTM) and ensemble LSTM with bagged and boosted approaches are implemented and evaluated for their accuracy in forecasting ET values using meteorological data from 2001 to 2023. The results demonstrate that the bagged LSTM approach accurately forecasts ET with limited meteorological conditions in Riyadh, Saudi Arabia, with the coefficient of determination (R2) of 0.94 compared to the boosted LSTM and off-the-shelf LSTM with R2 of 0.91 and 0.77, respectively. The bagged LSTM model is also more efficient with small values of root mean squared error (RMSE) and mean squared error (MSE) of 0.42 and 0.53 compared to the boosted LSTM and off-the-shelf LSTM models.

Exploring association and forecasting of evapotranspiration based on meteorological factors over megacity Lahore (Pakistan) and central place of Indo-Gangetic Basin

Exploring association and forecasting of evapotranspiration based on meteorological factors over megacity Lahore (Pakistan) and central place of Indo-Gangetic Basin

Machine learning models to predict daily actual evapotranspiration of citrus orchards under regulated deficit irrigation

Precise estimations of actual evapotranspiration (ETa) are essential for various environmental issues, including those related to agricultural ecosystem sustainability and water management. Indeed, the increasing demands of agricultural production, coupled with increasingly frequent drought events in many parts of the world, necessitate a more careful evaluation of crop water requirements.Artificial Intelligence-based models represent a promising alternative to the most common measurement techniques, e.g. using expensive Eddy Covariance (EC) towers. In this context, the main challenges are choosing the best possible model and selecting the most representative features. The objective of this research is to evaluate two different machine learning algorithms, namely Multi-Layer Perceptron (MLP) and Random Forest (RF), to predict daily actual evapotranspiration (ETa) in a citrus orchard typical of the Mediterranean ecosystem using different feature combinations. With many features available coming from various infield sensors, a thorough analysis was performed to measure feature importance, scatter matrix observations, and Pearson’s correlation coefficient calculation, which resulted in the selection of 12 promising feature combinations. The models were calibrated under regulated deficit irrigation (RDI) conditions to estimate ETa and save irrigation water. On average up to 38.5% water savings were obtained, compared to full irrigation. Moreover, among the different input variables adopted, the soil water content (SWC) feature appears to have a prominent role in the prediction of ETa. Indeed, the presented results show that by choosing the appropriate input features, the accuracy of the proposed machine learning models remains acceptable even when the number of features is reduced to only 4. The best performance was achieved by the Random Forest method, with seven input features, obtaining a root mean square error (RMSE) and a coefficient of determination (R2) of 0.39 mm/day and 0.84, respectively. Finally, the results show that the joint use of SWC, weather and satellite data significantly improves the performance of evapotranspiration forecasts compared to models using only meteorological variables.

Fitting statistical probability distribution for evaporation in western Brazil

This study aims to find evaporation (ETo) probabilities for the central west region of Brazil, including Campo Grande, Cuiabá, Goiânia and Brasilia. In this case, a maximum likelihood estimation was used for the probability distributions adjusting to the monthly ETo from 1960 to 2020. The work estimates three probability distributions, including the Generalized Extreme Values (GEV), Gumbel (GUM) and Log-Normal (LN). The Kolmogorov-Smirnov test, the Akaike information criterion (AIC), the Bayesian information criterion (BIC), the root mean square error (RMSE) and the coefficient of determination (R2) were applied to measure the quality of the adjustment. The evapotranspiration forecasts were made for the return periods of 10, 30, 50, and 100 years. The GEV and GUM distributions are advised for use during the summer. Altering LN distribution to a sequence of ETo throughout the winter months is the best option. Moreover, deforestation, combustion, extensive fires, aerosol pollutants, and climate change are the primary causes of excessive ETo in the area under study.

Vector Autoregression Model-Based Forecasting of Reference Evapotranspiration in Malaysia

Evapotranspiration is one of the hydrological cycle’s most important elements in water management across economic sectors. Critical applications in the agriculture domain include irrigation practice improvement and efficiency, as well as water resource preservation. The main objective of this research is to forecast reference evapotranspiration using the vector autoregression (VAR) model and investigate the meteorological variables’ causal relationship with reference evapotranspiration using a statistical approach. The acquired 20-year, 1-year, and 2-month research climate datasets from Penang, Malaysia, were split into 80% training data and 20% validation data. Public weather data are used to train the initial VAR model. A Raspberry Pi IoT device connected to a DHT11 temperature sensor was outfitted at the designated experimental crop site. In situ data acquisition was done using DHT11 temperature sensors to measure the ambient temperature and humidity. The collected temperature and humidity data were used in conjunction with the vector autoregression (VAR) model to calculate the reference evapotranspiration forecast. The results demonstrated that the 20-year dataset showed better performance and consistent results in forecasting general reference evapotranspiration, derived using root mean square error (RMSE) and correlation coefficient (CORR) of 1.1663 and −0.0048, respectively. As for the 1-year dataset model, RMSE and CORR were recorded at 1.571 and −0.3932, respectively. However, the 2-month dataset model demonstrated both positive and negative performance due to seasonal effects in Penang. The RMSE ranged between 0.5297 to 2.3562 in 2020, 0.8022 to 1.8539 in 2019, and 0.8022 to 2.0921 in 2018. As for CORR, it ranged between −0.5803 to 0.2825 in 2020, −0.3817 to 0.2714 in 2019, and −0.3817 to 0.2714 in 2018. In conclusion, the model tested using 20-year, 1-year, and 2-month meteorological datasets for estimating reference evapotranspiration (ET0) based on smaller RMSEs demonstrates better performance at predicting the true values, as well as producing both positive and negative CORR performance due to seasonal variations in Penang.

An attention-based LSTM model for long-term runoff forecasting and factor recognition

With advances in artificial intelligence, machine learning-based models such as long short-term memory (LSTM) models have shown much promise in forecasting long-term runoff by mapping pathways between large-scale climate patterns and catchment runoff responses without considering physical processes. The recognition of key factors plays a vital role and thus affects the performance of the model. However, there is no conclusion on which recognition algorithm is the most suitable. To address this issue, an LSTM model combined with two attention mechanisms both in the input and hidden layers, namely AT-LSTM, is proposed for long-term runoff forecasting at Yichang and Pingshan stations in China. The added attention mechanisms automatically assign weights to 130 climate phenomenon indexes, avoiding the use of subjectively set recognition algorithms. Results show that the AT-LSTM model outperforms the Pearson’s correlation based LSTM model in terms of four evaluation metrics for monthly runoff forecasting. Further, the set indirect runoff prediction method verifies that the AT-LSTM model also performs effectively in precipitation and potential evapotranspiration forecasting, and the indirect runoff prediction is inferior to the AT-LSTM model to establish a direct link between climate factors and runoff. Finally, four key factors related to runoff are identified by the attention mechanism and their impacts on runoff are analyzed on intra- and inter-annual scales. The proposed AT-LSTM model can effectively improve the accuracy of long-term forecasting and identify the dynamic influence of input factors.

Short–term forecasting of daily evapotranspiration from rice using a modified Priestley–Taylor model and public weather forecasts

Reliable crop evapotranspiration (ET) forecasting is critical for water resource management and allocation. However, most estimation models cannot be directly used for ET forecasting with public weather forecasting information. Here, we propose a modified Priestley–Taylor (PT) model using only solar radiation (Rs) and air temperature (Ta) as inputs that can be easily obtained from public weather forecasts. The commonly used generalized sunshine– and temperature–based Rs models using forecasted weather types (transferred to sunshine duration) and Ta data as inputs were compared to identify the more suitable model for forecasting daily Rs. Results showed that the modified PT model using measured Rs and Ta data provided a reasonable accuracy to estimate daily rice ET across three energy flux sites. The regression coefficient (m), correlation coefficient (R), root mean squared error (RMSE), and index of agreement (dIA) were in the range of 0.949–1.044, 0.916–0.964, 0.396–0.645 mm d–1, and 0.954–0.980, respectively. The generalized temperature–based Rs model is more suitable for forecasting daily Rs for a 1–7 day lead time during the growth period of rice at present as indicated by its superior performance to sunshine–based model at 70% of 30 sites. The forecasted accuracy of rice ET was acceptable and decreased gradually as lead time increased. The mean values of m, R, RMSE, and dIA were decreased from 0.887 to 0.981, 0.641–0.782, 1.160–1.243 mm d–1, and 0.783–0.870 in 1–day lead time to 0.834–0.941, 0.376–0.563, 1.406–1.740 mm d–1, and 0.639–0.743, respectively, in 7–day lead time during 2018–2020 at Nanjing when Rs was forecasted based on the generalized temperature–based model. This study provides an important approach for short–term prediction of rice ET using public weather forecasting information.

Applicability of hybrid bionic optimization models with kernel-based extreme learning machine algorithm for predicting daily reference evapotranspiration: a case study in arid and semiarid regions, China.

The accurate prediction of daily reference crop evapotranspiration (ETO) enables effective management decision-making for agricultural water resources; this is crucial for developing water-efficient agriculture. To improve the accuracy of ETO forecasts in data-deficient areas, this study uses a decision tree algorithm (classification and regression tree [CART]) to obtain the effects of various factors on ETO at typical stations in arid and semiarid regions of China. A combination of factors with considerable influence on the model was selected as the input for constructing a kernel-extreme-learning-machine (KELM) daily reference evapotranspiration prediction model, and three bionic optimization algorithms (i.e., sparrow search optimization algorithm, Harris Hawks optimization algorithm, and lion swarm optimization algorithm) were integrated to optimize KELM prediction model parameters and improve the accuracy of daily reference evapotranspiration prediction. The results indicate that temperature (maximum or minimum temperature) is the primary factor influencing ETO, and the range of importance is 0.399-0.554. RH and Ra are also key factors influencing ETO; the hybrid model optimized using the bionic optimization algorithm provides advantages over the independent KELM model, and the SSA-KELM model has the highest accuracy among hybrid models, with a root-mean-square error of 0.408-1.964, R2 of 0.545-0.982, mean absolute error of 0.273-1.086, and Nash-Sutcliffe efficiency coefficient of 0.658-0.967. The top five factors extracted using the CART algorithm are recommended as inputs for constructing the SSA-KELM model for ETO estimation in arid and semiarid regions of China, and this model can also serve as a reference for ETO forecasting in similar regions.

Evaluation of Five Equations for Short-Term Reference Evapotranspiration Forecasting Using Public Temperature Forecasts for North China Plain

Accurate short-term forecasts of daily reference evapotranspiration (ET0) are essential for real-time irrigation scheduling. Many models rely on current and historical temperature data to estimate daily ET0. However, easily accessible temperature forecasts are relatively less reported in short-term ET0 forecasting. Furthermore, the accuracy of ET0 forecasting from different models varies locally and also across regions. We used five temperature-dependent models to forecast daily ET0 for a 7-day horizon in the North China Plain (NCP): the McCloud (MC), Hargreaves-Samani (HS), Blaney-Criddle (BC), Thornthwaite (TH), and reduced-set Penman–Monteith (RPM) models. Daily meteorological data collected between 1 January 2000 and 31 December 2014 at 17 weather stations in NCP to calibrate and validate the five ET0 models against the ASCE Penman–Monteith (ASCE-PM). Forecast temperatures for up to 7 d ahead for 1 January 2015–19 June 2021 were input to the five calibrated models to forecast ET0. The performance of the five models improved for forecasts at all stations after calibration. The calibrated RPM is the preferred choice for forecasting ET0 in NCP. In descending order of preference, the remaining models were ranked as HS, TH, BC, and MC. Sensitivity analysis showed that a change in maximum temperature influenced the accuracy of ET0 forecasting by the five models, especially RPM, HS, and TH, more than other variables. Meanwhile, the calibrated RPM and HS equations were better than the other models, and thus, these two equations were recommended for short-term ET0 forecasting in NCP.