ARTIFICIAL NEURAL NETWORKS FOR SUBSURFACE DRAINAGE AND SUBIRRIGATION SYSTEMS IN ONTARIO,CANADA1

ABSTRACT: Artificial neural network (ANN) models were developed to simulate fluctuations in midspan water table depths (WTD) given rainfall, potential evapotranspiration, and irrigation inputs on a Brookston clay loam in Woodslee, Ontario, having a dual‐purpose subsurface drainage/subirrigation setup. Water table depths and meteorologic data collected at this site from 1992 to 1994 and from 1996 to 1997 were used to train the ANNs. The ANNs were then used for real‐time control and time series simulations. The lowest root mean squared errors (RMSE) for the various ANNs were 60.6 mm for real‐time control simulation, and 88.4 mm for time‐series simulation of water table depths. It was possible to simulate WTD for the different modes of water table management in one network by incorporating an indicator for switching from one to the other. The ANN simulations were quite good even though the training data sets had irregular measurement intervals. With fewer input parameters and small network structures, ANNs still provided accurate results and required little time for training and execution. ANNs are therefore easier and faster to develop and run than conventional models and can contribute to the proper management of subsurface drainage and subirrigation systems.     

SOIL MOISTURE PREDICTION USING SUPPORT VECTOR MACHINES1

ABSTRACT: Herein, a recently developed methodology, Support Vector Machines (SVMs), is presented and applied to the challenge of soil moisture prediction. Support Vector Machines are derived from statistical learning theory and can be used to predict a quantity forward in time based on training that uses past data, hence providing a statistically sound approach to solving inverse problems. The principal strength of SVMs lies in the fact that they employ Structural Risk Minimization (SRM) instead of Empirical Risk Minimization (ERM). The SVMs formulate a quadratic optimization problem that ensures a global optimum, which makes them superior to traditional learning algorithms such as Artificial Neural Networks (ANNs). The resulting model is sparse and not characterized by the “curse of dimensionality.” Soil moisture distribution and variation is helpful in predicting and understanding various hydrologic processes, including weather changes, energy and moisture fluxes, drought, irrigation scheduling, and rainfall/runoff generation. Soil moisture and meteorological data are used to generate SVM predictions for four and seven days ahead. Predictions show good agreement with actual soil moisture measurements. Results from the SVM modeling are compared with predictions obtained from ANN models and show that SVM models performed better for soil moisture forecasting than ANN models.     

Comparison of ANN,MVR, and SWR models for computing thermal efficiency of a solar still

In this paper, the viability of modeling the instantaneous thermal efficiency (η_ith) of a solar still was determined using meteorological and operational data with an artificial neural network (ANN), multivariate regression (MVR), and stepwise regression (SWR). This study used meteorological and operational variables to hypothesize the effect of solar still performance. In the ANN model, nine variables were used as input parameters: Julian day, ambient temperature, relative humidity, wind speed, solar radiation, feed water temperature, brine water temperature, total dissolved solids of feed water, and total dissolved solids of brine water. The η_ith was represented by one node in the output layer. The same parameters were used in the MVR and SWR models. The advantages and disadvantages were discussed to provide different points of view for the models. The performance evaluation criteria indicated that the ANN model was better than the MVR and SWR models. The mean coefficient of determination for the ANN model was about 13% and14% more accurate than those of the MVR and SWR models, respectively. In addition, the mean root mean square error values of 6.534% and 6.589% for the MVR and SWR models, respectively, were almost double that of the mean values for the ANN model. Although both MVR and SWR models provided similar results, those for the MVR were comparatively better. The relative errors of predicted η_ith values for the ANN model were mostly in the vicinity of ±10%. Consequently, the use of the ANN model is preferred, due to its high precision in predicting η_ith compared to the MVR and SWR models. This study should be extremely beneficial to those coping with the design of solar stills.     

Modeling biological oxygen demand of the Melen River in Turkey using an artificial neural network technique

Artificial neural networks (ANNs) are being used increasingly to predict and forecast water resources' variables. The feed-forward neural network modeling technique is the most widely used ANN type in water resources applications. The main purpose of the study is to investigate the abilities of an artificial neural networks' (ANNs) model to improve the accuracy of the biological oxygen demand (BOD) estimation. Many of the water quality variables (chemical oxygen demand, temperature, dissolved oxygen, water flow, chlorophyll a and nutrients, ammonia, nitrite, nitrate) that affect biological oxygen demand concentrations were collected at 11 sampling sites in the Melen River Basin during 2001-2002. To develop an ANN model for estimating BOD, the available data set was partitioned into a training set and a test set according to station. In order to reach an optimum amount of hidden layer nodes, nodes 2, 3, 5, 10 were tested. Within this range, the ANN architecture having 8 inputs and 1 hidden layer with 3 nodes gives the best choice. Comparison of results reveals that the ANN model gives reasonable estimates for the BOD prediction.     

Radionuclide transport above a near-surface water table: III. Soil migration and crop uptake of three gamma-emitting radionuclides, 1990 to 1993

This paper summarizes the vertical distributions of 22Na, 137Cs, and 60Co above controlled water tables in deep and shallow lysimeters during a four-year experiment. The activity concentration profiles were all determined at the time of harvest of a winter wheat (Triticum aestivum L. cv. Pastiche) crop. Activity concentrations in different crop tissues were determined and crop uptake expressed as both an inventory ratio (IR) and a transfer factor (TFw), weighted to account for root and radionuclide distributions within the soil profile. Experimental variates were subjected to analysis of variance to determine the single and combined effects of the soil depth and the year of the experiment on the results obtained. Each radionuclide showed significant variations in activity concentration with soil depth, but the significance of these variations from year to year was dependent on radionuclide. A distinction in the behavior of weakly sorbed (22Na) and more highly sorbed (137Cs and 60Co) radionuclides was observed. The former exhibited significant variations in its distribution in the soil profile from year-to-year whereas the latter did not. Relatively high TF, values for 22Na were maintained throughout the experiment, whereas for 137Cs and 60Co, the highest TFw values were recorded in 1990 followed by a significant decline in 1991, with TFw remaining low in 1992 and 1993. The TFw values were, in general, significantly higher for deep lysimeters than for shallow lysimeters. This is thought to provide evidence of enhanced radionuclide absorption by the relatively small fraction of roots in the vicinity of the deeper water table.     

Estimating monthly total nitrogen concentration in streams by using artificial neural network

Artificial Neural Network (ANN) is a flexible and popular tool for predicting the non-linear behavior in the environmental system. Here, the feed-forward ANN model was used to investigate the relationship among the land use, fertilizer, and hydrometerological conditions in 59 river basins over Japan and then applied to estimate the monthly river total nitrogen concentration (TNC). It was shown by the sensitivity analysis, that precipitation, temperature, river discharge, forest area and urban area have high relationships with TNC. The ANN structure having eight inputs and one hidden layer with seven nodes gives the best estimate of TNC. The 1:1 scatter plots of predicted versus measured TNC were closely aligned and provided coefficients of errors of 0.98 and 0.93 for ANNs calibration and validation, respectively. From the results obtained, the ANN model gave satisfactory predictions of stream TNC and appears to be a useful tool for prediction of TNC in Japanese streams. It indicates that the ANN model was able to provide accurate estimates of nitrogen concentration in streams. Its application to such environmental data will encourage further studies on prediction of stream TNC in ungauged rivers and provide a useful tool for water resource and environment managers to obtain a quick preliminary assessment of TNC variations.     

IMPACT OF WATER CUTBACK ON SUGARCANE PRODUCTION IN THE FLORIDA EVERGLADES1

ABSTRACT: Sugarcane (Saccharum spp.) was planted in six lysimeters containing Pahokee muck (Lithic Mediaprist) where water tables were maintained at 30, 60, and 90 cm depths. The main objective was to study the impact of a 40 percent water cutback (108 mm) on sugarcane production during the period near the end of the dry season (i.e., May). The water cutback treatment was simulated through manipulation of water table depth. Due to the high available water capacity of the muck soil and selection of a sugarcane cultivar ‘CP63-588’ (which has a high tolerance of water table fluctuations), the sugarcane growth, and the yields of sugarcane biomass and sugar were not significantly different as a result of the treatments with and without 40 percent water cutback during a period of two months. This result is in good agreement with the 1981 cane yield in the Everglades Agricultural Area where a 35 percent water cutback was imposed during the 1981 drought.     

MODELING THE FOREST HYDROLOGY OF WETLAND-UPLAND ECOSYSTEMS IN FLORIDA1

ABSTRACT: Few hydrological models are applicable to pine flat-woods which are a mosaic of pine plantations and cypress swamps. Unique features of this system include ephemeral sheet flow, shallow dynamic ground water table, high rainfall and evapotranspiration, and high infiltration rates. A FLATWOODS model has been developed specifically for the cypress wetland-pine upland landscape by integrating a 2-D ground water model, a Variable-Source-Area (VAS)-based surface flow model, an evapotranspiration (ET) model, and an unsaturated water flow model. The FLATWOODS model utilizes a distributed approach by dividing the entire simulation domain into regular cells. It has the capability to continuously simulate the daily values of ground water table depth, ET, and soil moisture content distributions in a watershed. The model has been calibrated and validated with a 15-year runoff and a four-year ground water table data set from two different pine flat woods research watersheds in northern Florida. This model may be used for predicting hydrologic impacts of different forest management practices in the coastal regions.     

Predicting Debris Yield From Burned Watersheds: Comparison of Statistical and Artificial Neural Network Models1

Abstract: Alluvial fans in southern California are continuously being developed for residential, industrial, commercial, and agricultural purposes. Development and alteration of alluvial fans often require consideration of mud and debris flows from burned mountain watersheds. Accurate prediction of sediment (hyper‐concentrated sediment or debris) yield is essential for the design, operation, and maintenance of debris basins to safeguard properly the general population. This paper presents results based on a statistical model and Artificial Neural Network (ANN) models. The models predict sediment yield caused by storms following wildfire events in burned mountainous watersheds. Both sediment yield prediction models have been developed for use in relatively small watersheds (50‐800 ha) in the greater Los Angeles area. The statistical model was developed using multiple regression analysis on sediment yield data collected from 1938 to 1983. Following the multiple regression analysis, a method for multi‐sequence sediment yield prediction under burned watershed conditions was developed. The statistical model was then calibrated based on 17 years of sediment yield, fire, and precipitation data collected between 1984 and 2000. The present study also evaluated ANN models created to predict the sediment yields. The training of the ANN models utilized single storm event data generated for the 17‐year period between 1984 and 2000 as the training input data. Training patterns and neural network architectures were varied to further study the ANN performance. Results from these models were compared with the available field data obtained from several debris basins within Los Angeles County. Both predictive models were then applied for hind‐casting the sediment prediction of several post 2000 events. Both the statistical and ANN models yield remarkably consistent results when compared with the measured field data. The results show that these models are very useful tools for predicting sediment yield sequences. The results can be used for scheduling cleanout operation of debris basins. It can be of great help in the planning of emergency response for burned areas to minimize the damage to properties and lives.     

Non-linear autoregressive neural network approach for inside air temperature prediction of a pillar cooler

The volcanic plate made pillar cooler system is designed for outdoor spaces as a heat exchanging medium and reduces the outcoming air temperature which flows through the exhaust port. This paper proposes the use of artificial neural networks (ANNs) to predict inside air temperature of a pillar cooler. For this purpose, at first, three statistically significant factors (outside temperature, airing and watering) influencing the inside air temperature of the pillar cooler are identified as input parameters for predicting the output (inside air temperature) and then an ANN was employed to predict the output. In addition, 70%, 15% and 15% data was chosen from a previously obtained data set during the field trial of the pillar cooler for training, testing and validation, respectively, and then an ANN was employed to predict inside air temperature. The training (0.99918), testing (0.99799) and validation errors (0.99432) obtained from the model indicate that the artificial neural network model (NARX) may be used to predict inside air temperature of pillar cooler. This study reveals that, an ANN approach can be used successfully for predicting the performance of pillar cooler.     

Evaluation of PCA and Gamma test techniques on ANN operation for weekly solid waste prediction

Artificial neural networks (ANNs) are suitable for modeling solid waste generation. In the present study, four training functions, including resilient backpropagation (RP), scale conjugate gradient (SCG), one step secant (OSS), and Levenberg–Marquardt (LM) algorithms have been used. The main goal of this research is to develop an ANN model with a simple structure and ample accuracy. In the first step, an appropriate ANN model with 13 input variables is developed using the afore-mentioned algorithms to optimize the network parameters for weekly solid waste prediction in Mashhad, Iran. Subsequently, principal component analysis (PCA) and Gamma test (GT) techniques are used to reduce the number of input variables. Finally, comparison amongst the operation of ANN, PCA-ANN, and GT-ANN models is made. Findings indicated that the PCA-ANN and GT-ANN models have more effective results than the ANN model. These two models decrease the number of input variables from 13 to 7 and 5, respectively.     

Use of artificial neural network black-box modeling for the prediction of wastewater treatment plants performance

A reliable model for any wastewater treatment plant is essential in order to provide a tool for predicting its performance and to form a basis for controlling the operation of the process. This would minimize the operation costs and assess the stability of environmental balance. This process is complex and attains a high degree of nonlinearity due to the presence of bio-organic constituents that are difficult to model using mechanistic approaches. Predicting the plant operational parameters using conventional experimental techniques is also a time consuming step and is an obstacle in the way of efficient control of such processes. In this work, an artificial neural network (ANN) black-box modeling approach was used to acquire the knowledge base of a real wastewater plant and then used as a process model. The study signifies that the ANNs are capable of capturing the plant operation characteristics with a good degree of accuracy. A computer model is developed that incorporates the trained ANN plant model. The developed program is implemented and validated using plant-scale data obtained from a local wastewater treatment plant, namely the Doha West wastewater treatment plant (WWTP). It is used as a valuable performance assessment tool for plant operators and decision makers. The ANN model provided accurate predictions of the effluent stream, in terms of biological oxygen demand (BOD), chemical oxygen demand (COD) and total suspended solids (TSS) when using COD as an input in the crude supply stream. It can be said that the ANN predictions based on three crude supply inputs together, namely BOD, COD and TSS, resulted in better ANN predictions when using only one crude supply input. Graphical user interface representation of the ANN for the Doha West WWTP data is performed and presented.     

Evaluation of the Use of Artificial Neural Networks for the Simulation of Hybrid Solar Collectors

Abstract

In the last decade, Artificial Neural Networks (ANNs) have been receiving an increasing attention for simulating engineering systems due to some interesting characteristics such as learning capability, fault tolerance, speed and nonlinearity. This article describes an alternative approach to assess two types of hybrid solar collector/heat pipe systems (plate heat pipe type and tube heat pipe type) using ANNs. Multiple Layer Perceptrons (MLPs) and Radial Basis Networks (RBFs) were considered. The networks were trained using results from mathematical models generated by Monte Carlo simulation. The mathematical models were based on energy balances and resulted in a system of nonlinear equations. The solution of the models was very sensitive to initial estimates, and convergence was not obtained under certain conditions. Between the two neural models, MLPs performed slightly better than RBFs. It can be concluded that similar configurations were adequate for both collector systems. It was found that ANNs simulated both collector efficiency and heat output with high accuracy when “unseen” data were presented to the networks. An important advantage of a trained ANN over the mathematical models is that convergence is not an issue and the result is obtained almost instantaneously.     

Estimating Root Zone Soil Moisture at Continental Scale Using Neural Networks

This study investigates the feasibility of artificial neural networks (ANNs) to retrieve root zone soil moisture (RZSM) at the depths of 20 cm (SM20) and 50 cm (SM50) at a continental scale, using surface information. To train the ANNs to capture interactions between land surface and various climatic patterns, data of 557 stations over the continental United States were collected. A sensitivity analysis revealed that the ANNs were able to identify input variables that directly affect the water and energy balance in root zone. The data important for RZSM retrieval in a large area included soil texture, surface soil moisture, and the cumulative values of air temperature, surface soil temperature, rainfall, and snowfall. The results showed that the ANNs had high skill in retrieving SM20 with a correlation coefficient above 0.7 in most cases, but were less effective at estimating SM50. The comparison of the ANNs showed that using soil texture data improved the model performance, especially for the estimation of SM50. It was demonstrated that the ANNs had high flexibility for applications in different climatic regions. The method was used to generate RZSM in North America using Soil Moisture and Ocean Salinity (SMOS) soil moisture data, and achieved a spatial soil moisture pattern comparable to that of Global Land Data Assimilation System Noah model with comparable performance to the SMOS surface soil moisture retrievals. The models can be efficient alternatives to assimilate remote sensing soil moisture data for shallow RZSM retrieval.     

ANN OUTPUT UPDATING OF LUMPED CONCEPTUAL RAINFALL/RUNOFF FORECASTING MODELS1

ABSTRACT: Artificial neural networks (ANNs) are tested for the output updating of one‐day‐ahead and three‐day‐ahead streamflow forecasts derived from three lumped conceptual rainfall/runoff (R‐R) models: the GR4J, the IHAC, and the TOPMO. ANN output updating proved superior to a parameter updating scheme and to the ‘simple’ output updating scheme, which always replicates the last observed forecast error. In fact, ANN output updating was able to compensate for large differences in the initial performance of the three tested lumped conceptual R‐R models, which the other tested updating approaches were not able to achieve. This is done mainly by incorporating input vectors usually exploited for ANN R‐R modeling such as previous rainfall and streamflow observations, in addition to the previous observed error. For one‐day‐ahead forecasts, the performance of all three lumped conceptual R‐R models, used in conjunction with ANN output updating, was equivalent to that of the ANN R‐R model. For three‐day‐ahead forecasts, the performance of the ANN‐output‐updated conceptual models was even superior to that of the ANN R‐R model, revealing that the conceptual models are probably performing some tasks that the ANN R‐R model cannot map. However, further testing is needed to substantiate the last statement.     

Dissolved organic carbon fluxes under bare soil

The flux of dissolved organic carbon (DOC) in soil facilitates transport of nutrients and contaminants in soil. There is little information on DOC fluxes and the relationship between DOC concentration and water flux in agricultural soils. The DOC fluxes and concentrations were measured during 2.5 yr using 30 automatic equilibrium tension plate lysimeters (AETPLs) at 0.4 m and 30 AETPLs at 1.20-m depth in a bare luvisol, previously used as an arable soil. Average annual DOC fluxes of the 30 AETPLS were 4.9 g C m(-2) y(-1) at 0.4 m and 2.4 g C m(-2) y(-1) at 1.2 m depth. The average leachate DOC concentrations were 17 mg C L(-1) (0.4 m) and 9 mg C L(-1) (1.2 m). The DOC concentrations were unrelated to soil moisture content or average temperature and rarely dropped below 9 mg C L(-1) (0.4 m) and 5 mg C L(-1) (1.2 m). The variability in cumulative DOC fluxes among the plates was positively related to leachate volume and not to average DOC concentrations at both depths. This suggests that water fluxes are the main determinants of spatial variability in DOC fluxes. However, the largest DOC concentrations were inversely proportional to the mean water velocity between succeeding sampling periods, suggesting that the maximal net DOC mobilization rate in the topsoil is limited. Elevated DOC concentrations, up to 90 mg C L(-1), were only observed at low water velocities, reducing the risks of DOC-facilitated transport of contaminants to groundwater. The study emphasizes that water flux and velocity are important parameters for DOC fluxes and concentrations.     

AN EVALUATION OF THE AVAILABLE TECHNIQUES FOR ESTIMATING MISSING FECAL COLIFORM DATA1

ABSTRACT: This paper presents the findings of a study aimed at evaluating the available techniques for estimating missing fecal coliform (FC) data on a temporal basis. The techniques investigated include: linear and nonlinear regression analysis and interpolation functions, and the use of artificial neural networks (ANNs). In all, seven interpolation, two regression, and one ANN model structures were investigated. This paper also investigates the validity of a hypothesis that estimating missing FC data by developing different models using different data corresponding to different dynamics associated with different trends in the FC data may result in a better model performance. The FC data (counts/100 ml) derived from the North Fork of the Kentucky River in Kentucky were employed to calibrate and validate various models. The performance of various models was evaluated using a wide variety of standard statistical measures. The results obtained in this study are able to demonstrate that the ANNs can be preferred over the conventional techniques in estimating missing FC data in a watershed. The regression technique was not found suitable in estimating missing FC data on a temporal basis. Further, it has been found that it is possible to achieve a better model performance by first decomposing the whole data set into different categories corresponding to different dynamics and then developing separate models for separate categories rather than developing a single model for the composite data set.     

WATER TABLE INDUCED SHRUB-HERBACEOUS ECOTONE: HYDROLOGIC MANAGEMENT IMPLICATIONS1

ABSTRACT: Environmental factors were investigated across a shrub-herbaceous ecotone (sharp zone of change) on a sloping site underlain by shallow groundwater on the arid floor of Owens Valley, California. Dominant plant species were salt rabbitbrush (Chrysothamnus nauseosus ssp. consimilis [E. Greene] Hall and Clements) and saltgrass (Distichlis spicata var. stricta EL.] E. Greene); typical of many similar habitats across the Great Basin. Historic air photographs were analyzed, and soil properties, water table levels and shrub and herbaceous cover were measured at discrete sample points. To investigate soil and vegetation spatial properties, sample points were apportioned on both sides of the ecotone. Land management practices and fire were ruled out as causal factors for the ecotone which remained stable through a 45-year period of air photo record. Soil textural, chemical and hydraulic properties were similar across the ecotone and were uniform throughout the site. Only depth to the water table changed significantly in a gradient perpendicular to the ecotone. The shrub-herbaceous ecotone was located where the water table depth fluctuated periodically between 0.8 and 1.2 m; deeper water tables than this range favors shrub cover while shallower depths favors meadow vegetation. When extrapolated to hydrologic management such as groundwater pumping, such a shallow depth and a narrow range of amplitude could restrict options for water development if maintenance of meadow vegetation is a goal.     

Headcut erosive regimes influenced by groundwater on disturbed agricultural soils

A series of simulated rainfall experiments, testing several soils and slope gradients in a 10 m × 0.8 m laboratory flume, displayed close correlations between initial development of a water table at a 10 cm depth and highly erosive headcut formation. On some soils and gradients, highly erosive headcuts formed consistently and predictably within minutes or seconds of initial water table rise. However, headcuts alone were not good indicators of increased erosion. In most experiments some headcuts formed early, often when surface hydraulic parameter values reached established rill initiation thresholds, but resulted in little or no erosion increase. Later, at initial water table rise, other headcuts formed coincident with major erosion increase, often with surface hydraulic values then less than rill initiation thresholds. On the four soils tested, highly erosive headcuts never formed without groundwater development, except on steep 9° slopes.     

Testing of a GIS model ofEucalyptus Iargiflorens health on a semiarid,saline floodplain

Irrigated agriculture has resulted in substantial changes in water flows to the lower reaches of the River Murray. These changes have led to large-scale occurrences of dieback inEucalyptus largiflorens (black box) woodlands as well as increased inputs of salt to the river. Management options to address problems of this scale call for the use of spatial data sets via geographic information systems (GIS). A GIS exists for one floodplain of the River Murray at Chowilla, and a simple model predicted six health classes ofEucalyptus largiflorens based on groundwater salinity, flooding frequency, and groundwater depth.To determine the usefulness of the model for vegetation management, the quality of both the model and the GIS data sets were tested. Success of the testing procedure was judged by the degree of spatial matching between the model's predictions of health and that assessed from aerial photographs and by field truthing. Analyses at 80 sites showed that tree health was significantly greater where groundwater salinity was less than 40 dS/m or flooding occurred more frequently than 1 in 10 years or depth to groundwater exceeded 4 m. Testing of the GIS data sets found that vegetation was misclassified at 15% of sites. Association was shown between GIS-predicted values and field-truthed values of groundwater salinity but not groundwater depth. The GIS model of health is a useful starting point for future vegetation management and can be further improved by increasing the quality of the data coverages and further refining of the model to optimize parameters and thresholds.