Predicting coral bleaching in response to environmental stressors using 8 years of global-scale data

Coral reefs have experienced extensive mortality over the past few decades as a result of temperature-induced mass bleaching events. There is an increasing realization that other environmental factors, including water mixing, solar radiation, water depth, and water clarity, interact with temperature to either exacerbate bleaching or protect coral from mass bleaching. The relative contribution of these factors to variability in mass bleaching at a global scale has not been quantified, but can provide insights when making large-scale predictions of mass bleaching events. Using data from 708 bleaching surveys across the globe, a framework was developed to predict the probability of moderate or severe bleaching as a function of key environmental variables derived from global-scale remote-sensing data. The ability of models to explain spatial and temporal variability in mass bleaching events was quantified. Results indicated approximately 20% improved accuracy of predictions of bleaching when solar radiation and water mixing, in addition to elevated temperature, were incorporated into models, but predictive accuracy was variable among regions. Results provide insights into the effects of environmental parameters on bleaching at a global scale.     

Eutrophication Prediction Using a Markov Chain Model: Application to Lakes in the Yangtze River Basin,China

Lake eutrophication is harmful and difficult to predict due to its complex evolution. As an alternative to existing mechanistic models, a Markov chain model was developed to predict the development of lake eutrophication based on an 11-year dataset in 41 lakes of the Yangtze River Basin. This model was validated using a real-time update strategy and was demonstrated to be reliable. Based on the dataset, the lake eutrophication dynamics from 2000 to 2010 were analyzed. Lakes with different trophic states from 2011 to 2050 and their responses to different water management practices were simulated based on the developed model. The simulation results show that lake eutrophication would worsen from 2011 to 2040; however, eutrophication could be significantly alleviated by changing 100 km² of hypereutrophic lakes into eutrophic lakes per year from 2010 to 2020. The nutrient conditions in most of the lakes in the Yangtze River Basin show that phosphorus control would be more efficient than nitrogen control in eutrophication management practices. This case study demonstrates the utility of Markov chain models in using prior information to predict the long-term evolution of lake eutrophication at large spatial scales. The Markov chain technique can be easily adapted to predict evolutionary processes in other disciplines.     

Estimating overall persistence and long-range transport potential of persistent organic pollutants: a comparison of seven multimedia mass balance models and atmospheric transport models

Two different approaches to modeling the environmental fate of organic chemicals have been developed in recent years. The first approach is applied in multimedia box models, calculating average concentrations in homogeneous boxes which represent the different environmental media, based on intermedia partitioning, transport, and degradation processes. In the second approach, used in atmospheric transport models, the spatially and temporally variable atmospheric dynamics form the basis for calculating the environmental distribution of chemicals, from which also exchange processes to other environmental media are modeled. The main goal of the present study was to investigate if the multimedia mass balance models CliMoChem, SimpleBox, EVn-BETR, G-CIEMS, OECD Tool and the atmospheric transport models MSCE-POP and ADEPT predict the same rankings of the overall persistence (P(ov)) and long-range transport potential (LRTP) of POPs, and to explain differences and similarities between the rankings by the mass distributions and inter-compartment mass flows. The study was performed for a group of 14 reference chemicals. For P(ov), the models yield consistent results, owing to the large influence of phase partitioning parameters and degradation rate constants, which are used similarly by all models. Concerning LRTP, there are larger differences between the models than for P(ov), due to different LRTP calculation methods and spatial model resolutions. Between atmospheric transport models and multimedia fate models, no large differences in mass distributions and inter-compartment flows can be recognized. Deviations in mass flows are mainly caused by the geometrical design of the models.     

An experimental method to verify soil conservation by check dams on the Loess Plateau, China

A successful experiment with a physical model requires necessary conditions of similarity. This study presents an experimental method with a semi-scale physical model. The model is used to monitor and verify soil conservation by check dams in a small watershed on the Loess Plateau of China. During experiments, the model–prototype ratio of geomorphic variables was kept constant under each rainfall event. Consequently, experimental data are available for verification of soil erosion processes in the field and for predicting soil loss in a model watershed with check dams. Thus, it can predict the amount of soil loss in a catchment. This study also mentions four criteria: similarities of watershed geometry, grain size and bare land, Froude number (Fr) for rainfall event, and soil erosion in downscaled models. The efficacy of the proposed method was confirmed using these criteria in two different downscaled model experiments. The B-Model, a large scale model, simulates watershed prototype. The two small scale models, D_a and D_b, have different erosion rates, but are the same size. These two models simulate hydraulic processes in the B-Model. Experiment results show that while soil loss in the small scale models was converted by multiplying the soil loss scale number, it was very close to that of the B-Model. Obviously, with a semi-scale physical model, experiments are available to verify and predict soil loss in a small watershed area with check dam system on the Loess Plateau, China.     

Modeling of NO x Emissions from Oil Refinery Furnaces

The current paper investigates the possibility of establishing an empirically based model for predicting the emission rate of nitrogen oxides (NO _x ) from oil refinery furnaces, in order to continually track emissions with respect to environmental licence limits. Model input data were collected by direct stack monitoring using an electrochemical cell NO _x analyser, as well as a range of telemetry sensors to obtain refinery process parameters. Principal Component Analysis (PCA), in conjunction with Partial Least Squares (PLS) regression was then used to build a series of models able to predict NO _x emissions from the furnaces. The models produced were proven to be robust, with a relatively high accuracy, and are able to predict NO _x levels over the range of operating conditions which were sampled. It was found that due to structural/operational variations a separate model is usually required for each furnace. The models can be integrated with the refinery operating system to predict NO _x emission rates on a continuous basis. Two models representing structurally different furnaces are considered in this paper. This revised version was published online in July 2006 with corrections to the Cover Date.     

Predicting copper concentrations in acid mine drainage: a comparative analysis of five machine learning techniques

Acid mine drainage (AMD) is a global problem that may have serious human health and environmental implications. Laboratory and field tests are commonly used for predicting AMD, however, this is challenging since its formation varies from site-to-site for a number of reasons. Furthermore, these tests are often conducted at small-scale over a short period of time. Subsequently, extrapolation of these results into large-scale setting of mine sites introduce huge uncertainties for decision-makers. This study presents machine learning techniques to develop models to predict AMD quality using historical monitoring data of a mine site. The machine learning techniques explored in this study include artificial neural networks (ANN), support vector machine with polynomial (SVM-Poly) and radial base function (SVM-RBF) kernels, model tree (M5P), and K-nearest neighbors (K-NN). Input variables (physico-chemical parameters) that influence drainage dynamics are identified and used to develop models to predict copper concentrations. For these selected techniques, the predictive accuracy and uncertainty were evaluated based on different statistical measures. The results showed that SVM-Poly performed best, followed by the SVM-RBF, ANN, M5P, and KNN techniques. Overall, this study demonstrates that the machine learning techniques are promising tools for predicting AMD quality.     

The effect of thermodynamic data on computer model predictions of uranium speciation in natural water systems

Computer models have found widespread application in order to help elucidate and predict changes in environmental systems. One such application is the prediction of trace metal speciation in aqueous systems. This is achieved by solving a set of non-linear equations involving equilibrium constants for all the components in the system, within mass and charge balance constraints. In this study a comparison of the predicted uranium speciation from two computer programs, WHAM and PHREEQCI, is used to illustrate the effect variations in thermodynamic data can have on the models produced. Using the original thermodynamic data provided with the models, WHAM predicted the UO2(2+) ion as the major species (84%) while PHREEQCI predicted UO2(HPO4)2(2-) as the major species (86%). Substituting uranium data from the Nuclear Energy Agency Thermochemical Database project (NEA-TDB) into both programs produced similar results from each program, with UO2F+ predicted to dominate (68%) in a groundwater sample. Natural water samples often contain humic substances. The possible interaction of such substances with uranium was also modelled. The WHAM program includes a discreet site electrostatic humic substance model, however in order to use the PHREEQCI program to model humic substance interactions, a 'model fulvic acid' dataset was added to the program. These models predicted 85 to 98% uranium-humic substance species at neutral pH. This indicates that humic substances do need to be taken into account when modelling uranium speciation in natural water samples.     

Development of an Ozone Forecasting Model for Non-Attainment Areas in the State of Ohio

Ground level ozone is responsible for the formation ofsmog, and for a variety of adverse effects on bothhuman and plant life. High concentrations of groundlevel ozone occur during the summer months. This paperdescribes the development of a model to forecast themaximum daily concentration of ozone as a function ofthe maximum surface temperature, for ozonenon-attainment regions in Ohio. The model wasdeveloped by statistical analysis of existing data.Site-specific models were developed initially. Theverification and evaluation of the performancecriteria of the model at each site were explored bycomparing the model with an independent datasetcollected from that site. A generalized statewidemodel was developed from the site-specific models. Theperformance criteria of this model were verified andevaluated by employing the same independent datasetsemployed for the site-specific models. An exceedencemodel to predict the occurrence of ozone exceedencesover 100 ppb has also been presented.     

A Comparison Between Regression Models and Genetic Programming for Predictions of Chlorophyll-a Concentrations in Northern Lakes

Chlorophyll-a (chl-a) concentrations are often used as a proxy for water quality problems as well as phytoplankton blooms. Available chl-a models range from simple phosphorus loading models to complex regression and dynamic models. A comparison of multiple regression models was made with genetic programming (GP) techniques to predict chl-a concentrations over a large range of 104 Swedish lakes. Independent variables used were lake area, mean depth, iron, latitude, ammonium, nitrogen + nitrate, pH, phosphate, secchi depth, silicon, temperature, total phosphorus, total nitrogen and total organic carbon. GP is a method based on the Darwinian evolution theory. This implies that a program will be able to test different mathematical equations, iterating and improving each equation using fundamental ideas from evolution theory to increase the predictive power. A good correspondence was found between the multiple regression and the GP modelling approach. No significant improvement of the predictive power was found using GP, and it is therefore recommended that multiple regression methods should be preferred when predicting chl-a concentrations as these models tend to be less complex and the modelling approach is easier to use. Results from GP were in some cases more accurate compared to multiple regressions; however, the best model was created by multiple regressions which used concentrations of total phosphorus, total nitrogen and latitude as independent variables. These findings will be an important note for limnologists and modelling managers when developing future models of chl-a concentrations in lakes.     

The effectiveness of digital soil mapping to predict soil properties over low-relief areas

This study investigates the ability of different digital soil mapping (DSM) approaches to predict some of physical and chemical topsoil properties in the Shahrekord plain of Chaharmahal-Va-Bakhtiari province, Iran. According to a semi-detailed soil survey, 120 soil samples were collected from 0 to 30 cm depth with approximate distance of 750 m. Particle size distribution, coarse fragments (CFs), electrical conductivity (EC), pH, organic carbon (OC), and calcium carbonate equivalent (CCE) were determined. Four machine learning techniques, namely, artificial neural networks (ANNs), boosted regression tree (BRT), generalized linear model (GLM), and multiple linear regression (MLR), were used to identify the relationship between soil properties and auxiliary information (terrain attributes, remote sensing indices, geology map, existing soil map, and geomorphology map). Root-mean-square error (RMSE) and mean error (ME) were considered to determine the performance of the models. Among the studied models, GLM showed the highest performance to predict pH, EC, clay, silt, sand, and CCE, whereas the best model is not necessarily able to make accurate estimation. According to RMSE%, DSM has a good efficiency to predict soil properties with low and moderate variabilities. Terrain attributes were the main predictors among different studied auxiliary information. The accuracy of the estimations with more observations is recommended to give a better understanding about the performance of DSM approach over low-relief areas.     

Nutrient dynamics between sediment and overlying water in the inner part of Izmir Bay, Eastern Aegean

In this article a concept is described in order to predict and map the occurrence of benthic communities within and near the German Exclusive Economic Zone (EEZ) of the North Sea. The approach consists of two work steps: (1) geostatistical analysis of abiotic measurement data and (2) calculation of benthic provinces by means of Classification and Regression Trees (CART) and GIS-techniques. From bottom water measurements on salinity, temperature, silicate and nutrients as well as from punctual data on grain size ranges (0–20, 20–63, 63–2,000 μ) raster maps were calculated by use of geostatistical methods. At first the autocorrelation structure was examined and modelled with help of variogram analysis. The resulting variogram models were then used to calculate raster maps by applying ordinary kriging procedures. After intersecting these raster maps with punctual data on eight benthic communities a decision tree was derived to predict the occurrence of these communities within the study area. Since such a CART tree corresponds to a hierarchically ordered set of decision rules it was applied to the geostatistically estimated raster data to predict benthic habitats within and near the EEZ.     

The placement strategies of structural best management practices for different moving rainstorms

Hydrological responses and pollutant exports are always highly related to rainfall characteristics. Many studies have demonstrated that the influence of moving rainstorm on flows and mass transport process in hydrologic systems cannot be ignored. Best management practices (BMPs) are popularly applied for controlling water quantity and water quality in a watershed. Since the movements of rainstorm can influence watershed responses, BMP placement strategies should be suitably adjusted in different moving rainstorms. This study designed an intermediate rainfall pattern with varied movement behavior and tried to find the optimal BMP placement strategies, which cannot only satisfy environmental standards but also improve economic benefits, for the rainfall events. The result shows that the control efficiency of pollutant and runoff can highly improve when the BMPs are set near the outlet of a watershed. Since the economic efficiency is always regarded as an important factor, the BMP placement strategy is significant for watershed conservation and management.     

NEMS and MARKAL-MACRO Models for Energy-Environmental-Economic Analysis: A Comparison of the Electricity and Carbon Reduction Projections

The Annual Energy Outlook forecasts published by the United States Energy Information Administration (EIA) of the Department of Energy are based on results from the National Energy Modeling system (NEMS). This paper compares NEMS, which is used only in the U.S., with the U.S. version of MARKAL-MACRO (USMM) model, which is used in more than thirty-five countries. The two models predict similar results for the base 1999 US Annual Energy Outlook (AEO), but their results with carbon constraints are quite different. The differences of the models and those of their predictions are explained. USMM can be used to provide an alternative and complementary approach to projections of renewable technologies penetration and their potential in reducing carbon dioxide emissions in the USA.     

Interpretation and Discrimination of Marshy Wetlands by Soil Factors in the Kuan-Tu Natural Park, Taiwan

This work investigated soil samples collected from Kuan–Tu wetlands, Taiwan. Factor analysis was performed to explain the impact of various soil factors on this marshy wetlands located in suburban Taipei. The results indicated that the latent factors were heavy metals, salinity, and soil organic matter. Canonical discriminant analysis was used to improve an existing vegetation classi–fication scheme by identifying the physical-chemical properties of sediment in Kaun–Tu wetlands, Taiwan. Predictive discriminant analysis was used to examine the ability of the models to predict class membership for unknown soil sample. Multivariate analysis of the spatial patterns of soil quality and vegetation types showed that different properties of soil grew different types of vegetation and absorbed contaminants differently. We can feasibly conserve a suitable habitat for wetland biology by processing these unstable predictor variables. The methodology and results provide useful information concerning the Kuan–Tu wetlands and may be applicable to other wetlands with similar properties that are experiencing similar environmental issues.     

Geostatistical analysis of Palmerton soil survey data

This paper presents a literature review focused on predictive technique audits, one of the types of audit considered to have the greatest potential role in improving environmental impact assessment practice. The literature review is limited to US literature with the exception of a few UK audits, one undertaken by Tomlinson at the University of Aberdeen. The authors are, however, aware that literature from other countries exists on this subject, for example from Canada and South Africa.In the review, predictive technique audits performed for or by the US Bureau of Land Management, the Electric Power Research Institute, the US Nuclear Regulatory Commission, the US Corps of Engineers, together with the Wisconsin Power Plant Impact Study are described. In addition, articles describing the auditing of models designed to predict environmental change are reviewed, before details of auditing activity in the UK are presented.     

An expert system for water quality modelling

The RAISON-micro (Regional Analysis by Intelligent System ON a micro-computer) expert system is being used to predict the effects of mine effluents on receiving waters in Ontario. The potential of this system to assist regulatory agencies and mining industries to define more acceptable effluent limits was shown in an initial study. This system has been further developed so that the expert system helps the model user choose the most appropriate model for a particular application from a hierarchy of models. The system currently contains seven models which range from steady state to time dependent models, for both conservative and nonconservative substances in rivers and lakes. The menu driven expert system prompts the model user for information such as the nature of the receiving water system, the type of effluent being considered, and the range of background data available for use as input to the models. The system can also be used to determine the nature of the environmental conditions at the site which are not available in the textual information database, such as the components of river flow. Applications of the water quality expert system are presented for representative mine sites in the Timmins area of Ontario.     

Prediction of daily ground-level ozone concentration maxima over New Delhi

The pollution levels in New Delhi from industrial, residential, and transportation sources are continuously growing. As one of the major pollutants, ground-level ozone is responsible for various adverse effects on both humans and foliage. The present study aims to predict daily ground-level ozone concentration maxima over a site situated in New Delhi through neural networks (NN) and multiple-regression (MR) analysis. Although these methodologies are case and site specific, they are being developed and used widely. Therefore, to test these methodologies for New Delhi where no such study is available for ground-level ozone, six models have been developed based on NNs and MR using the same input data set. The changes in the performance capability of the two methods are sensitive to the selection of input parameters. The results are encouraging, and remarkable improvements in the performance of the models have been observed.     

Regional data refine local predictions: modeling the distribution of plant species abundance on a portion of the central plains

Species distribution models are frequently used to predict species occurrences in novel conditions, yet few studies have examined the consequences of extrapolating locally collected data to regional landscapes. Similarly, the process of using regional data to inform local prediction for species distribution models has not been adequately evaluated. Using boosted regression trees, we examined errors associated with extrapolating models developed with locally collected abundance data to regional-scale spatial extents and associated with using regional data for predictions at a local extent for a native and non-native plant species across the northeastern central plains of Colorado. Our objectives were to compare model results and accuracy between those developed locally and extrapolated regionally, those developed regionally and extrapolated locally, and to evaluate extending species distribution modeling from predicting the probability of presence to predicting abundance. We developed models to predict the spatial distribution of plant species abundance using topographic, remotely sensed, land cover and soil taxonomic predictor variables. We compared model predicted mean and range abundance values to observed values between local and regional. We also evaluated model prediction performance based on Pearson's correlation coefficient. We show that: (1) extrapolating local models to regional extents may restrict predictions, (2) regional data can help refine and improve local predictions, and (3) boosted regression trees can be useful to model and predict plant species abundance. Regional sampling designed in concert with large sampling frameworks such as the National Ecological Observatory Network may improve our ability to monitor changes in local species abundance.     

Analysing the sensitivity behaviour of two hydrology models

Recently, the New Morris Method has been presented as an effective sensitivity analysis tool for mathematical models. The New Morris Method estimates the sensitivity of an output parameter to a given set of input parameters (first-order effects) and the extent these parameters interact with each other (second-order effects). This method requires the specification of two parameters (runs and resolution) that control the sampling of the output parameter to determine its sensitivity to various inputs. The criteria for these parameters have been set on the analysis of a well-behaved analytical function (see Cropp and Braddock, Reliab. Eng. Syst. Saf. 78:77–83, 2002), which may not be applicable to other physical models that describe complex processes. This paper will investigate the appropriateness of the criteria from (Cropp and Braddock, 2002) and hence the effectiveness of the New Morris Method to determine the sensitivity behaviour of two hydrologic models: the Soil Erosion and Deposition System and Griffith University Representation of Urban Hydrology. In the first case, this paper will separately analyse the sensitivity of an output parameter on a set of input parameters (first- and second-order effects) for each model and discuss the physical meaning of these sensitivities. This will be followed by an investigation into the sampling criteria by exploring the convergence of the sensitivity behaviour for each model as the sampling of the parameter space is increased. By comparing these trends to the convergence behaviour from Cropp and Braddock (2002), we will determine how well the New Morris Method estimates the sensitivity for each model and whether the sampling criteria are appropriate for these models. It will be shown that the New Morris Method can provide additional insight into the functioning of these models, and that, under a different metric, the sensitivity behaviour of these models does converge confirming the sampling criteria set by Cropp and Braddock.