Multilayer perceptron neural network for flow prediction

Artificial neural networks (ANNs) have proven to be a tool for characterizing, modeling and predicting many of the non-linear hydrological processes such as rainfall-runoff, groundwater evaluation or simulation of water quality. After proper training they are able to generate satisfactory predictive results for many of these processes. In this paper they have been used to predict 1 or 2 days ahead the average and maximum daily flow of a river in a small forest headwaters in northwestern Spain. The inputs used were the flow and climate data (precipitation, temperature, relative humidity, solar radiation and wind speed) as recorded in the basin between 2003 and 2008. Climatic data have been utilized in a disaggregated form by considering each one as an input variable in ANN(1), or in an aggregated form by its use in the calculation of evapotranspiration and using this as input variable in ANN(2). Both ANN(1) and ANN(2), after being trained with the data for the period 2003-2007, have provided a good fit between estimated and observed data, with R(2) values exceeding 0.95. Subsequently, its operation has been verified making use of the data for the year 2008. The correlation coefficients obtained between the data estimated by ANNs and those observed were in all cases superior to 0.85, confirming the capacity of ANNs as a model for predicting average and maximum daily flow 1 or 2 days in advance.     

Monitoring spatiotemporal variations of diel radon concentrations in peatland and forest ecosystems based on neural network and regression models

Concentrations of outdoor radon-222 (²²²Rn) in temperate grazed peatland and deciduous forest in northwestern Turkey were measured, compared, and modeled using artificial neural networks (ANNs) and multiple nonlinear regression (MNLR) models. The best-performing multilayer perceptron model selected out of 28 ANNs considerably enhanced accuracy metrics in emulating ²²²Rn concentrations relative to the MNLR model. The two ecosystems had similar diel patterns with the lowest ²²²Rn concentrations in the afternoon and the highest ones near dawn. Mean level (5.1?+?2.5 Bq?m^?3 h^?1) of ²²²Rn in the forest was three times smaller than that (15.8?+?9.7 Bq?m^?3) of ²²²Rn in the peatland. Mean ²²²Rn level had negative and positive relationships with air temperature and relative humidity, respectively.     

Artificial neural network based carbon monoxide persistence models for episodic urban air quality management

This paper describes the development of artificial neural network (ANN) based carbon monoxide (CO) persistence (ANNCOP) models to forecast 8-h average CO concentration using 1-h maximum predicted CO data for the critical (winter) period (November–March). The models have been developed for three 8-h groupings of 10 p.m. to 6 a.m., 6 a.m. to 2 p.m. and 2–10 p.m., at two air quality control regions (AQCRs) in Delhi city, representing an urban intersection and an arterial road consisting heterogeneous traffic flows. The result indicates that time grouping of 2–10 pm is dominantly affected by inversion conditions and peak traffic flow. The ANNCOP model corresponding to this grouping predicts the 8-h average CO concentrations within the accuracy range of 68–71%. The CO persistence values derived from ANNCOP model are comparable with the persistence values as suggested by the Environmental Protection Agency (EPA), USA. This work demonstrates that ANN based model is capable of describing winter period CO persistence phenomena.     

Assessment of some remote sensing techniques used to detect land use/land cover changes in South-East Transilvania,Romania

This paper assesses the image differencing technique for the Normalized Difference Vegetation Index (NDVI), the second principal component (PC2), and the TM 4 band (TM 4), as well as the post-classification comparison (PCC) in order to analyze the land use/land cover changes in the South-East Transilvania, Romania. The analysis was performed using two frames from Landsat 5 TM satellite images acquired on August 5, 1993 and July 24, 2009. After applying the NDVI, PC2, and TM 4 image differencing techniques, the images obtained were transformed into change/no change maps. The thresholds identified to highlight the changes were set at 0.6 s for NDVI and 0.7 s for PC2 and TM 4. Before applying the PCC technique, the satellite images were classified through the supervised classification method. The overall accuracy obtained was 85.91 % and the kappa statistics 0.8249 for 1993, 88.18 % and 0.8497 for 2009, respectively. The assessment of the changes detection methods in the studied area shows that the first place is occupied by NDVI image differencing with an overall accuracy of 83.80 %, followed by PCC method with 83.20 %, PC2 difference with an overall accuracy of 81.60 %, and TM 4 difference with an overall accuracy of 79.40 %.     

Integrating multiple vegetation indices via an artificial neural network model for estimating the leaf chlorophyll content of <Emphasis Type="Italic">Spartina alterniflora</Emphasis> under interspecies competition

The invasive species Spartina alterniflora and native species Phragmites australis display a significant co-occurrence zonation pattern and this co-exist region exerts most competitive situations between these two species, competing for the limited space, directly influencing the co-exist distribution in the future. However, these two species have different growth ratios in this area, which increase the difficulty to detect the distribution situation directly by remote sensing. As chlorophyll content is a key indicator of plant growth and physiological status, the objective of this study was to reduce the effect of interspecies competition when estimating Cab content; we evaluated 79 published representative indices to determine the optimal indices for estimating the chlorophyll a and b (Cab) content. After performing a sensitivity analysis for all 79 spectral indices, five spectral indices were selected and integrated using an artificial neural network (ANN) to estimate the Cab content of different competition ratios: the Gitelson ratio green index, the transformed chlorophyll absorption ratio index/optimized soil-adjusted vegetation index, the modified normalized difference vegetation index, the chlorophyll fluorescence index, and the Vogelmann chlorophyll index. The ANN method yielded better results (R ² = 0.7110 and RMSE = 8.3829 μg cm^?2) on average than the best single spectral index (R ² = 0.6319 and RMSE = 9.3535 μg cm^?2), representing an increase of 10.78% in R ² and a decrease of 10.38% in RMSE. Our results indicated that integrating multiple vegetation indices with an ANN can alleviate the impact of interspecies competition and achieve higher estimation accuracy than the traditional approach using a single index.     

Hybrid Neural Networks and Boosted Regression Tree Models for Predicting Roadside Particulate Matter

This paper examines the application of artificial neural network (ANN) and boosted regression tree (BRT) methods in air quality modelling. The methods were applied to developing air quality models for predicting roadside particle mass concentration (PM₁₀, PM_2.5) and particle number counts (PNC) based on air pollution, traffic and meteorological data from Marylebone Road in London. Elastic net, Lasso and principal components analysis were used as feature selection methods for the ANN models to reduce the number of predictor variables and improve their generalisation. The performance of the ANN with feature selection (ANN hybrid) and the BRT models was evaluated and compared using statistical performance metrics. The performance parameters include root mean square error (RMSE), fraction of prediction within a factor of two of the observation (FAC2), mean bias (MB), mean gross error (MGE), the coefficient of correlation (R) and coefficient of efficiency (CoE) values. The input variables selected by the elastic net produced the best performing ANN models. The ANN hybrid produced models performed only slightly better than the BRT models. The R values of the ANN elastic net and BRT models were 0.96 and 0.95 for PM₁₀, 0.96 and 0.96 for PM_2.5 and 0.89 and 0.87 for PNC, respectively. Their corresponding CoE values were 0.72 and 0.70 for PM₁₀, 0.74 and 0.76 for PM_2.5 and 0.81 and 0.71 for PNC respectively. About 80–99% of all the model predictions are within a factor of two of the observed particle concentrations. The BRT models offer more advantages regarding model interpretation and permit feature selection. Therefore, the study recommends the use of BRT over ANN where the model interpretation is a priority.     

Application of artificial neural network approach and remotely sensed imagery for regional eco-environmental quality evaluation

Eco-environment quality evaluation is an important research theme in environment management. In the present study, Fuzhou city in China was selected as a study area and a limited number of 222 sampling field sites were first investigated in situ with the help of a GPS device. Every sampling site was assessed by ecological experts and given an Eco-environment Background Value (EBV) based on a scoring and ranking system. The higher the EBV, the better the ecological environmental quality. Then, three types of eco-environmental attributes that are physically-based and easily-quantifiable at a grid level were extracted: (1) remote sensing derived attributes (vegetation index, wetness index, soil brightness index, surface land temperature index), (2) meteorological attributes (annual temperature and annual precipitation), and (3) terrain attribute (elevation). A Back Propagation (BP) Artificial Neural Network (ANN) model was proposed for the EBV validation and prediction. A three-layer BP ANN model was designed to automatically learn the internal relationship using a training set of known EBV and eco-environmental attributes, followed by the application of the model for predicting EBV values across the whole study area. It was found that the performance of the BP ANN model was satisfactory and capable of an overall prediction accuracy of 82.4%, with a Kappa coefficient of 0.801 in the validation. The evaluation results showed that the eco-environmental quality of Fuzhou city is considered as satisfactory. Through analyzing the spatial correlation between the eco-environmental quality and land uses, it was found that the best eco-environmental areas were related to forest lands, whereas the urban area had the relatively worst eco-environmental quality. Human activities are still considered as a major impact on the eco-environmental quality in this area.     

Using remote sensing and GIS to detect and monitor land use and land cover change in Dhaka Metropolitan of Bangladesh during 1960–2005

This paper illustrates the result of land use/cover change in Dhaka Metropolitan of Bangladesh using topographic maps and multi-temporal remotely sensed data from 1960 to 2005. The Maximum likelihood supervised classification technique was used to extract information from satellite data, and post-classification change detection method was employed to detect and monitor land use/cover change. Derived land use/cover maps were further validated by using high resolution images such as SPOT, IRS, IKONOS and field data. The overall accuracy of land cover change maps, generated from Landsat and IRS-1D data, ranged from 85% to 90%. The analysis indicated that the urban expansion of Dhaka Metropolitan resulted in the considerable reduction of wetlands, cultivated land, vegetation and water bodies. The maps showed that between 1960 and 2005 built-up areas increased approximately 15,924 ha, while agricultural land decreased 7,614 ha, vegetation decreased 2,336 ha, wetland/lowland decreased 6,385 ha, and water bodies decreased about 864 ha. The amount of urban land increased from 11% (in 1960) to 344% in 2005. Similarly, the growth of landfill/bare soils category was about 256% in the same period. Much of the city's rapid growth in population has been accommodated in informal settlements with little attempt being made to limit the risk of environmental impairments. The study quantified the patterns of land use/cover change for the last 45 years for Dhaka Metropolitan that forms valuable resources for urban planners and decision makers to devise sustainable land use and environmental planning.     

Spatial 3D distribution of soil organic carbon under different land use types

Soil organic carbon (SOC) has been assessed in three dimension (3D) in several studies, but little is known about the combined effects of land use and soil depth on SOC stocks in semi-arid areas. This paper investigates the 3D distribution of SOC to a depth of 1 m in a 4600-ha area in southeastern Iran with different land uses under the irrigated farming (IF), dry farming (DF), orchards (Or), range plants on the Gachsaran formation (RaG), and range plants on a quaternary formation (RaQ). Predictions were made using the artificial neural networks (ANNs), regression trees (RTs), and spline functions with auxiliary covariates derived from a digital elevation model (DEM), the Landsat 8 imagery, and land use types. Correlation analysis showed that the main predictors for SOC in the topsoil were covariates derived from the imagery; however, for the lower depths, covariates derived from both the DEM and imagery were important. ANNs showed more efficiency than did RTs in predicting SOC. The results showed that 3D distribution of SOC was significantly affected by land use types. SOC stocks of soils under Or and IF were significantly higher than those under DF, RaG, and RaQ. The SOC below 30 cm accounted for about 59% of the total soil stock. Results showed that depth functions combined with digital soil mapping techniques provide a promising approach to evaluate 3D SOC distribution under different land uses in semi-arid regions and could be used to assess changes in time to determine appropriate management strategies.     

The geostatistic-based spatial distribution variations of soil salts under long-term wastewater irrigation

The purpose of this study was to determine and evaluate the spatial changes in soil salinity by using geostatistical methods. The study focused on the suburb area of Beijing, where urban development led to water shortage and accelerated wastewater reuse to farm irrigation for more than 30 years. The data were then processed by GIS using three different interpolation techniques of ordinary kriging (OK), disjunctive kriging (DK), and universal kriging (UK). The normality test and overall trend analysis were applied for each interpolation technique to select the best fitted model for soil parameters. Results showed that OK was suitable for soil sodium adsorption ratio (SAR) and Na⁺ interpolation; UK was suitable for soil Cl^? and pH; DK was suitable for soil Ca²⁺. The nugget-to-sill ratio was applied to evaluate the effects of structural and stochastic factors. The maps showed that the areas of non-saline soil and slight salinity soil accounted for 6.39 and 93.61 %, respectively. The spatial distribution and accumulation of soil salt were significantly affected by the irrigation probabilities and drainage situation under long-term wastewater irrigation.     

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.     

Capability Maturity Model (CMM) method for assessing the performance of low-carbon city practice

Carbon emissions are considered as major factor affecting sustainable urban development. Cities have been promoting low carbon city (LCC) strategies to reduce carbon emissions and various methods have been introduced to assess the performance of LCC strategies. However, most of the existing assessment methods focus on overall LCC performance at city level, but the effects of individual dimensions like industrial structure and energy efficiency are ignored. The research question in this study therefore is whether an alternative method can be established to measure both the overall and dimensional LCC performance. This paper introduces a method for assessing LCC performance by using Capability Maturity Model (CMM), which is called LCC-CMM. The proposed method of LCC-CMM can help identify LCC maturity grade through assessing the performance of individual LCC dimensions which contributes to overall performance. There are four procedures in applying LCC-CMM, including to identify the Key Process Areas (KPAs) in the context of LCC, denoted as LCC-KPAs, to define the indicators for measuring the performance of LCC-KPAs, to calculate the performance score for each LCC-KPA, and to define criteria for specifying different grades of LCC capability maturity. The proposed method is proven effective through a case demonstration. The method can help policy makers to identify weak areas in LCC practice and introduce tailor-made policy measures to improve the weak areas.     

Fine-resolution mapping of wildfire fuel types for Canada: Fuzzy logic modeling for an Alberta pilot area

The Canadian Forest Fire Danger Rating System (CFFDRS) is used daily across Canada for evaluating forest fire danger. Fuel-type information is one of the inputs required by the models used in the CFFDRS. In this project, three fuel-type maps with a 25 m resolution were produced for a pilot study area located in Alberta using land cover only; land cover and biomass; and, land cover, biomass and leaf area index data derived from satellite imagery. The relationships between inputs and fuel types were determined mainly by the opinions of forest fire scientists and incorporated into a computer program using fuzzy set methodology. Not all the CFFDRS fuel types could be distinguished using these inputs; three of the coniferous types had to be grouped into one common fuel type. Overall accuracy was between 74 and 83% based on ground-truth comparisons. The most accurate map resulted from land cover and biomass data. Detailed accuracy assessment indicated that the overall accuracy increased up to 86% if ambiguous fuel type identification was considered. No combination of inputs was able to define a fuel type with absolute certainty, which is a reflection of differing expert opinions and the small number of inputs used to produce the maps.     

Noise tolerance of algorithms for estimating chlorophyll a concentration in turbid waters

The accuracy and noise tolerance of 13 global models and 5 Case II chlorophyll a (chl a) retrieval models were evaluated using three dataset. It was found that if 5 % input noise related to atmospheric correction is considered, then the uncertainty associated with noise tolerance varied from 5.5 % to 55.6 %, and these uncertainties generally accounts for 15.63 % to 24.75 % of the total uncertainty. This observation suggests that an optimal algorithm not only should have a strong chl a concentration prediction ability but also should possess high insensitivity to the noise of remote-sensing imagery. The accuracy evaluations of chl a models were based on comparisons of chl a predicted models with chl a concentration measured analytically for field measurements. The results indicate that none of the selected chl a estimation algorithms provide accurate retrievals of chl a in turbid waters. This may be attributed to the strong optical influence of organic and inorganic matter at the blue green range, and the non-negligible of non-organic matter absorption at the red and near-infrared ranges. In order to solve this problem, the chl a concentration retrieval models must be further optimized. After being optimized using the empirical optimized method constructed in this paper, a single parameterized NDCI (normalized difference chl a index) model produces accurate retrievals in the Yellow River Estuary, Taihu Lake and Chesapeake Bay. If 5 % input noise associated with residual uncertainty 0of atmospheric correction is taken into account, the model produces only 29.96 % uncertainty for the remote sensing of chl a concentration in these three turbid waters.     

Determining land use changes by radar-optic fused images and monitoring its environmental impacts in Edremit region of western Turkey

Rapid and unplanned urbanization and industrialization are the main reasons of environmental problems. If urban growth is not based on resource sustainability criteria and urban plans are not applied, natural and human resources are damaged dramatically. In this study, land use change and urban expansion in Edremit region of Turkey is determined by means of remote sensing techniques between 1971 and 2002. To improve the accuracy of land use/cover maps, the contribution of SAR images to optic images in defining land cover types was investigated. To determine the situation of land use/cover types in 2002 accurately, Landsat-5 images and Radarsat-1 images were fused, and the land use/cover types were defined from the fused images. Comparisons with the ground truth reveal that land use maps generated using the fuse technique are improved about 6% with an accuracy of 81.20%. To define land use types and urban expansion, screen digitizing and classification methods were used. The results of the study indicate that the urban areas have been increased 1,826 ha across the agricultural fields which are in land use capability classes of I and II, and significant environmental changes such as land degradation and degeneration of ground water quality occurred.     

LaSVM-based big data learning system for dynamic prediction of air pollution in Tehran

Due to critical impacts of air pollution, prediction and monitoring of air quality in urban areas are important tasks. However, because of the dynamic nature and high spatio-temporal variability, prediction of the air pollutant concentrations is a complex spatio-temporal problem. Distribution of pollutant concentration is influenced by various factors such as the historical pollution data and weather conditions. Conventional methods such as the support vector machine (SVM) or artificial neural networks (ANN) show some deficiencies when huge amount of streaming data have to be analyzed for urban air pollution prediction. In order to overcome the limitations of the conventional methods and improve the performance of urban air pollution prediction in Tehran, a spatio-temporal system is designed using a LaSVM-based online algorithm. Pollutant concentration and meteorological data along with geographical parameters are continually fed to the developed online forecasting system. Performance of the system is evaluated by comparing the prediction results of the Air Quality Index (AQI) with those of a traditional SVM algorithm. Results show an outstanding increase of speed by the online algorithm while preserving the accuracy of the SVM classifier. Comparison of the hourly predictions for next coming 24 h, with those of the measured pollution data in Tehran pollution monitoring stations shows an overall accuracy of 0.71, root mean square error of 0.54 and coefficient of determination of 0.81. These results are indicators of the practical usefulness of the online algorithm for real-time spatial and temporal prediction of the urban air quality.     

Modeling seasonal variations of long-term soil CO<Subscript>2</Subscript> emissions in an orchard plantation in a semiarid area,SE Turkey

Emissions of soil CO₂ under different management systems have a significant effect on the carbon balance in the atmosphere. Soil CO₂ emissions were measured from an apricot orchard at two different locations: under the crown of trees (CO₂-UC) and between tree rows (CO₂-BR). For comparison, one other measurement was performed on bare soil (CO₂-BS) located next to the orchard field. Analytical data were obtained weekly during 8 years from April 2008 to December 2016. Various environmental parameters such as air temperature, soil temperature at different depths, soil moisture, rainfall, and relative humidity were used for modeling and estimating the long-term seasonal variations in soil CO₂ emissions using two different methods: generalized linear model (GLM) and artificial neural network (ANN). Before modeling, data were randomly split into two parts, one for calibration and the second for validation, with a varying number of samples in each part. Performances of the models were compared and evaluated using means absolute of estimations (MAE), square root of mean of prediction (RMSEP), and coefficient of determination (R²) values. CO₂-UC, CO₂-BR, and CO₂-BS values ranged from 11 to 3985, from 9 to 2365, and from 8 to 1722 kg ha^?1 week^?1, respectively. Soil CO₂ emissions were significantly correlated (p?<?0.05) with some environmental variables. The results showed that GLM and ANN models provided similar accuracies in modeling and estimating soil CO₂ emissions, as the number of samples in the validation data set increased. The ANN was more advantageous than GLM models by providing a better fit between actual observations and predictions and lower RMSEP and MAE values. The results suggested that the success of environmental variables for estimations of CO₂ emissions using the two methods was moderate.     

Field monitoring of a LID-BMP treatment train system in China

In order to assess the urban runoff control effectiveness of a low-impact development best management practice (LID-BMP) treatment train system, a field test of selected LID-BMPs was conducted in China. The LID-BMPs selected include three grassed swales, a buffer strip, a bioretention cell, two infiltration pits, and a constructed wetland. The test site is in a campus in southern China. The LID-BMPs, connected in a series, received stormwater runoff from four tennis courts with an area of 2808 m² and eight basketball courts with an area of 4864 m². Construction of the LID-BMPs was completed in early spring of 2012, and the sampling was conducted during May of 2012 to September of 2013. During the sampling effort, besides the performance evaluations of grassed swales and the bioretention cell in controlling runoff quantity as well as quality, the emphasis was also on determining the performance of the LID-BMP treatment train system. A total of 19 storm events were monitored, with nine producing no runoff and ten producing runoff. Data collected from the ten storm events were analyzed for estimating runoff quantity (peak flow rate and total runoff volume) and quality reduction by the LID-BMPs. The sum of loads (SOL) method was used for calculating the water quality performance of LID-BMPs. Results indicated that, for peak flow rate, a bioretention cell reduction of 50–84 % was obtained, and grassed swale reduction was 17–79 %, with a runoff volume reduction of 47–80 and 9–74 %, respectively. For water quality, the bioretention cell in general showed good removal for zinc (nearly 100 %), copper (69 %), NH₃-N (ammonia nitrogen) (51 %), and total nitrogen (TN) (49 %); fair removal for chemical oxygen demand (COD) (18 %); and poor removal for total suspended solids (TSS) (?11 %) and total phosphorus (TP) (?21 %). And its performance effectiveness for pollutant removal increased in the second year after 1 year of stabilizing. When considering the aggregated effect of the LID-BMP treatment train system, it showed excellent removal for NH₃-N (73 %), TN (74 %), and TP (95 %) and fair removal for COD (19 %) and TSS (35 %). The assessment results of the LID-BMP treatment train system provide valuable information on how to link the different types of LID-BMP facilities and maximize the integrated effectiveness on urban runoff control.     

Improving Estimation of Specific Surface Area by Artificial Neural Network Ensembles Using Fractal and Particle Size Distribution Curve Parameters as Predictors

Specific surface area (SSA) is one of the principal soil properties used in modeling soil processes. In this study, artificial neural network (ANN) ensembles were evaluated to predict SSA. Complete soil particle-size distribution was estimated from sand, silt, and clay fractions using the model by Skaggs et al. and then the particle-size distribution curve parameters (PSDCPs) and fractal parameters were calculated. The PSDCPs were used to predict 20 particle-size classes for a soil sample’s particle size distribution. Fractal parameters were calculated by the model of Bird et al. In addition, total soil-specific surface area (TSS) was calculated using the above 20 size classes. Pedotransfer functions were developed for SSA and TSS using ANN ensembles from 63 pieces of SSA data taken from the literature. Fractal parameters, PSDCPs, and some other soil properties were used to predict SSA and TSS. Introducing fractal parameters and PSDCPs improved the SSA estimations by 12.5 and 11.1 %, respectively. The improvements were even better for TSS estimations (27.7 and 27.0 %, respectively). The use of fractal parameters as estimators described 44 and 92.8 % of the variation in SSA and TSS, respectively, while PSDCPs explained 42 and 6.6 % of the variation in SSA and TSS, respectively. The results suggested that fractal parameters and PSDCPs could be successfully used as predictors in ANN ensembles to predict SSA and TSS.     

Delineation of Environmental Units by Multivariate Techniques in the Duero River Watershed,Michoacán,Mexico

Environmental or hydrological landscape units can integrate various environmental characteristics to support proper management of natural resources. To delineate these units, quantitative methods such as ordination, clustering, and classification of abiotic factor information are used. In the present work, environmental units were delineated in the Duero River watershed of Michoacán, Mexico. This will enhance understanding of the hydrologic landscape, which is a fundamental to natural resource management. A digital elevation model was used to generate sub-basins. Climatic data were obtained from 16 meteorological stations. Sixty-nine soil and 150 water samples were collected and analyzed in the laboratory. Geostatistical methods for spatial prediction of the environmental variables were used. Mean data for each sub-basin were obtained from the environmental variable grids, generating an abiotic factor data matrix. A multivariate analysis was conducted. Exponential, linear, spherical, and Gaussian models were fit to an empirical variogram. Spatial prediction of the environmental data was done via universal and ordinary kriging. Based on principal component analysis, abiotic factors evaporation, total nitrogen, soil pH, and sodium absorption ratio of water were selected for cluster analysis. Five environmental units were delineated in the Duero watershed. One environmental unit (number 4) provided greater than 50 % of the payment for ecosystem services. The general trend is an increase of urban area. The urban surface in 1983 and 2014 was 1724 and 4750 ha, respectively, an increase of 275 %. Environmental unit 1 showed the greatest urban area growth (1336 ha) during the latter period.