A Comparison of Statistical Approaches for Predicting Stream Temperatures Across Heterogeneous Landscapes1

Abstract: Estimating stream temperatures across broad spatial extents is important for regional conservation of running waters. Although statistical models can be useful in this endeavor, little information exists to aid in the selection of a particular statistical approach. Our objective was to compare the accuracy of ordinary least‐squares multiple linear regression, generalized additive modeling, ordinary kriging, and linear mixed modeling (LMM) using July mean stream temperatures in Michigan and Wisconsin. Although LMM using low‐rank thin‐plate smoothing splines to measure the spatial autocorrelation in stream temperatures was the most accurate modeling approach; overall, there were only slight differences in prediction accuracy among the evaluated approaches. This suggests that managers and researchers can select a stream temperature modeling approach that meets their level of expertise without sacrificing substantial amounts of prediction accuracy. The most accurate models for Michigan and Wisconsin had root mean square errors of 2.0‐2.3°C, suggesting that only relatively coarse predictions can be produced from landscape‐based statistical models at regional scales. Explaining substantially more variability in stream temperatures likely will require the collection of finer‐scale hydrologic and physiographic data, which may be cost prohibitive for monitoring and assessing stream temperatures at regional scales.     

Evaluation of aerial photography for predicting trends in structural attributes of Australian woodland including comparison with ground-based monitoring data

The accurate assessment of trends in the woody structure of savannas has important implications for greenhouse accounting and land-use industries such as pastoralism. Two recent assessments of live woody biomass change from north-east Australian eucalypt woodland between the 1980s and 1990s present divergent results. The first estimate is derived from a network of permanent monitoring plots and the second from woody cover assessments from aerial photography. The differences between the studies are reviewed and include sample density, spatial scale and design. Further analyses targeting potential biases in the indirect aerial photography technique are conducted including a comparison of basal area estimates derived from 28 permanent monitoring sites with basal area estimates derived by the aerial photography technique. It is concluded that the effect of photo-scale; or the failure to include appropriate back-transformation of biomass estimates in the aerial photography study are not likely to have contributed significantly to the discrepancy. However, temporal changes in the structure of woodlands, for example, woodlands maturing from many smaller trees to fewer larger trees or seasonal changes, which affect the relationship between cover and basal area could impact on the detection of trends using the aerial photography technique. It is also possible that issues concerning photo-quality may bias assessments through time, and that the limited sample of the permanent monitoring network may inadequately represent change at regional scales.     

A quantitative evaluation of monitoring networks for region-specific nitrate reduction policies

Modern environmental policy implementation in many developed countries is increasingly regionally oriented. Regional governments have undertaken measures designed for the specific needs of the region but, so far, the resulting change in environmental quality has hardly been monitored. This study develops a method for the quantitative evaluation of region-specific environmental policy. Two forms of monitoring are distinguished: monitoring of performance of policy measures and monitoring of environmental quality. This study investigates two regions in the Netherlands where region-specific policy has been implemented. Performance monitoring focused on reduction of nitrate leaching to groundwater. Leaching reductions were calculated with simple models or rules of thumb. Quality monitoring focused on nitrate concentrations under grassland and other land use. An analysis of variance showed significant differences between different soil units. Stratified kriging was applied for spatial interpolation, showing within-unit variability. The current monitoring network could be improved by using a more regular type of grid.     

区域土地利用方式变化对土壤性质影响研究——以土壤钾为例

选取经济发达区原锡山市作为研究区域,以1982年和2005年土壤速效钾作为研究对象,分析探讨不同时空状态下区域土地利用变化对土壤钾含量的影响,以期获得较长时空条件下土壤质量对土地利用变化的响应。20年来,通过统计分析和独立样本t检验表明,原锡山市土壤速效钾含量增加了36.9mg/kg。变异函数分析表明,土壤速效钾的变异函数理论模型呈指数型,块金方差与基台值的比值、自相关阀值发生较大变化。kriging插值分析结果表明,土壤速效钾空间分布较简单,但空间变异明显。     

Land Evaluation for Maize Based on Fuzzy Set and Interpolation

The objective of this article is to apply fuzzy set and interpolation techniques for land suitability evaluation for maize in Northern Ghana. Land suitability indices were computed at point observations using the Semantic Import (SI) model, whereas spatial interpolation was carried out by block kriging. Interpolated land suitability shows a high correlation (R2 = 0.87) with observed maize yield at the village level. This indicates that land suitability is closely related to maize yield in the study area. Membership functions were further used to assess the degree of limitation of land characteristics to maize. Sixty percent of the data has membership functions ranging from 0.23 for ECEC to 1.00 for drainage. ECEC, organic C, and clay are the major constraints to maize yield. The use of the fuzzy technique is helpful for land suitability evaluation, especially in applications in which subtle differences in soil quality are of a major interest. Furthermore, the use of kriging that exploits spatial variability of data is useful in producing continuous land suitability maps and in estimating uncertainties associated with predicted land suitability indices.     

Spatial Variability of Groundwater Depth and Quality Parameters in the National Capital Territory of Delhi

The groundwater quantity and quality scenario is of much concern in the National Capital Territory of Delhi, India, which necessitates an investigation to envisage the extent of spatial variability of groundwater depth and pollutant concentration levels in this region. Therefore, in this study, an effort was made to generate the spatial variability map of groundwater depth and quality parameters (viz. chloride, electrical conductivity, fluoride, magnesium, and nitrate). Ordinary kriging was used to analyze the spatial variability of groundwater depth and quality parameters, whereas indicator kriging was used to analyze groundwater quality parameters equal to or greater than the pollution threshold values. It was observed that the semivariogram parameters fitted well in the exponential model for water depth and in the spherical model for water quality parameters. The generated spatial variability maps indicated that in 43% of the study area, groundwater depth was within 20 m. The salinity level was higher than 2.5 dS m⁻¹ in 69% of the study area and the nitrate concentration exceeded 45 mg l⁻¹ in 36% of the area. The probability maps showed that about 24% of the area had the highest probability (0.8–1.0) of exceedence of the threshold electrical conductivity value and an area of 2% exhibited the highest probability of exceedence of the threshold value of nitrate concentration in the groundwater. The generated spatial variability and probability maps will assist water resource managers and policymakers in development of guidelines in judicious management of groundwater resources for agricultural and drinking purposes in the study area.     

HYDROLOGIC PARAMETERIZATION OF WATERSHEDS FOR RUNOFF PREDICTION USING SWAT1

ABSTRACT: The use of continuous time, distributed parameter hydrologic models like SWAT (Soil and Water Assessment Tool) has opened several opportunities to improve watershed modeling accuracy. However, it has also placed a heavy burden on users with respect to the amount of work involved in parameterizing the watershed in general and in adequately representing the spatial variability of the watershed in particular. Recent developments in Geographical Information Systems (GIS) have alleviated some of the difficulties associated with managing spatial data. However, the user must still choose among various parameterization approaches that are available within the model. This paper describes the important parameterization issues involved when modeling watershed hydrology for runoff prediction using SWAT with emphasis on how to improve model performance without resorting to tedious and arbitrary parameter by parameter calibration. Synthetic and actual watersheds in Indiana and Mississippi were used to illustrate the sensitivity of runoff prediction to spatial variability, watershed decomposition, and spatial and temporal adjustment of curve numbers and return flow contribution. SWAT was also used to predict stream runoff from actual watersheds in Indiana that have extensive subsurface drainage. The results of this study provide useful information for improving SWAT performance in terms of stream runoff prediction in a manner that is particularly useful for modeling ungaged watersheds wherein observed data for calibration is not available.     

Site-specific phosphorus application based on the kriging fertilizer-phosphorus availability index of soils

Site-specific phosphorus management is done to optimize crop production and minimize P loss from soils. The spatial variability of the available P prior to fertilizer application and the P-fixation tendency of soil both need to be taken into account for variable-rate P application. The objectives of this research were to document the spatial variability of the fertilizer-P availability index, which shows the P-fixation tendency, and to develop a strategy that takes the spatial distribution of this index into account for site-specific phosphorus application. In this study, the spatial patterns of the fertilizer-P availability index were characterized by using geostatistics. The ordinary kriging was used for spatial interpolation of the fertilizer-P availability index. Because the fertilizer-P availability index of soil is related to oxalate-extractable Fe and Al and because measuring oxalate-extractable Fe and Al is much easier than directly determining the fertilizer-P availability index, the spatial distribution of the fertilizer-P availability index can be obtained using the oxalate-extractable Fe and Al data. The spatial distribution of Olsen-extractable P, which was used to measure the available-P status prior to fertilizer-P application, was also estimated by using ordinary kriging. The required fertilizer-P amounts were then determined using the kriging estimates of the fertilizer-P availability index and Olsen-extractable P. A fertilizer-P recommendation map for the 430-ha study site in Changhua county, Taiwan was generated by using this approach for illustration. The proposed method for generating fertilizer-P recommendation maps can be used for variable-rate application to maintain an adequate P status for crop production and to potentially reduce the P loss from soils.     

Geostatistical analysis and risk assessment on soil total nitrogen and total soil phosphorus in the Dongting Lake plain area, China

Nonpoint-source pollution and water body eutrophication have become increasing concerns for scientists and policymakers. Nitrogen and phosphorus affect environmental pollution, especially lake eutrophication. To assess the environmental risk of soil total nitrogen (TN) and total phosphorus (TP) pollution, a typical ecological unit of Dongting Lake plain was selected as the experimental site. To verify the stationary of the data, a moving windows technique was adopted. Our results showed that Box-Cox transformation achieved normality in the data set and dampened the effect of outliers. The best theoretical model for semivariogram of TN and TP was a spherical model. The ordinary kriging estimates of TN and TP concentrations were mapped. The integrative comparisons of semivariogram parameters with different trends to the kriging prediction errors of TN and TP indicated that the two-order trend is preferable. Kriging SDs provided valuable information that will increase the accuracy of TN and TP mapping. The probability kriging method is useful to assess the risk of N and P pollution by providing the conditional probability of N and P concentrations exceeding the threshold concentrations of 3.2 and 1.05 g/kg, respectively. The probability distribution of TN and TP at different levels will be helpful to conduct risk assessment, optimize fertilization, and control the pollution of N and P.     

Optimal spatial resolution for collection of ground data and multi-sensor image mapping of a soil erosion cover factor

Military training activities disturb ground and vegetation cover of landscapes and increases potential soil erosion. To monitor the dynamics of soil erosion, there is an important need for an optimal sampling design in which determining the optimal spatial resolutions in terms of size of sample plots used for the collection of ground data and the size of pixels for mapping. Given a sample size, an optimal spatial resolution should be cost-efficient in both sampling costs and map accuracy. This study presents a spatial variability-based method for that purpose and compared it with the traditional methods in a study area in which a soil erosion cover factor was sampled and mapped with multiple plot sizes and multi-sensor images. The results showed that the optimal spatial resolutions obtained using the spatial variability-based method were 12 and 20m for years 1999 and 2000, respectively, and were consistent with those using the traditional methods. Moreover, the most appropriate spatial resolutions using the high-resolution images were also consistent with those using ground sample data, which provides a potential to use the high-resolution images instead of ground data to determine the optimal spatial resolutions before sampling. The most appropriate spatial resolutions above were then verified in terms of cost-efficiency which was defined as the product of sampling cost and map error using ordinary kriging without images and sequential Gaussian co-simulation with images to generate maps.     

NATURAL EROSION RATES AND THEIR PREDICTION IN THE IDAHO BATHOLITH1

ABSTRACT: Natural rates of surface erosion on forested granitic soils in central Idaho were measured in 40 m² bordered erosion plots over a period of four years. In addition, we measured a variety of site variables, soil properties, and summer rainstorm intensities in order to relate erosion rates to site attributes. Median winter erosion rates are approximately twice summer period rates, however mean summer rates are nearly twice winter rates because of infrequent high erosion caused by summer rainstorms. Regression equation models and regression tree models were constructed to explore relationships between erosion and factors that control erosion rates. Ground cover is the single factor that has the greatest influence on erosion rates during both summer and winter periods. Rainstorm intensity (erosivity index) strongly influences summer erosion rates, even on soils with high ground cover percentages. Few summer storms were of sufficient duration and intensity to cause rilling on the plots, and the data set was too small to elucidate differences in rill vs. interrill erosion. The regression tree models are relatively less biased than the regression equations developed, and explained 70 and 84 percent of the variability in summer and winter erosion rates, respectively.     

Using rank-order geostatistics for spatial interpolation of highly skewed data in a heavy-metal contaminated site

The spatial distribution of a pollutant in contaminated soils is usually highly skewed. As a result, the sample variogram often differs considerably from its regional counterpart and the geostatistical interpolation is hindered. In this study, rank-order geostatistics with standardized rank transformation was used for the spatial interpolation of pollutants with a highly skewed distribution in contaminated soils when commonly used nonlinear methods, such as logarithmic and normal-scored transformations, are not suitable. A real data set of soil Cd concentrations with great variation and high skewness in a contaminated site of Taiwan was used for illustration. The spatial dependence of ranks transformed from Cd concentrations was identified and kriging estimation was readily performed in the standardized-rank space. The estimated standardized rank was back-transformed into the concentration space using the middle point model within a standardized-rank interval of the empirical distribution function (EDF). The spatial distribution of Cd concentrations was then obtained. The probability of Cd concentration being higher than a given cutoff value also can be estimated by using the estimated distribution of standardized ranks. The contour maps of Cd concentrations and the probabilities of Cd concentrations being higher than the cutoff value can be simultaneously used for delineation of hazardous areas of contaminated soils.     

Evaluation of a Regional Monitoring Program’s Statistical Power to Detect Temporal Trends in Forest Health Indicators

Forests are socioeconomically and ecologically important ecosystems that are exposed to a variety of natural and anthropogenic stressors. As such, monitoring forest condition and detecting temporal changes therein remain critical to sound public and private forestland management. The National Parks Service’s Vital Signs monitoring program collects information on many forest health indicators, including species richness, cover by exotics, browse pressure, and forest regeneration. We applied a mixed-model approach to partition variability in data for 30 forest health indicators collected from several national parks in the eastern United States. We then used the estimated variance components in a simulation model to evaluate trend detection capabilities for each indicator. We investigated the extent to which the following factors affected ability to detect trends: (a) sample design: using simple panel versus connected panel design, (b) effect size: increasing trend magnitude, (c) sample size: varying the number of plots sampled each year, and (d) stratified sampling: post-stratifying plots into vegetation domains. Statistical power varied among indicators; however, indicators that measured the proportion of a total yielded higher power when compared to indicators that measured absolute or average values. In addition, the total variability for an indicator appeared to influence power to detect temporal trends more than how total variance was partitioned among spatial and temporal sources. Based on these analyses and the monitoring objectives of the Vital Signs program, the current sampling design is likely overly intensive for detecting a 5 % trend·year^?1 for all indicators and is appropriate for detecting a 1 % trend·year^?1 in most indicators.     

SOURCES OF VARIABILITY IN CONDUCTING PEBBLE COUNTS: THEIR POTENTIAL INFLUENCE ON THE RESULTS OF STREAM MONITORING PROGRAMS1

Pebble counts have been used for a variety of monitoring projects and are an important component of stream evaluation efforts throughout the United States. The utility of pebble counts as a monitoring tool is, however, based on the monitoring objectives and the assumption that data are collected with sufficient precision to meet those objectives. Depending upon the objective, sources of variability that can limit the precision of pebble count data include substrate heterogeneity at a site, differences in substrate among sample locations within a stream reach, substrate variability among streams, differences in when the substrate sample is collected, differences in how and where technicians pick up substrate particles, and how consistently technicians measure the intermediate axis of a selected particle. This study found that each of these sources of variability is of sufficient magnitude to affect results of monitoring projects. Therefore, actions such as observer training, increasing the number of pebbles measured, evaluating several riffles within a reach, evaluating permanent sites, and narrowing the time window during which pebble counts are conducted should be considered in order to minimize variability. The failure to account for sources of variability associated with pebble counts within the study design may result in failing to meet monitoring objectives.     

Sampling and Mapping a Soil Erosion Cover Factor by Integrating Stratification,Model Updating and Cokriging with Images

Cost-efficient sample designs for collection of ground data and accurate mapping of variables are required to monitor natural resources and environmental and ecological systems. In this study, a sample design and mapping method was developed by integrating stratification, model updating, and cokriging with Landsat Thematic Mapper (TM) imagery. This method is based on the spatial autocorrelation of variables and the spatial cross-correlation among them. It can lead to sample designs with variable grid spacing, where sampling distances between plots vary depending on spatial variability of the variables from location to location. This has potential cost-efficiencies in terms of sample design and mapping. This method is also applicable for mapping in the case in which no ground data can be collected in some parts of a study area because of the high cost. The method was validated in a case study in which a ground and vegetation cover factor was sampled and mapped for monitoring soil erosion. The results showed that when the sample obtained with three strata using the developed method was used for sampling and mapping the cover factor, the sampling cost was greatly decreased, although the error of the map was slightly increased compared to that without stratification; that is, the sample cost-efficiency quantified by the product of cost and error was greatly increased. The increase of cost-efficiency was more obvious when the cover factor values of the plots within the no-significant-change stratum were updated by a model developed using the previous observations instead of remeasuring them in the field.     

MODELING ACTUAL EVAPOTRANSPIRATION FROM FORESTED WATERSHEDS ACROSS THE SOUTHEASTERN UNITED STATES1

ABSTRACT: About 50 to 80 percent of precipitation in the southeastern United States returns to the atmosphere by evapotranspiration. As evapotranspiration is a major component in the forest water balances, accurately quantifying it is critical to predicting the effects of forest management and global change on water, sediment, and nutrient yield from forested watersheds. However, direct measurement of forest evapotranspiration on a large basin or a regional scale is not possible. The objectives of this study were to develop an empirical model to estimate long‐term annual actual evapotranspiration (ART) for forested watersheds and to quantify spatial AET patterns across the southeast. A geographic information system (GIS) database including land cover, daily streamflow, and climate was developed using long term experimental and monitoring data from 39 forested watersheds across the region. Using the stepwise selection method implemented in a statistical modeling package, a long term annual AET model was constructed. The final multivariate linear model includes four independent variables—annual precipitation, watershed latitude, watershed elevation, and percentage of forest coverage. The model has an adjusted R² of 0.794 and is sufficient to predict long term annual ART for forested watersheds across the southeastern United States. The model developed by this study may be used to examine the spatial variability of water availability, estimate annual water loss from mesoscale watersheds, and project potential water yield change due to forest cover change.     

Uncertainty Assessment for Management of Soil Contaminants with Sparse Data

In order for soil resources to be sustainably managed, it is necessary to have reliable, valid data on the spatial distribution of their environmental impact. However, in practice, one often has to cope with spatial interpolation achieved from few data that show a skewed distribution and uncertain information about soil contamination. We present a case study with 76 soil samples taken from a site of 15 square km in order to assess the usability of information gleaned from sparse data. The soil was contaminated with cadmium predominantly as a result of airborne emissions from a metal smelter. The spatial interpolation applies lognormal anisotropic kriging and conditional simulation for log-transformed data. The uncertainty of cadmium concentration acquired through data sampling, sample preparation, analytical measurement, and interpolation is factor 2 within 68.3 % confidence. Uncertainty predominantly results from the spatial interpolation necessitated by low sampling density and spatial heterogeneity. The interpolation data are shown in maps presenting likelihoods of exceeding threshold values as a result of a lognormal probability distribution. Although the results are not deterministic, this procedure yields a quantified and transparent estimation of the contamination, which can be used to delineate areas for soil improvement, remediation, or restricted area use, based on the decision-makers probability safety requirement.     

Multiscale Analysis of Hydrology in a Mountaintop Mine‐Impacted Watershed

In the Appalachian region of the eastern United States, mountaintop removal mining (MTM) is a dominant driver of land‐cover change, impacting 6.8% of the largely forested 4.86 million ha coal fields region. Recent catastrophic flooding and documented biological impairment downstream of MTM has drawn sharp criticism to this practice. Despite its extent, scale, and use since the 1970s, the impact of MTM on hydrology is poorly understood. Therefore, the goal of this study was a multiscale evaluation to establish the nature of hydrologic impacts associated with MTM. To quantify the extent of MTM, land‐cover change over the lifetime of this practice is estimated for a mesoscale watershed in southern West Virginia. To assess hydrologic impacts, we conducted long‐term trend analyses to evaluate for systematic changes in hydrology at the mesoscale, and conducted hydrometric and response time modeling to characterize storm‐scale responses of a MTM‐impacted headwater catchment. Results show a general trend in the conversion of forests to mines, and significant decreases in maximum streamflow and variability, and increases in base‐flow ratio attributed to valley fills and deep mine drainage. Decreases in variability are shown across spatial and temporal scales having important implications for water quantity and quality. However, considerable research is necessary to understand how MTM impacts hydrology. In an effort to inform future research, we identify existing knowledge gaps and limitations of our study.     

USING ARTIFICIAL NEURAL NETWORKS TO ESTIMATE MISSING RAINFALL DATA1

ABSTRACT: Missing rainfall data from a time series or a spatial field of observations can present a serious obstacle to data analysis, modeling studies and operational forecasting in hydrology. Numerous schemes for replacing missing data have been proposed, ranging from simple weighted averages of data points that are nearby in time and space to complex statistically-based interpolation methods and function fitting schemes. This paper presents a technique for replacing missing spatial data using a backpropagation neural network applied to concurrent data from nearby gauges. Tests performed on a sample of gauges in the Middle Atlantic region of the United States show that this technique produces results that compare favorably to simple techniques such as arithmetic and distance-weighted averages of the values from nearby gauges, and also to linear optimization methods such as regression.     

Evaluation of the MIKE SHE Model for Application in the Loess Plateau,China1

Abstract: Quantifying the hydrologic responses to land use/land cover change and climate variability is essential for integrated sustainable watershed management in water limited regions such as the Loess Plateau in Northwestern China where an adaptive watershed management approach is being implemented. Traditional empirical modeling approach to quantifying the accumulated hydrologic effects of watershed management is limited due to its complex nature of soil and water conservation practices (e.g., biological, structural, and agricultural measures) in the region. Therefore, the objective of this study was to evaluate the ability of the distributed hydrologic model, MIKE SHE to simulate basin runoff. Streamflow data measured from an overland flow‐dominant watershed (12 km²) in northwestern China were used for model evaluation. Model calibration and validation suggested that the model could capture the dominant runoff process of the small watershed. We found that the physically based model required calibration at appropriate scales and estimated model parameters were influenced by both temporal and spatial scales of input data. We concluded that the model was useful for understanding the rainfall‐runoff mechanisms. However, more measured data with higher temporal resolution are needed to further test the model for regional applications.