Search

Is Atlantic Multidecadal Oscillation a genuine representation of natural variability in the climate system? Or perhaps is it strongly forced by external drivers? In this paper, a data-driven attribution investigation has been performed for the Atlantic Multidecadal Oscillation (AMO) behaviour in the past via a machine learning technique, NN modelling. We clearly see a forced nature of AMO in the last 150 years, with a strong contribution of the forcing coming from anthropogenic sulphates, which induces its typical oscillating behaviour. The following original application of our model to future predictions of the AMO behaviour shows that it shall probably lose its oscillating characteristic features. The only way to recover them is to consider an unrealistic increase in anthropogenic sulphates in the future under a strong mitigation scenario, and possibly a low-power solar regime. Due to the established influence of AMO on climate and meteorological phenomena in several regions of the world, our results can be important to better understand the past and envisage several future scenarios. [ABSTRACT FROM AUTHOR]

In traditional artificial neural networks (ANN) models, the relative importance of the individual meteorological input variables is often overlooked. A case study is presented in this paper to model monthly wind speed values using meteorological data (air pressure, air temperature, relative humidity, and precipitation), where the study also includes an estimate of the relative importance of these variables. Recorded monthly mean data are available at a gauging site in Tabriz, Azerbaijan, Iran, for the period from 2000 to 2005, gauged in the city at the outskirt of alluvial funneling mountains with an established microclimatic conditions and a diurnal wind regime. This provides a sufficiently severe test for the ANN model with a good predictive capability of 1 year of lead time but without any direct approach to refer the predicted results to local microclimatic conditions. A method is used in this paper to calculate the relative importance of each meteorological input parameters affecting wind speed, showing that air pressure and precipitation are the most and least influential parameters with approximate values of 40 and 10 %, respectively. This gained knowledge corresponds to the local knowledge of the microclimatic and geomorphologic conditions surrounding Tabriz. [ABSTRACT FROM AUTHOR]

Spatial interpolation methods usually differ in their underlying mathematical concepts. Each has inherent advantages and disadvantages, and choosing a method should be based on the type of data to be analyzed. This paper, therefore, compares and evaluates the performances of well-established interpolation techniques that can be used to estimate monthly rainfall in Thailand. The approaches analyzed include inverse distance weighting (IDW), inverse exponential weighting (IEW), multiple linear regression (MLR), artificial neural networks (ANN), and ordinary kriging (OK) methods. In addition, a search of the nearest stations has also been conducted for some of the aforementioned schemes. A k-fold cross-validation is exploited to assess the efficiency of each method. Results show that ANN might be the least desirable choice as it underperformed, with the remaining methods being roughly comparable. Considering both accuracy and computational flexibility, the IEW approach with a restricted number of neighboring stations is recommended in this study. [ABSTRACT FROM AUTHOR]

Changes in amount and frequency of runoff are caused by interaction of the climatic and anthropogenic factors in human-nature system. Those part of the runoff variations caused by anthropogenic loadings are more manageable compared with climate effects occurring over a long period. In this paper, we propose a framework to decompose the effect of anthropogenic and climatic time series on streamflow. To fulfill this task, anthropogenic time series attribution (ATSA) method is introduced. ATSA employs the output of climate elasticity (CE) method to extract human-affected time series and naturalizes the annual streamflow. In other words, ATSA makes it possible to exploit CE output in time series analysis. Furthermore, temporal downscaling process performed by a hybrid discrete wavelet transform and artificial neural network (DWT-ANN) enhances the temporal resolution of ATSA. We applied ATSA framework on Urmia Lake (northwest of Iran) inflow over the 1972–2010 period. CE application indicated that the contribution of climatic and anthropogenic factors in reduction of the mean inflow to the lake is about 60 and 40%, respectively. ATSA also revealed how human exploitation affected the time series of the lake inflow. Roughly, 14,000 MCM of the lake inflow reduction was attributed to the anthropogenic impacts between 1995 and 2010 such that the removal of anthropogenic loading could have led to rising water level up to 1275.5 meters above the sea level (i.e., 1.4 m above the ecological level) in 2010. Furthermore, the effect of de-trended inflow time series was investigated via DWT-ANN. Trend-free pre-whitening Mann-Kendall (TFPW-MK) test indicated significant increase in temperature, potential evapotranspiration, population, reservoir capacity, irrigated area, and the number of wells, over 39 years of the study period. [ABSTRACT FROM AUTHOR]

Several studies have been devoted to dynamic and statistical downscaling for both climate variability and climate change. This paper introduces an application of temporal neural networks for downscaling global climate model output and autocorrelation functions. This method is proposed for downscaling daily precipitation time series for a region in the Amazon Basin. The downscaling models were developed and validated using IPCC AR4 model output and observed daily precipitation. In this paper, five AOGCMs for the twentieth century (20C3M; 1970–1999) and three SRES scenarios (A2, A1B, and B1) were used. The performance in downscaling of the temporal neural network was compared to that of an autocorrelation statistical downscaling model with emphasis on its ability to reproduce the observed climate variability and tendency for the period 1970–1999. The model test results indicate that the neural network model significantly outperforms the statistical models for the downscaling of daily precipitation variability. [ABSTRACT FROM AUTHOR]

Due to the spatial-temporal inadequacy of large-scale general circulation models (GCMs), linking large-scale GCM data with small-scale local climatic data has found great interest. In this paper, in order to downscale minimum and maximum temperatures and precipitation predictands, the performance of three statistical downscaling techniques including Long Ashton Research Station-Weather Generator (LARS-WG), statistical downscaling model (SDSM), and artificial neural network (ANN) was compared based on Intergovernmental Panel on Climate Change-Fifth Assessment Report (IPCC-AR5) in northwest Iran. For this purpose, a nonparametric test named Mann-Whitney test, Spearman correlation coefficient, and the root mean square error (RMSE) were utilized to assess the efficiency of downscaling models. To scrutinize the climate change impacts, periods of 1961–1990 and 1991–2005 were considered as the baseline and verification periods, respectively. The findings revealed the superior performance of the ANN model for minimum and maximum temperatures, while for precipitation predictand, the SDSM represented the best performance among the models. Simulation results for future temperature indicated an ascending trend as 0.1–1.3 °C, 0.3–1.7 °C, and 0.5–2.1 °C for LARS-WG, SDSM, and ANN techniques, respectively. On the other hand, simulation outputs for the precipitation indicated a descending trend of 10–30% in future precipitation of the region according to downscaling models under Representative Concentration Pathway 8.5 (RCP8.5) pessimistic scenario of Hadley Center Coupled Model version 3 (HadCM3) GCM model. [ABSTRACT FROM AUTHOR]

The aim of this paper is to investigate the ability of artificial neural network (ANN) models for post-processing the monthly precipitation forecasts under North American multi-model ensemble (NMME) project and proposing a new multi-model ensemble neural network (MME-NN) model. Monthly precipitation hindcasts of eight models from NMME project are considered in this study. Multi-layer perceptron neural networks are employed for post-processing the output of the models in comparison with PERSIANN-CDR climatology data. Also, utilizing a multi-criteria decision-making approach, NMME models are ranked for each month. The study is implemented over Iran and detailed discussions are provided. The results indicate that the skill of NMME models is different for each month and for each region of the country. Also, it is shown that the neural network outperforms all NMME models for all months. By using the ranking of the models, for each month, the NMME models are ordered based on their skill and a monthly rank is devoted for each model. [ABSTRACT FROM AUTHOR]

Floods are considered as natural hazards. The parameter estimation of hydrologic methods is time-consuming in the flood routing of multiple inflows river systems. This paper presents the application of gene expression programming (GEP) and artificial neural network (ANN) as alternative approaches to predict the outflow hydrograph in downstream of multiple inflows systems. GEP and ANN models were compared with the equivalent Muskingum inflow model. The Gharesoo River Basin as a multiple inflows river system was applied for the calibration and verification phases of these models. The GEP obtained the formula as a function of the inflow branches based on the fitting data for simulating the outflow hydrograph of the multiple inflows system. GEP and ANN models investigated inflow hydrographs at different time steps. The obtained outflow hydrograph by the GEP model indicated an excellent performance compared with ANN and equivalent Muskingum inflow models in the case involving multiple inflows system. [ABSTRACT FROM AUTHOR]

The present paper reports a study, where growing hierarchical self-organising map (GHSOM) has been applied to achieve a visual cluster analysis to the Indian rainfall dataset consisting of 142 years of Indian rainfall data so that the yearly rainfall can be segregated into small groups to visualise the pattern of clustering behaviour of yearly rainfall due to changes in monthly rainfall for each year. Also, through support vector machine (SVM), it has been observed that generation of clusters impacts positively on the prediction of the Indian summer monsoon rainfall. Results have been presented through statistical and graphical analyses. [ABSTRACT FROM AUTHOR]

This paper aims to introduce an approach for multi-hour forecasting (915 h ahead) of hourly global horizontal solar radiation time series and forecasting of a small-scale solar radiation database (30- and 1-s scales) for a period of 1 day (47,000 s ahead) using commonly and available measured meteorological solar radiation. Three methods are considered in this study. First, autoregressive-moving-average (ARMA) model is used to predict future values of the global solar radiation time series. However, because of the non-stationarity of solar radiation time series, a phase of detrending is needed to stationarize the irradiation data; a 6-degree polynomial model is found to be the most stationary one. Secondly, due to the nonlinearity presented in solar radiation time series, a nonlinear autoregressive (NAR) neural network model is used for prediction purposes. Taking into account the advantages of both models, the goodness of ARMA for linear problems and NAR for nonlinear problems, a hybrid method combining ARMA and NAR is introduced to produce better results. The validation process for the site of Ghardaia in Algaria shows that the hybrid model gives a normalized root mean square error (NRMSE) equals to 0.2034 compared to a NRMSE equal to 0.2634 for NAR model and 0.3241 for ARMA model. [ABSTRACT FROM AUTHOR]

Precipitation prediction is crucial for various sectors, including agriculture, water resource management, and transportation. This paper addresses the gap in rainfall prediction studies that seldom separate spatial and temporal scales. The spatial distribution of rainfall in Zhejiang Province in 2015 is analyzed using multiple interpolation methods: Inverse Distance to a Power (IDP), Kriging Interpolation, Minimum Curvature Interpolation (MC), Modified Shepard’s Method (MSM), Nearest Neighbour Interpolation (NeaN), and Natural Neighbor(NN). Additionally, daily and monthly rainfall are simulated using a 30-year historical precipitation dataset (1990-2020) of an observation station from Zhejiang Province, employing Radial Basis Function Network (RBFN) and Back Propagation (BP) neural networks.Our results confirm the accuracy of these methods in both simulation and interpolation. The benefits of Artificial Neural Networks (ANNs) over traditional interpolation techniques in predicting rainfall are underscored, while the challenges ANNs face, such as nonlinear rainfall patterns, training time, computational complexity, and the configuration of meteorological stations are also acknowledged. Notably, RBFN outperformed the BP model in simulating rainfall, especially for longer forecast periods.In conclusion, the potential applications and future directions of ANNs in rainfall prediction are discussed, their utility across different spatial and temporal scales are emphasized. By comparing ANNs with classical interpolation methods, their respective strengths are highlighted, providing scientific insights for future precipitation forecasting at provincial administrative levels. [ABSTRACT FROM AUTHOR]

The target of the current paper was to examine the performance of three Markovian and seasonal based artificial neural network (ANN) models for one-step ahead and three-step ahead prediction of monthly precipitation which is the most important parameter of any hydrological study. The models proposed here are feed forward neural network (FFNN, as a classic ANN-based models), Wavelet-ANN (WANN, as a hybrid model), and Emotional-ANN (EANN, as a modern generation of ANN-based models). The models were used to precipitation prediction of seven stations located in the Northern Cyprus. Two scenarios were examined each having specific inputs set. The scenario 1 was developed for predicting each station's precipitation through its own data at previous time steps, while in scenario 2, the central station's data were also imposed into the models in addition to each station's data, as exogenous inputs. The obtained results showed the better performance of the EANN model in comparison with other models (FFNN and WANN) especially in three-step ahead prediction. The superiorities of the EANN model over other models are due to its ability in dealing with error magnification in multi-step ahead prediction. Also, the results indicated that the performance of the scenario 2 was better than scenario 1, showing improvement of modeling efficiency up to 17% and 26% in calibration and verification steps, respectively. [ABSTRACT FROM AUTHOR]

High resolution and global coverage of the satellite-based precipitation data have been found useful in many climate and hydrological studies, particularly in limited and nongauged areas. Due to systematic and nonsystematic factors, there are always deviations between ground-based and satellite-based precipitation. Among many satellite-based precipitation products, the PERSIANN (Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks) family products attracted more attention. In this paper, we evaluated the PERSIANN family of monthly, seasonal, and annual precipitation data over Fars Province, Iran, at four different spatial scales, namely point (station), pixel, regional, and provincial during the period of 2003–2015 using 132 rain gauge data as baseline. The performance of the products was first evaluated by calculating some statistical metrics. The result shows that in all spatial and temporal scales, the three products mirror the precipitation pattern and underestimate the precipitation in the province. The PERSIANN-CDR outperforms the other two products and is the superior product. The performance of PERSIANNN products is the best at the provincial scale followed by regional, pixel, and point scales. The performance of PERSIANN family products is also evaluated by the quantile–quantile plot from which a set of equations is proposed to accurately predict precipitation in Fars Province from the PERSIANN-CDR precipitation data. The proposed equations are verified by 2-year precipitation data that were not used in the assessment of the products. These equations provide highly accurate precipitation data from PERSIANN-CDR data particularly in nongauged sites at various spatiotemporal scales for application such as in statistical hydrology and hydroclimate-related projects. [ABSTRACT FROM AUTHOR]

In this paper, artificial neural network (ANN) was used for statistically downscale the outputs of general circulation models (GCMs) to assess future changes of precipitation and mean temperature in Tabriz synoptic station at north-west of Iran. Since one of the significant subjects in statistical downscaling of GCMs is to select the most dominant large scale climate variables (predictors) among huge number of potential predictors, the predictors screening methods including decision tree, mutual information (MI) and correlation coefficient (CC) were used to statistically downscale mean monthly precipitation and temperature. Three GCMs were used, including Can-ESM2 and BNU-ESM from IPCC AR5 models and CGCM3 from IPCC AR4 models. The results of downscaling in the base period (1951–2000) indicated that among feature extraction methods decision tree had superiority to MI and CC techniques. Therefore, the future projection of precipitation and mean temperature during 2020–2060 was implemented using ANN-based simulation according to the most efficient downscaling model (i.e., decision tree-based calibration). Different results according to different GCMs and scenarios were obtained for precipitation projection. In this way, the Can-ESM2 model under RCP8.5 showed 29.78% decrease in annual precipitation and CGCM3 model under B1 indicated 1.06% increase of annual precipitation. Temperature projection outcomes denoted that annual mean temperature will increase over the region and the most increase in mean temperature was determined by BNU-ESM model under RCP8.5. [ABSTRACT FROM AUTHOR]

In this paper, an improved clustered adaptive neuro-fuzzy inference system (ANFIS) to forecast an hour-ahead solar radiation data for 915 h is introduced. First, we have classified the history data of solar radiation time series to decrease the input sample size using clustering methods. Three methods are used, namely, fuzzy c-means (FCM), subtractive clustering, and grid partitioning. These methods allow classifying the input data into groups; each group has similar properties that help to understand the correlation between the data and by consequence simplify the forecasting process. Second, we designed an ANFIS structure that takes both advantages of fuzzy theory to describe the uncertain phenomena of the data and artificial neural network algorithm, which has a self-learning ability. Finally, by combining clustered data and ANFIS model, an hour-ahead forecasting is achieved, and it was validated using measured data. The advantage of the proposed method is that provides the ability to use implicitly the information associated with the forecasting problem, without a priori knowledge of the relationships between the different variables solar radiation. The comparison results show that the ANFIS with FCM clustering model gives the best results with RMSE equals to 112 W/m2 and high values of FS. [ABSTRACT FROM AUTHOR]

Using climate models with high performance to predict the future climate changes can increase the reliability of results. In this paper, six kinds of global climate models that selected from the Coupled Model Intercomparison Project Phase 5 (CMIP5) under Representative Concentration Path (RCP) 4.5 scenarios were compared to the measured data during baseline period (1960-2000) and evaluate the simulation performance on precipitation. Since the results of single climate models are often biased and highly uncertain, we examine the back propagation (BP) neural network and arithmetic mean method in assembling the precipitation of multi models. The delta method was used to calibrate the result of single model and multimodel ensembles by arithmetic mean method (MME-AM) during the validation period (2001-2010) and the predicting period (2011-2100). We then use the single models and multimodel ensembles to predict the future precipitation process and spatial distribution. The result shows that BNU-ESM model has the highest simulation effect among all the single models. The multimodel assembled by BP neural network (MME-BP) has a good simulation performance on the annual average precipitation process and the deterministic coefficient during the validation period is 0.814. The simulation capability on spatial distribution of precipitation is: calibrated MME-AM > MME-BP > calibrated BNU-ESM. The future precipitation predicted by all models tends to increase as the time period increases. The order of average increase amplitude of each season is: winter > spring > summer > autumn. These findings can provide useful information for decision makers to make climate-related disaster mitigation plans. [ABSTRACT FROM AUTHOR]

Accurate estimation of reference evapotranspiration (ET0) is essential for the computation of crop water requirements, irrigation scheduling, and water resources management. In this context, having a battery of alternative local calibrated ET0 estimation methods is of great interest for any irrigation advisory service. The development of irrigation advisory services will be a major breakthrough for West African agriculture. In the case of many West African countries, the high number of meteorological inputs required by the Penman-Monteith equation has been indicated as constraining. The present paper investigates for the first time in Ghana, the estimation ability of artificial intelligence-based models (Artificial Neural Networks (ANNs) and Gene Expression Programing (GEPs)), and ancillary/external approaches for modeling reference evapotranspiration (ET0) using limited weather data. According to the results of this study, GEPs have emerged as a very interesting alternative for ET0 estimation at all the locations of Ghana which have been evaluated in this study under different scenarios of meteorological data availability. The adoption of ancillary/external approaches has been also successful, moreover in the southern locations. The interesting results obtained in this study using GEPs and some ancillary approaches could be a reference for future studies about ET0 estimation in West Africa. [ABSTRACT FROM AUTHOR]

The right design and the high efficiency of solar energy systems require accurate information on the availability of solar radiation. Due to the cost of purchase and maintenance of the radiometers, these data are not readily available. Therefore, there is a need to develop alternative ways of generating such data. Artificial neural networks (ANNs) are excellent and effective tools for learning, pinpointing or generalising data regularities, as they have the ability to model nonlinear functions; they can also cope with complex 'noisy' data. The main objective of this paper is to show how to reach an optimal model of ANNs for applying in prediction of solar radiation. The measured data of the year 2007 in Ghardaïa city (Algeria) are used to demonstrate the optimisation methodology. The performance evaluation and the comparison of results of ANN models with measured data are made on the basis of mean absolute percentage error (MAPE). It is found that MAPE in the ANN optimal model reaches 1.17 %. Also, this model yields a root mean square error (RMSE) of 14.06 % and an MBE of 0.12. The accuracy of the outputs exceeded 97 % and reached up 99.29 %. Results obtained indicate that the optimisation strategy satisfies practical requirements. It can successfully be generalised for any location in the world and be used in other fields than solar radiation estimation. [ABSTRACT FROM AUTHOR]

In this paper, 3-layers MLP (Multi-Layers Perceptron) Artificial Neural Network (ANN) models have been developed and tested for predicting hourly and daily diffuse solar fractions at Fez city in Morocco. In parallel, some empirical models were tested. Three years of data (2009-2011) have been used for establishing the parameters of all tested models and 1 year (2012) to test their prediction performances. To select the best ANN (3-layers MLP) architecture, we have conducted several tests by using different combinations of inputs and by varying the number of neurons in the hidden layer. The output is only the diffuse solar fraction. The performances of each model were assessed on the basis of four statistic characteristics: mean absolute error (MAE), relative mean bias error (RMBE), relative root mean square error (RRMSE) and the degree of agreement (DA). Additionally, the coefficient of correlation ( R) is used to test the linear regression between predicted and observed data. The results indicate that the ANN model is more suitable for predicting diffuse solar fraction than the empirical tested models at Fez city in Morocco. [ABSTRACT FROM AUTHOR]

Sunshine duration data are desirable for calculating daily solar radiation ( R) and subsequent reference evapotranspiration (ET) using the Penman-Monteith (PM) method. In the absence of measured R data, the Ångström equation has been recommended by the Food and Agriculture Organization (FAO) of the United Nations. This equation requires actual sunshine duration that is not commonly observed at many weather stations. This paper examines the potential for the use of artificial neural networks (ANNs) to estimate sunshine duration based on air temperature and humidity data under arid environment. This is important because these data are commonly available parameters. The impact of the estimated sunshine duration on estimation of R and ET was also conducted. The four weather stations selected for this study are located in Sistan and Baluchestan Province (southeast of Iran). The study demonstrated that modelling of sunshine duration through the use of ANN technique made acceptable estimates. Models were compared using the determination coefficient ( R), the root mean square error (RMSE) and the mean bias error (MBE). Average R, RMSE and MBE for the comparison between measured and estimated sunshine duration were calculated resulting 0.81, 6.3 % and 0.1 %, respectively. Our analyses also demonstrate that the difference between the measured and estimated sunshine duration has less effect on the estimated R and ET by using Ångström and FAO-PM equations, respectively. [ABSTRACT FROM AUTHOR]

This paper aimed to investigate the potential of using spectral and textural features in the Meteosat Second Generation-Spinning Enhanced Visible and Infrared Imager (MSG-SEVIRI) data for developing techniques capable of classifying convective and stratiform rain areas in the satellite image. Two different classification methods were introduced that use the brightness temperature (BT) T and brightness temperature differences T- T T- T T- T T- T T- T T- T, and T- T as spectral parameters, along with textural parameters derived from the thermal infrared MSG-SEVIRI channel to discriminate between convective and stratiform rainy clouds. The first is an algorithm based on the probability of convective rainfall (PCR) for each pixel of the MSG satellite data and the second is an artificial neural network multilayer perceptron (MLP) model that relies on the correlation of satellite data with convective and stratiform rain. Both schemes were trained using as reference convective/stratiform classification data from 88 stations in Greece for 20 rainy days with high convective activity and evaluated against an independent dataset of gauge data for six rainy days. Two PCR and two MLP algorithms were constructed based on the previous results: the PCR1 and MLP1 algorithms that use only spectral measures and the PCR2 and MLP2 algorithms based on both spectral and textural measures. It was found that the introduction of textural parameters as additional information to a technique (PCR2 and MLP2 ) works in improving the delineation between convective and stratiform rainy pixels compared to the techniques based only on spectral information (PCR1 and MLP1). The comparison of the two schemes revealed a clear outperformance of the MLP techniques over the PCR techniques. Best skill is provided by MLP2 (POD = 74.5 %, FAR = 44.3 %, POFD = 22.5 %, CSI = 0.47, ETS = 0.31, bias = 1.34) followed by MLP1 and the two PCR techniques. All algorithms overestimate the convective rain occurrences observed by the rain gauge network. These findings showed that the combined use of spectral and textural features in the MSG-SEVIRI imagery can be beneficial for the classification of convective and stratiform precipitating clouds. [ABSTRACT FROM AUTHOR]

The prediction of meteorological time series plays very important role in several fields. In this paper, an application of least squares support vector machine (LS-SVM) for short-term prediction of meteorological time series (e.g. solar irradiation, air temperature, relative humidity, wind speed, wind direction and pressure) is presented. In order to check the generalization capability of the LS-SVM approach, a K-fold cross-validation and Kolmogorov-Smirnov test have been carried out. A comparison between LS-SVM and different artificial neural network (ANN) architectures (recurrent neural network, multi-layered perceptron, radial basis function and probabilistic neural network) is presented and discussed. The comparison showed that the LS-SVM produced significantly better results than ANN architectures. It also indicates that LS-SVM provides promising results for short-term prediction of meteorological data. [ABSTRACT FROM AUTHOR]

This paper designs a comprehensive approach based on the engineering machine/ system concept, to model, analyze, and assess the level of carbon dioxide (CO) exchange between the atmosphere and terrestrial ecosystems, which is an important factor in understanding changes in global climate. The focus of this article is on spatial patterns and on the correlation between levels of CO fluxes and a variety of influencing factors in eco-environments. The engineering/machine concept used is a system protocol that includes the sequential activities of design, test, observe, and model. This concept is applied to explicitly include various influencing factors and interactions associated with CO fluxes. To formulate effective models of a large and complex climate system, this article introduces a modeling technique that will be referred to as stochastic filtering analysis of variance (SF-ANOVA). The CO flux data observed from some sites of AmeriFlux are used to illustrate and validate the analysis, prediction, and globalization capabilities of the proposed engineering approach and the SF-ANOVA technique. The SF-ANOVA modeling approach was compared to stepwise regression, ridge regression, and neural networks. The comparison indicated that the proposed approach is a valid and effective tool with similar accuracy and less complexity than the other procedures. [ABSTRACT FROM AUTHOR]

Climatic variables are frequently used as weighting factors to indicate the degree of clearness for interpreting sky patterns. However, such important parameters are not always widely available and their criteria to define a sky condition are not clear-cut. In addition, certain variables may be more effective than the others in terms of sky identification. This paper studies the capability of various daylight parameters, namely zenith luminance, global, direct-beam and sky-diffuse illuminance, and solar altitude for categorizing the 15 International Commission on Illumination (CIE) standard skies. A new form of artificial neural networks called probabilistic neural network (PNN) which is a powerful technique for pattern recognition was used for the analysis. The findings suggested that the PNN is an appropriate tool when a number of climatic parameters of various criteria for differentiating sky standards are employed, and the ratio of zenith luminance to diffuse illuminance ( L/ D) and solar altitude ( α) are respectively the most and the least significant input parameters for discriminating between the 15 CIE skies. [ABSTRACT FROM AUTHOR]

This paper examines the potential for the use of artificial neural networks (ANNs) to estimate the reference crop evapotranspiration (ET0) based on air temperature data under humid subtropical conditions on the southern coast of the Caspian Sea situated in the north of Iran. The input variables for the networks were the maximum and minimum air temperature and extraterrestrial radiation. The temperature data were obtained from eight meteorological stations with a range of latitude, longitude, and elevation throughout the study area. A comparison of the estimates provided by the ANNs and by Hargreaves equation was also conducted. The FAO-56 Penman–Monteith model was used as a reference model for assessing the performance of the two approaches. The results of this study showed that ANNs using air temperature data successfully estimated the daily ET0 and that the ANNs with an R2 of 0.95 and a root mean square error (RMSE) of 0.41 mm day−1 simulated ET0 better than the Hargreaves equation, which had an R2 of 0.91 and a RMSE of 0.51 mm day−1. [ABSTRACT FROM AUTHOR]

Since the middle of the twentieth century, artificial intelligence (AI) models have been used widely in engineering and science problems. Water resource variable modeling and prediction are the most challenging issues in water engineering. Artificial neural network (ANN) is a common approach used to tackle this problem by using viable and efficient models. Numerous ANN models have been successfully developed to achieve more accurate results. In the current review, different ANN models in water resource applications and hydrological variable predictions are reviewed and outlined. In addition, recent hybrid models and their structures, input preprocessing, and optimization techniques are discussed and the results are compared with similar previous studies. Moreover, to achieve a comprehensive view of the literature, many articles that applied ANN models together with other techniques are included. Consequently, coupling procedure, model evaluation, and performance comparison of hybrid models with conventional ANN models are assessed, as well as, taxonomy and hybrid ANN models structures. Finally, current challenges and recommendations for future researches are indicated and new hybrid approaches are proposed. [ABSTRACT FROM AUTHOR]

This study aimed to assess monthly average daily global solar irradiation ( H g m) estimates in Southeastern Brazil from empirical models and artificial intelligence (AI) techniques using extreme air temperatures (maximum—Tx and minimum—Tn) and extraterrestrial solar irradiation ( H 0 m ) and evaluate the geographic and climate factors in the performance of the models or the AI techniques. The H g m , Tx and Tn data series from 95 automatic weather stations of the Brazilian National Institute of Meteorology (INMET) were used. The Hargreaves-Samani (HS) and Bristow-Campbell (BC) models were fitted, and AI techniques (artificial neural network—ANN and support vector machines—SVM) were trained, and assessed according to local climate conditions. Three input schemes in the AI techniques were assessed: (i) H 0 m , Tx, and Tn; (ii) H 0 m and air temperature amplitude (ΔT); and (iii) H 0 m , Tx, Tn, and ΔT. The coefficients of the HS (0.134 ≤ kr ≤ 0.262) and BC (0.539 ≤ β0 ≤ 0.796; 0.004 ≤ β1 ≤ 0.843; 0.35 ≤ β2 ≤ 2.60) models and their performance were influenced by climate, continentality/maritime effects, altitude and weather systems. The performance of the HS model was inferior to that of BC model and the SVM and ANN techniques. Overall, the SVN and ANN techniques performed better than the BC method. However, BC had similar or superior precision and accuracy to SVM and ANN for some climatic conditions and combinations of input variables. The quality of the H g m series, the combination of input variables, and climate influenced the estimates of the SVM and ANN techniques. [ABSTRACT FROM AUTHOR]

This work proposes a probabilistic forecast of seasonal precipitation in the basins of the Limay, Neuquén, and Negro rivers in the north of Argentine Patagonia. The Comahue region is particularly important because part of the country's hydroelectric energy is generated there. The amount of winter precipitation modulates the flow of rivers, and therefore, prior knowledge of possible precipitation thresholds is very useful for decision-makers. Ensembles made up of statistical models that explain more than 50% of the precipitation were used and were generated with multiple techniques such as linear regression, generalized additive models, support vector regression, and artificial neural networks. The result showed that the forecasts are better in Limay and Neuquén river basins in winter than in Negro River basin. Brier Skill Score values indicate that the probabilistic forecast is better than the climatology in winter, in Neuquén and Limay basins for below and above normal categories. [ABSTRACT FROM AUTHOR]

Crop yield forecasting is the art of predicting yield before harvest and is crucial for sound planning and policy making at various levels. Rapeseed and mustard (R&M) is the predominant oilseed crop in Assam and is very much sensitive to climatic factors compared to other crops grown during rabi season. In the context of recent climate variability observed in the Brahmaputra valley, development of efficient yield forecasting models for R&M based on weather parameters would certainly help the policy makers in formulating appropriate policies and decision-making. The present study was undertaken to develop R&M pre-harvest yield forecast models at F1 (flowering), F2 (siliqua formation) and F3 (siliqua development & maturity) stages for 15 districts of Assam using stepwise multiple linear regression (SMLR) and artificial neural network (ANN) techniques by analysing the yield and weather data for 27 years and to investigate the impact of weather factors in determining the yield of R&M. The results were enumerated based on coefficient of determination (R2), root mean square error (RMSE) and normalized root mean square error (RMSEn) values. ANN models were found to have greater R2 values over SMLR at all the three stages of forecast except in 2–3 districts. Similarly, profound improvement was observed in RMSE and RMSEn values of the forecast models using ANN technique. Analysing the percent error between observed and forecasted yield, ANN weather-based models were found to give more accurate pre-harvest yield prediction of R&M in the Brahmaputra valley of Assam. Temperature and relative humidity were found to be the most significant parameters in affecting R&M yield in most of the districts during all the three stages of forecast. [ABSTRACT FROM AUTHOR]

Wind profiler radars are capable of measuring three-dimensional wind profiles at various altitudes of the atmosphere, at very high temporal and spatial resolution. The Advanced Centre for Atmospheric Radar Research (ACARR), located at Cochin University of Science and Technology (CUSAT), operates the world's first 205 MHz stratosphere-troposphere wind profiler radar which provides three-dimensional wind profiles for an altitude range of 315 m to 20 km. During non-rainy condition, the radar Doppler power spectrum bears the signature of ambient air motion whereas during rainy conditions, it contains signatures of both ambient air motion and fall velocity of rain droplets. The classification of Doppler power spectra for rainy (Precipitation) and non-rainy (Clear) conditions is necessary as wind profile retrieval from the former needs careful separation of ambient air motion from fall velocity of droplets. A manual classification of the power spectrum is cumbersome, time-consuming, and therefore not practical due to the vast database. This work intends to automate Doppler power spectra classification using the deep learning Convolutional Neural Network (CNN). The proposed Convolutional Neural Network model gives a k-fold validation accuracy of 99.77% and testing accuracy of 99.60% for power spectra classification. The performance of CNN is compared against other popular machine learning classifiers such as Support Vector Machine, Decision Tree, K Nearest Neighbour and Naive Bayes. The performance comparison results show that the proposed CNN outperforms other models in radar Doppler power spectra classification. [ABSTRACT FROM AUTHOR]

The skill of the diverse-based precipitation products is investigated in comparison with the observed-derived HYBAM and with GoAmazon and TRMM-LBA field campaigns data. The performance of eight remote sensing-based datasets (CHIRPS, MSWEP, TRMM, CMORPH, IMERG, PERSIANN-CDR, PERSIANN-CCS-CDR, and PERSIANN-CCS) is evaluated from 1998 to 2009 considering different timescales (diurnal, intraseasonal, and seasonal) for the Amazon Basin (AB). To compare the rainfall products, we applied cluster analysis, the Seasonality Index, the Kling-Gupta Efficiency metric, categorical indices, spectral analysis, and composing technique. Overall, the databases poorly represent the precipitation in the northwest of the AB (NWAB), a region with a lack of in situ measurements. CHIRPS underestimates NWAB precipitation at the seasonal scale, while at the intraseasonal timescale most databases do not adequately represent the precipitation anomalies in the NWAB, except for MSWEP. The PERSIANN-CCS-CDR product overestimates precipitation and extreme rainfall, while CMORPH overestimates the number of no rain events. At the diurnal timescale, most databases overestimate nighttime precipitation and underestimate it in the afternoon. Although there is no single database more suitable for the AB at all timescales, CHIRPS, MSWEP, and PERSIANN-CDR are the most accurate databases at seasonal timescale, while at the intraseasonal scale MSWEP would be the most appropriate. The estimates that combine reanalysis data, surface measurements, and infrared and microwave satellite data through more sophisticated techniques, such as artificial neural networks, can improve the representation of the rainfall in the AB, especially at the diurnal timescale. [ABSTRACT FROM AUTHOR]

Estimation of solar radiation can play a key role in environmental management as well as other fields of energy, agriculture, and hydrological and ecological modeling. In some areas, there are not enough solar radiation data due to a lack of pyranometer or its breakdown from time to time. Hence, having an estimation set at hand to estimate solar radiation based on other climatic variables is crucial. In order to develop an estimation tool, two models are applied simultaneously as a new hybrid model for estimation of monthly global solar radiation for three regions in Iran as case studies of this research work: (1): an artificial neural network (ANN) optimized with Harris hawk's optimization (HHO) algorithm (ANNHHO) and (2) phase space reconstruction (PSR) integrated with the ANNHHO hybrid model (PSR-ANNHHO). Monthly meteorological data of minimum temperature (Tmin), maximum temperature (Tmax), mean temperature (Tmean), sunshine hours (SH), wind speed (U2), and relative humidity (RH) of 37 years (1985–2018) from three regions in Iran with different climate types were employed for training and testing the developed models. To select appropriate input variables for the models, a relief algorithm was applied. The performance of the new hybrid models is compared with the stand-alone ANN model. The obtained results revealed that although all the intelligent models perform satisfactorily, the hybrid PSR-ANNHHO model outperforms the hybrid ANNHHO and stand-alone ANN models in all regions. The hybrid ANN-HHO model follows the PSR-ANNHHO model as the second most accurate model. [ABSTRACT FROM AUTHOR]

The development of the stage-discharge relationship is a fundamental issue in hydrological modeling. Due to the complexity of the stage-discharge relationship, discharge prediction plays an essential role in planning and water resource management. The present study was conducted to model discharge at the Gaula Barrage site in the Uttarakhand state of India. The study evaluated adaptive neuro-fuzzy inference system (ANFIS)-, artificial neural network (ANN)-, and wavelet-based artificial neural system (WANN)-based models to estimate the discharge. The daily data of 12 years (2007–2018) were used to train and test the models. The gamma test was used to identify the best model for discharge prediction. The input data having a stage with 1-day lag and discharge with 1- and 2-day lags and current-day discharge as output was used for discharge modeling. In the case of ANN models, the back propagation algorithm and hyperbolic tangent sigmoid activation function were used. WANN used Haar, a trous-based wavelet function. In ANFIS models, triangular, psig, generalized bell, and Gaussian membership functions were used to train and test the models. The models were evaluated qualitatively and quantitatively using correlation coefficient, and root means square error, Willmott index, and efficiency coefficient. It was found that the ANFIS model performed better than ANN- and WANN-based models for discharge prediction at the Gaula Barrage. [ABSTRACT FROM AUTHOR]

Soil erosion is one of the most serious problems that cause environmental degradation and loss of agricultural productivity in semiarid regions. To investigate the applicability of satellite-estimated data for assessing rainfall and erosivity in the eastern portion of the Brazilian semiarid region, data from the Climate Hazards Group Infrared Precipitation with Stations (CHIRPS) and Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks – Climate Data Record (PERSIANN-CDR) were evaluated and compared with measured rainfall data. Rainfall dynamics simultaneously considering precipitation temporal series and geomorphology were analyzed. To evaluate satellite-estimated rainfall data and estimated erosivity, the correlation coefficient (R), relative bias (BIAS), average error (ME), and root mean square error (RMSE) were used based on rain gauge-measured data. The results showed that both estimated rainfall datasets perform well when compared to the rain gauge-measured rainfall data. CHIRPS-estimated rainfall data were slightly more accurate on a monthly scale and captured the spatial variability of the rainfall when compared to the other dataset. Finally, the study pointed to CHIRPS- and PERSIANN-CDR-estimated rainfall data as good alternatives to estimate rainfall and erosivity in a Brazilian semiarid region. [ABSTRACT FROM AUTHOR]

Prediction of local scale frost events can be helpful for farmers to minimize crop loss due to frost damage. This study aims to detect a temporal trend in the occurrence of frost events and develop frost prediction models using multivariate statistical techniques like logistic regression, artificial neural network model, and thumb rules for two diverse locations of India (Palampur and Ludhiana). In these statistical models, eight daily meteorological parameters viz., maximum temperature (Tmax), minimum temperature (Tmin), wind speed, precipitation, sunshine duration, cumulative pan evaporation, morning relative humidity (RH1), and afternoon relative humidity (RH2) 1 to 5 days preceding the frost events for the period of 2004–2016 and 1982–2013 at Palampur and Ludhiana, respectively were used. Principal Component Analysis was performed to select the weather parameter that has maximum effect on the occurrence of frost event. Ten different skill scores like accuracy, bias, and probability of false detection were used to evaluate the accuracy of frost prediction models. The Mann–Kendall trend test showed a significant increasing annual trend in the number of frost events at Ludhiana, with a remarkable increase in December. The results also showed that lower afternoon relative humidity 1-day preceding the frost event at Palampur and calm wind and lower evaporation 1-day preceding at Ludhiana augmented the occurrence of frost events. Among the techniques for developing frost prediction models, the logistic regression model performed better over artificial neural network and thumb rule-based models. The logistic regression model performed better for the plain region (Ludhiana) than for the hilly area (Palampur). The developed models are most suitable for predicting the radiation frost. [ABSTRACT FROM AUTHOR]

Finding an accurate computational method for estimating pan evaporation (EPm) can be useful in the application of these methods for the development of sustainable agricultural systems and water resources management. In the present study, the proposed hybrid method called multiple model-support vector machine (MM-SVM) with the aim of showing the increasing, decreasing, and constant accuracy behavior of this hybrid model and improving the results of estimating EP compared to the two models ANN and SVM on a monthly scale of EPm in four meteorological stations (Ardabil, Khalkhal, Manjil (from Iran), and Grand Island (from the USA)) located in semi-arid regions, using the output of artificial intelligence (AI) models (i.e., artificial neural network (ANN) and support vector machine (SVM)), was evaluated. The results of intelligent models using several statistical indices (i.e., root mean square error (RMSE), mean absolute error value (MAE), Kling-Gupta (KGE), and coefficient of determination (R2)) and with the help of case visual indicators were compared. According to the results of evaluation indicators in the test phase, MM-SVM-6, ANN-5, MM-SVM-3, and MM-SVM-7 with RMSE = 1.088, 0.761, 0.829, and 0.134 mm/day; MAE = 0.79, 0.54, 0.589, and 0.105 mm/day; KGE = 0.819, 0.903, 0.972, and 0.981; and R2 = 0.939, 0.962, 0.967, and 0.996 and with four input variables were introduced as the best models in Ardabil, Khalkhal, Manjil, and Grand Island stations, respectively. The proposed hybrid model (MM-SVM) was able to use its multi-model strategy with inputs estimated by independent models, its power to estimate EPm in scenarios where there is a high correlation between its components with EPm, in a feasible state Accept to show. So that the incremental, constant, and decreasing modes in EPm estimation accuracy by this hybrid model under the semi-arid climatic conditions of the studied areas were quite clear. Therefore, the results of the proposed and superior models in the present study can help local stakeholders in discussing water resources management. [ABSTRACT FROM AUTHOR]

Changes in precipitation as a result of climate change are becoming a widespread issue all around the world. A lack of rainfall causes a meteorological drought. The short-term Standardized Precipitation Index (SPI-6) index was used to estimate meteorological drought conditions in Saudi Arabia's Asir region from 1970 to 2017. Innovative trend analysis (ITA), the Modified Mann–Kendall test (MMK), the Sequential Mann–Kendall test, and Morlet wavelet transformation were used to detect trend and periodicity in meteorological drought conditions in the Asir region. In addition, the meteorological drought conditions were forecasted by integrating Particle Swarm Optimization (PSO) ensemble machine learning algorithm and an artificial neural network (ANN). Droughts of varying severity have become more frequent in Asir, according to the findings. In most stations, ITA and MMK tests have revealed a significant increase in drought. In all stations, the SQMK test revealed a big sudden year-over-year drought trend. With the exception of one station, all stations experienced extreme drought frequency discovered using Morlet Wavelet Transformation over a long period of time (10 years or more) (station 34). The PSO-ANN hybrid learning algorithm predicted SPI-6 values that had a strong correlation with actual SPI-6 values and also had lower error values, indicating that this model performed well. The PSO-ANN model predicts that the Asir region of Saudi Arabia will experience major moderate to extreme drought events in the coming years (2018–2025). The findings of this analysis will assist planners and policymakers in planning for the acquisition of sustainable agriculture in the study area. [ABSTRACT FROM AUTHOR]

Soil temperature is one of the most important glacio-meteorological parameters that play a critical role in glacier energy and mass balance dynamics, surface hydrological processes, and glacier-atmosphere interaction. However, the availability of the data is acutely scarce in the Himalayan glaciated region. In this study, we applied artificial neural network (ANN) models for the prediction of soil temperature of glacial forefield region of the Pindari Glacier (Central Himalaya). Three-layer feed-forward ANN models were developed and tested for estimating multi-depth soil temperatures using concurrent and antecedent air-soil temperature data for one complete annual cycle as inputs for the models. Models with different combinations of input variables were tested, and best sets of variables were selected based on the prediction accuracy. Rigorous statistics were further employed to compare the performances of different models. High concurrence was obtained between ANN-estimated and measured soil and air temperatures as evident by various correlation coefficients and error ranges. In a boarder perspective, our results point toward the applicability of developed ANN models to provide robust soil temperature prediction for the glacial forefield regions of the Central Himalaya. [ABSTRACT FROM AUTHOR]

Solar radiation is an important parameter that affects the atmosphere-earth thermal balance and many water and soil processes such as evapotranspiration and plant growth. The modeling of the daily and monthly solar radiation by Gaussian process regression (GPR) with K-fold cross-validation model has been discussed recently. This study evaluated different neural models such as artificial neural network (ANN), support vector machine (SVM), adaptive network-based fuzzy inference system (ANFIS), and multiple linear regression (MLR) for estimating the global solar radiation (daily and monthly) with K-fold cross-validation method. For the appropriate comparison of the models, the randomized complete block (RCB) design applied in the training and test phases. Also, different data sets were evaluated by K-fold cross-validation in each model. The results showed that radial basis function (RBF) model has the lowest error for estimating the monthly and daily solar radiation. In this study, the result of RBF was compared with the GPR models. The conclusion indicated that RBF methodology can predict solar radiation with higher accuracy relative to the GPR model. The results of yearly solar radiation estimation (2009–2014) showed that the RBF model can estimate solar radiation with the MAPE and RMSE of 5.1% and 0.29, respectively. Also, the coefficient of correlation (R2) between actual and estimated values throughout the year is 98% and can be used by the engineers and other researchers for solar and thermal applications. [ABSTRACT FROM AUTHOR]

The reference evapotranspiration (ET0) estimates is important for water resources and irrigation management. The Penman-Monteith equation is known for its accuracy but requires a high number of climatic parameters that are not always available. Thus, this study aimed to evaluate the performance of machine learning techniques (cubist regression, artificial neural network with Bayesian regularization, support vector machine with linear kernel function) and stepwise multiple linear regression method to estimate daily ET0 with limited weather data in a Brazilian agricultural frontier (MATOPIBA). Climatic data from 2000 to 2016 obtained from 23 weather stations were used. Five data scenarios were evaluated: (i) all variables, (ii) radiation and temperature, (iii) temperature and relative humidity, (iv) wind speed and temperature, and (v) temperature. The results showed that the machine learning methods are robust in estimating ET0, even in the absence of some variables. Among the methods evaluated using only temperature data, the cubist regression showed better performance. When estimating water demand for soybean and maize crops using only temperature, the cubist regression and calibrated Hargreaves-Samani equation showed the smallest errors. [ABSTRACT FROM AUTHOR]

Efficient and reliable forecasting techniques for various climatic conditions are indispensable in agricultural dependent country like India. In this context, rainfall forecasting is one of the most challenging tasks because of the existence of three patterns, viz., temporal, spatial, and non-linear, simultaneously. Space-Time Autoregressive Moving Average (STARMA) model is one of the promising and popular approaches for modelling spatio-temporal time series data. However, the observed features of many space-time rainfall data comprise complex non-linear dynamics and modelling these patterns often go beyond the capability of conventional STARMA model. Moreover, despite the popularity of artificial neural network (ANN) and support vector machine (SVM) for modelling complex non-linear dynamics, they are not capable to deal with spatial patterns. To overcome the problem, a new spatio-temporal hybrid modelling approach has been proposed by integrating STARMA, ANN, and SVM as well. The proposed approach has been empirically illustrated on annual precipitation data of six districts of northern part of West Bengal, India. The study reveals that proposed spatio-temporal hybrid approach has better modelling and forecasting precision over conventional STARMA as well as most widely used Autoregressive Integrated Moving Average (ARIMA) model. [ABSTRACT FROM AUTHOR]

A comparative study between classic linear and intelligent nonlinear time series approaches for short-term maximum wave height forecasting is presented in this study. The applied models to accomplish a use case for onshore measurements from the Mediterranean Sea include ordinary linear regression (LR), autoregressive integrated moving average (ARIMA), artificial neural networks (ANN), and genetic programming (GP). The study also introduces a new evolutionary ensemble model called ensemble GP, which integrates effective models' forecasts through an evolutionary procedure. The results from standalone models showed that both linear and nonlinear models provide the same accuracy for short-term maximum wave height hindcasting on a seasonal scale. The proposed ensemble model can enhance the forecasting accuracy of standalone models markedly. The new model can forecast maximum wave heights with the root mean squared errors less than 5 cm and Nash-Sutcliff efficiency more than 0.97. It is explicit and secures parsimony conditions, thus it is proposed to be used in practice. [ABSTRACT FROM AUTHOR]

The forecasting of lake water level is one of the complex problems in the hydrology field owing to the incorporating with various hydrological and morphological characteristics. In this research, newly hybrid data intelligence (DI) model based on the integration of the Multilayer Perceptron (MLP) and Whale Optimization Algorithm (WOA) is developed for lake water level forecasting. The potential of the proposed hybrid MLP_WOA model is validated against several well-established DI models over the literature including the Cascade-Correlation Neural Network Model (CCNNM), Self-Organizing Map (SOM), Decision Tree Regression (DTR), Random Forest Regression (RFR), and classical MLP. The applied predictive models are examined to forecast the Van Lake water level fluctuation with monthly scale over seven-decade time period (1943–2016). The input variables are abstracted using statistical correlation analysis procedure. The modeling is diagnosed using multiple statistical metrics (i.e., root mean square error (RMSE), mean absolute error (MAE), Nash-Sutcliffe coefficient (NSE), Willmott's Index (WI), Legate and McCabe's Index (LMI), determination coefficient (R2)). In addition, graphical distribution data such as the Taylor diagram, violin plot, and point density are investigated. Results indicated that the MLP_WOA model performed superior prediction results over the comparable models based on forecasting performance. Five-month lead times performed the best results for the prediction procedure. In quantitative terms, the RMSE and MAE are reduced by 29.8% and 33.9%, 48.3% and 52%, 57.6% and 59.7%, 53.9% and 58.3%, and 25.3% and 23.9% using the MLP_WOA model over CCNNM, SOM, DTR, RFR, and MLP models, respectively. In comparison with the literature studies, using longer span of historical data elevated the forecasting accuracy. In summary, MLP_WOA model provided an applicable and simple methodology for Van Lake water level forecasting owing to its simple learning procedure. [ABSTRACT FROM AUTHOR]

Evapotranspiration (ET) is a fundamental element in the hydrological cycle and plays a vital role in simulating hydrological effects of climate change. Accurate estimation of reference evapotranspiration (ETo) plays a significant role in the field of hydrometeorology and agrometeorology. There are many direct or indirect methods that are employed for estimation of ETo. Furthermore, soft computing approaches are also used to solve complex problems of estimation in various disciplines. The application of the soft computing techniques for estimating evapotranspiration has only been limited to few approaches including artificial neural network, fuzzy logic, and genetic algorithm. This study is conducted to apply non-conventional soft computing approaches for estimating ETo which include single decision tree (SDT), tree boost (TB), and decision tree forest (DTF). Monthly meteorological data spanning over 30 years (1987 to 2016) for six different cities located in arid, semi-arid, and humid regions of Pakistan is used for the estimation of ETo in this study. Seventeen input combinations comprising of various climatic parameters were developed to evaluate the impact of different parameters. A total of 306 models were developed using the SDT, TB, and DTF approaches for all six selected cities of Pakistan. The results of the developed models are then compared with ETo calculated by the standard method of modified Penmen equation. The TB modeling technique is found to perform best for various cities located in different climatic zones of Pakistan. Only two climatic parameters, the mean temperature and wind velocity, are found to be sufficient for efficient ETo prediction. To further validate the results, the outcome of the study was applied on some other climatic regions located in the USA, New Zealand, and China. Finally, it is concluded that these non-conventional approaches can also be considered superior alternatives to the available methods of estimating ETo. [ABSTRACT FROM AUTHOR]

A precise forecast of streamflow in intermittent rivers is of major difficulties and challenges in watershed management, particularly in arid and semiarid regions. The present research study introduces an ensemble gene expression programming (EGEP) modeling approach to 1- and 2-day ahead streamflow forecasts that meet both accuracy and simplicity criteria of an applied model. Three main components of the proposed EGEP approach which are capable of producing a parsimonious model include (i) creating a population of suitable solutions using classic genetic programming (GP) instead of a single solution, (ii) combining the solutions throughout gene expression programming, and (iii) parsimony selection based upon trade-off analysis between the complexity and accuracy of the best-evolved solutions at the holdout validation set. The EGEP model was trained and verified using the streamflow measurements from the Shahrchay River lying northwest of Iran. Several statistical indicators were computed for verification of the ensemble models' accuracy with that of classic GP and artificial neural network models developed as the benchmarks. Our results revealed that the EGEP outperforms the benchmarks. It is an explicit, simple, and precise approach and, therefore, worthy to be used in practice. [ABSTRACT FROM AUTHOR]

Evapotranspiration (ET) is a main factor of the hydrologic balance. Estimating precise ET is necessary for managing the water supply in a basin. In this study, daily barley standard evapotranspiration (DBSE) is obtained (1) directly by weighing lysimeter and (2) indirect methods. In the first step, DBSE was obtained by two weighing lysimeters in a semi-arid region (Kooshkak, Iran). In the next step, indirect methods for estimating the DBSE, the Penman–Monteith (PM), and the artificial neural networks (ANNs), including the radial basis functions (RBF), and the multi-layer perceptron (MLP), were utilized. Results showed that DBSE can be successfully calculated in semi-arid region by MLP-ANN, RBF-ANN, and PM methods. The ANN methods were offered as the best method because they need fewer input data and can be easily used for other developed programs that applied for water allocation and therefore solve the conflicts between stakeholders and optimize water usage. Finally, the sensitivity of ANNs to input data was investigated by relating changes in the daily metrological data to the dimensionless scaled sensitivities (DSS) index. Results showed that in the multi-layer perceptron, ETc was more sensitive to sunshine hours and less sensitive to wind speed and the radial basic function has different patterns which are more sensitive to sunshine hours. When the sunshine hours decrease by more than 10%, the standard crop evapotranspiration (ETc) was more sensitive to average humidity and less sensitive to wind speed. [ABSTRACT FROM AUTHOR]

In recent years, satellite data has been used to estimate precipitation with the aim of increasing the accuracy of rainfall spatial distribution especially at ungauged locations. In this research, the satellite data, including visible and infrared reflection data from the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor and observation data, consists of rainfall records (10 years 2005–2015) from three synoptic stations in Semnan province, were used to simulate rainfall using an artificial neural network (ANN) method. The network performance is evaluated through three performance criteria, i.e., correlation coefficient (R), root mean square error (RMSE), and Nash–Sutcliffe (NS). Findings show that using a combination of visible reflection data of band 3 and infrared reelection data of bands 5, 18, and 19 as input data results in better performance compared with other possible combinations. In this model, the values of R, NS, and RMSE for test period data were 0.93, 0.81, and 1.49, respectively. [ABSTRACT FROM AUTHOR]

The Yarlung Zangbo River Basin (YZRB) is the longest plateau river in China and is one of the highest rivers in the world. In the context of climate change, the ecological environment of the YZRB has become increasingly fragile because of its unique location and environment. In this study, four machine learning techniques, multiple linear regression (MLR), artificial neural network (ANN), support vector machine (SVM), and random forest (RF) model, were applied to downscale the daily extreme temperatures (maximum and minimum) at 20 meteorological stations located in and around the YZRB. The performance of these methods was evaluated using four comparison criteria. The best identified model was adopted to simulate future temperatures under two extreme scenarios (the lowest rate emission scenario (RCP2.6) and the highest rate emission scenario (RCP8.5)) from 2016 to 2050 using outputs from the MPI-ESM-LR climate model. The four comparison criteria showed that the RF model yielded the highest efficiency; therefore, this model was chosen to simulate the future temperatures. The results indicate that the extreme temperatures at the 20 stations increase continually under both extreme scenarios. The increases in the maximum temperature at the 20 stations under the two extreme emission scenarios are 0.46 and 0.83 °C, and the increases in the minimum temperature at the 20 stations are 0.30 and 0.68 °C for the period 2016–2050, respectively. [ABSTRACT FROM AUTHOR]

Uncertainties in rainfall have increased in the recent past exacerbating climate risks which are projected to be higher in semiarid environments. This study investigates the associated features of rainfall such as rain onset, cessation, length of the rain season (LRS), and dry spell frequency (DSF) as part of climate risk management in Botswana. Their trends were analysed using Mann-Kendall test statistic and Sen's Slope estimator. The rainfall-evapotranspiration relationships were used in formulating the rain onset and cessation criteria. To understand some of the complexities arising from such uncertainties, artificial neural network (ANN) is used to predict onset and cessation of rain. Results reveal higher coefficients of variation in onset dates as compared to cessation of rain. Pandamatenga experiences the earliest onset on 28th of November while Tsabong the latest on 14th of January. Likewise, earliest cessation is observed at Tshane on 22nd of February and the latest on 30th of March at Shakawe. The shortest LRS of 45 days is registered at Tsabong whereas the northern locations show LRS greater than 100 days. Stations across the country experience strong negative correlation between onset and LRS of − 0.9. DSF shows increasing trends in 50% of the stations but only significant at Mahalapye, Pandamatenga, and Shakawe. Combining the LRS criteria and DSF, Kasane, Pandamatenga, and Shakawe were identified to be suitable for rainfed agriculture in Botswana especially for short to medium maturing cereal varieties. Predictions of onset and cessation indicate the possibility of delayed onset by 2-5 weeks in the next 5 years. Information generated from this study could help Botswana in climate risk management in the context of rainfed farming. [ABSTRACT FROM AUTHOR]

Hydrological impacts of global climate change on regional scale are generally assessed by downscaling large-scale climatic variables, simulated by General Circulation Models (GCMs), to regional, small-scale hydrometeorological variables like precipitation, temperature, etc. In this study, we propose a new statistical downscaling model based on Recurrent Neural Network with Long Short-Term Memory which captures the spatio-temporal dependencies in local rainfall. The previous studies have used several other methods such as linear regression, quantile regression, kernel regression, beta regression, and artificial neural networks. Deep neural networks and recurrent neural networks have been shown to be highly promising in modeling complex and highly non-linear relationships between input and output variables in different domains and hence we investigated their performance in the task of statistical downscaling. We have tested this model on two datasets—one on precipitation in Mahanadi basin in India and the second on precipitation in Campbell River basin in Canada. Our autoencoder coupled long short-term memory recurrent neural network model performs the best compared to other existing methods on both the datasets with respect to temporal cross-correlation, mean squared error, and capturing the extremes. [ABSTRACT FROM AUTHOR]