AN ALGORITHM FOR ESTIMATING SURFACE SUSPENDED SEDIMENT CONCENTRATIONS WITH LANDSAT MSS DIGITAL DATA1

ABSTRACT: Algorithms for Landsat MSS digital data are needed to reduce the necessity of calibrating each Landsat scene if these data are to be useful in monitoring programs for surface suspended sediments. In this study digital data were extracted from 16 Landsat Multispectral Scanner (MSS) scenes collected between March 1987 and August 1988 over Enid Reservoir in North Central Mississippi. These data were converted to radiance and reflectance data for comparison with field measurements of surface suspended sediment concentrations. Concentrations ranged from 2 to 168 mg/1 during the study with only four greater than 100 mg/l. Linear and polynomial regression analyses were used to relate the surface suspended sediment concentrations with radiance and reflectance. Reflectance in MSS band 2 (0.6 to 0.7 μm) and MSS band 3 (0.7 to 0.8 μm) were best related to the surface suspended sediment concentrations with coefficients of determination accounting for 71 percent and 68 percent of the variation in the data, respectively. Regressions with radiance data accounted for 36 percent (band 2) or less of the variation. Logarithmic transformations of either reflectance or sediment concentrations increase the coefficients of determination for MSS band 2 reflectance data to 81 percent. Regressions between the ratio of MSS band 1 to MSS band 2 reflectances and concentrations also accounted for 80 percent of the variation. An equation Log_e SS (mg/l) = 9.21R½+ 2.71R½2 + 8.45, where S is surface suspended sediment concentrations and R¹/₂ is the ratio of MSS band 1 to MSS band 2 reflectances, provided the best fit to the data with a coefficient of determination of 0.82. This equation is essentially the same as an algorithm proposed by Topliss et at. (1990), for estimating surface suspended sediment concentrations in Canadian coastal waters. These equations for Enid Reservoir and Canadian waters suggest that it may be possible to develop an algorithm for widespread use for estimating surface suspended sediments.     

THE APPLICATION OF RIDGE REGRESSION ANALYSIS TO A HYDROLOGIC TARGET-CONTROL MODEL1

ABSTRACT: Ridge regression analysis is used to investigate the stability of regression estimates over twenty-three years of data in a target-control model. Two target stations in the Wind River Range in Wyoming are studied using two sets of control variables. The predictive ability of ridge regression analysis is compared to that of ordinary regression analysis. The results of this study indicate improved stability of the estimates of ridge regression over ordinary regression. The predictive ability of the ridge regression estimates is as good or better than regression estimates.     

PREDICTING FECAL COLIFORM BACTERIA LEVELS IN THE CHARLES RIVER,MASSACHUSETTS, USA1

In Massachusetts, the Charles River Watershed Association conducts a regular water quality monitoring and public notification program in the Charles River Basin during the recreational season to inform users of the river's health. This program has relied on laboratory analyses of river samples for fecal coliform bacteria levels, however, results are not available until at least 24 hours after sampling. To avoid the need for laboratory analyses, ordinary least squares (OLS) and logistic regression models were developed to predict fecal coliform bacteria concentrations and the probabilities of exceeding the Massachusetts secondary contact recreation standard for bacteria based on meteorological conditions and streamflow. The OLS models resulted in adjusted R²s ranging from 50 to 60 percent. An uncertainty analysis reveals that of the total variability of fecal coliform bacteria concentrations, 45 percent is explained by the OLS regression model, 15 percent is explained by both measurement and space sampling error, and 40 percent is explained by time sampling error. Higher accuracy in future bacteria forecasting models would likely result from reductions in laboratory measurement errors and improved sampling designs.     

EFFECTIVE USE OF LANDSAT DATA IN HYDROLOGIC MODELS1

ABSTRACT: In a cooperative demonstration project, NASA and the U.S. Army Corps of Engineers (Corps) compared conventional and Landsat-derived land-use data for use in hydrologic models, and the resulting discharge frequency curves were analyzed. When a grid-based data-management system was used on a cell-by-cell basis (size about 1.1 acres or 0.45 hectare), Landsat classification accuracy was only 64 percent, but, when the grid cells were aggregated into watersheds, the classification accuracy increased to about 95 percent. When both conventional and Landsat land-use data were input to the HEC-1 model for generating discharge frequency curves, the differences in calculated discharge were judged insignificant for subbasins as small as 1.0mi² (2.59 km²). For basins larger than 10mi² (25.9km²), use of the Landsat approach is more cost-effective than use of conventional methods. Digital Landsat data can also be used effectively by local and regional agencies for hydrologic analysis by incorporating the data into grid-based data-management systems. The transfer of this new technology is well under way through inclusion in some Corps training courses and through use by both county government personnel and private consultants.     

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.     

Evaluating the Eco‐Geomorphological Condition of Restored Streams Using Visual Assessment and Macroinvertebrate Metrics

The Stream Performance Assessment (SPA), a new rapid assessment method, was applied to 93 restored, 21 impaired, 29 reference, and 13 reference streams with some incision throughout North Carolina. Principal component analysis (PCA) indicated restored streams align more closely with reference streams rather than impaired streams. Further, PCA‐based factor analysis revealed restored streams were similar to reference streams in terms of morphologic condition, but exhibited a greater range of scores relative to aquatic habitat and bedform. Macroinvertebrate sampling and GIS watershed analyses were conducted on 84 restored streams. SPA and watershed data were compared to Ephemeroptera, Plecoptera, and Trichoptera (EPT) taxa to determine which factors indicate stream health. SPA and watershed factors were used in least squares, ridge, and principal component regression (PCR) to develop a prediction model for EPT taxa. All three methods produced reasonable predictions for EPT taxa. Cross‐validation indicated ridge regression resulted in the lowest prediction error. The ridge model was then used to predict EPT taxa numbers for 21 impaired and 25 reference streams in addition to the 84 restored streams. Statistical comparisons of the predicted scores indicated urban streams (>10% impervious watershed cover) have lower expected numbers of EPT taxa. Rural restored streams have macroinvertebrate metric scores similar to those predicted for rural reference streams.     

Comparison of Macroinvertebrate‐Derived Stream Quality Metrics Between Snag and Riffle Habitats1

Abstract: We compared benthic macroinvertebrate assemblage structure at snag and riffle habitats in 43 Wisconsin streams across a range of watershed urbanization using a variety of stream quality metrics. Discriminant analysis indicated that dominant taxa at riffles and snags differed; Hydropsychid caddisflies (Hydropsyche betteni and Cheumatopsyche spp.) and elmid beetles (Optioservus spp. and Stenemlis spp.) typified riffles, whereas isopods (Asellus intermedius) and amphipods (Hyalella azteca and Gammarus pseudolimnaeus) predominated in snags. Analysis of covariance indicated that samples from snag and riffle habitats differed significantly in their response to the urbanization gradient for the Hilsenhoff biotic index (BI), Shannon’s diversity index, and percent of filterers, shredders, and pollution intolerant Ephemeroptera, Plecoptera, and Trichoptera (EPT) at each stream site (p ≤ 0.10). These differences suggest that although macroinvertebrate assemblages present in either habitat type are sensitive to detecting the effects of urbanization, metrics derived from different habitats should not be intermixed when assessing stream quality through biomonitoring. This can be a limitation to resource managers who wish to compare water quality among streams where the same habitat type is not available at all stream locations, or where a specific habitat type (i.e., a riffle) is required to determine a metric value (i.e., BI). To account for differences in stream quality at sites lacking riffle habitat, snag‐derived metric values can be adjusted based on those obtained from riffles that have been exposed to the same level of urbanization. Comparison of nonlinear regression equations that related stream quality metric values from the two habitat types to percent watershed urbanization indicated that snag habitats had on average 30.2 fewer percent EPT individuals, a lower diversity index value than riffles, and a BI value of 0.29 greater than riffles.     

ESTIMATING CENTRAL TENDENCY FROM UNCENSORED TRACE LEVEL MEASUREMENTS1

ABSTRACT: The statistical analysis of data which have trace level measurements has traditionally been a two-step process in which data are first censored using criteria based on measurement precision, and then analyzed with statistical methods for censored data. The process might be more informative if data were left uncensored. In this paper, information loss attributable to censoring and measurement noise are assessed by comparing the sample mean and median of uncensored measurements with a log regression mean and median based on censored data. Measurements are derived from lognormal parent distributions which have random variability characteristic of trace level measurement. The relative performance of estimators used with error-free samples and with samples having measurement noise can be explained by differences between the probability distributions of parents and measurements. Measurement introduces bias and dispersion and transforms lognormal parent distributions toward greater symmetry. Estimates using uncensored data are less biased and more accurate than the log regression mean and median when censoring exceeds about 50 percent, and are not much worse at any fraction censored. For data with many (80 percent) results below the limit of detection, bias may be quite severe.     

STATISTICAL LOW FLOW ESTIMATION USING GIS ANALYSIS IN HUMID MONTANE REGIONS IN PUERTO RICO1

ABSTRACT: Statistical analysis of watershed parameters derived using a Geographical Information system (GIS) was done to develop equations for estimating the 7d–10yr, 30d–10yr, and 7d–2yr low flow for watersheds in humid montane regions of Puerto Rico. Digital elevation models and land use, geology, soils, and stream network coverages were used to evaluate 21 geomorphic, 10 stream channel, 9 relief, 7 geology, 4 climate, and 2 soil parameters for each watershed. To assess which parameters should be used for further investigation, a correlation analysis was used to determine the independence and collinearity among these parameters and their relationship with low flows. Multiple regression analyses using the selected parameters were then performed to develop the statistical models of low flows. The final models were selected in the basis of the Mallow Cp statistic, the adjusted R², the Press statistic, the degree of collinearity, and an analysis of the residuals. In the final models, drainage density, the ratio of length of tributaries to the length of the main channel, the percent of drainage area with northeast aspect, and the average weighted slope of the drainage were the most significant parameters. The final models had adjusted standard errors of 58.7 percent, 59.2 percent, and 48.6 percent for the 7d–10yr, 30d–10yr, and 7d–2yr low flows respectively. For comparison, the best model based on watershed parameters that can be easily measured without a GIS had an adjusted standard error of 82.8 percent.     

GENERALIZED LOW-FLOW FREQUENCY RELATIONSHIPS FOR UNGAGED SITES IN MASSACHUSETTS1

ABSTRACT: Regional hydrologic procedures such as generalized least squares regression and streamflow record augmentation have been advocated for obtaining estimates of both flood-flow and low-flow statistics at ungaged sites. While such procedures are extremely useful in regional flood-flow studies, no evaluation of their merit in regional low-flow estimation has been made using actual streamflow data. This study develops generalized regional regression equations for estimating the d-day, T-year low-flow discharge, Q_{d, t}, at ungaged sites in Massachusetts where d = 3, 7, 14, and 30 days. A two-parameter lognormal distribution is fit to sequences of annual minimum d-day low-flows and the estimated parameters of the lognormal distribution are then related to two drainage basin characteristics: drainage area and relief. The resulting models are general, simple to use, and about as precise as most previous models that only provide estimates of a single statistic such as Q_7,10. Comparisons are provided of the impact of using ordinary least squares regression, generalized least squares regression, and streamflow record augmentation procedures to fit regional low-flow frequency models in Massachusetts.     

Effects of Urban Spatial Structure,Sociodemographics, and Climate on Residential Water Consumption in Hillsboro,Oregon1

House-Peters, Lily, Bethany Pratt, and Heejun Chang, 2010. Effects of Urban Spatial Structure, Sociodemographics, and Climate on Residential Water Consumption in Hillsboro, Oregon. Journal of the American Water Resources Association (JAWRA) 46(3):461-472. DOI: 10.1111/j.1752-1688.2009.00415.x Abstract: In the Portland metropolitan area, suburban growth in cities such as Hillsboro is projected to increase as people seek affordable housing near a burgeoning metropolis. The most significant determinants for increases in water demand are population growth, climate change, and the type of urban development that occurs. This study analyzes the spatial patterns of single family residential (SFR) water consumption in Hillsboro, Oregon, at the census block scale. The following research questions are addressed: (1) What are the significant determinants of SFR water consumption in Hillsboro, Oregon? (2) Is SFR water demand sensitive to drought conditions and interannual climate variation? (3) To what magnitude do particular census blocks react to drought conditions and interannual climate variation? Using ordinary least squares multiple regression and spatial regression methods, we found that base use, representing indoor water use, is dependent on household size and that seasonal use, representing external water use is dependent on both education level and the size of the property’s outdoor space. Spatial analysis techniques determined that although the water demand of the study area as a whole is not sensitive to drought conditions, certain individual census blocks do respond with a higher magnitude of water use. The most climate-sensitive census blocks tend to contain newer and larger homes, and have higher property values and more affluent and well-educated residents.     

PERFORMANCE VARIATIONS OF COD AND NITROGEN REMOVAL BY VEGETATED SUBMERGED BED WETLANDS1

ABSTRACT: Vegetated submerged bed wetlands can supplement treatment of onsite wastewater systems. This study evaluated vegetation, media, and seasonal impacts on system performance in six meso scale rock plant filters with and without narrow leaf cattails (Typha augustifolia). Daily chemical oxygen demand (COD) reductions in planted cells averaged 85 percent and weekly total nitrogen (TN) reductions averaged 50 percent. Planted cells had 17 percent greater COD reduction and 76 percent greater TN reduction than unplanted cells, both significant differences. Media type affected COD reduction, particularly in unplanted cells. COD treatment in planted cells was highest for fine crushed limestone (87±13 percent) and least variable for coarse river gravel (85±11 percent). No significant difference in TN reduction was observed for different media types (48 to 51 percent range). Seasonal influences on treatment included a significant decrease during late fall and early spring and a significant increase with temperature. After a step increase in organic loading, treatment efficiency decreased sharply but returned to prior levels after an adaptation period of about one month. Planted cells not only exhibited higher treatment efficiency but also had a retarded organic matter breakthrough, appearing after three to seven times the period for a bromide tracer. This supports a hypothesis that retardation of contaminant movement through the treatment cells results in enhanced removal. These results support the use of rock plant filters, but demonstrate the need to account for performance variations in system design. (KEY TERMS: constructed wetlands; seasonal effects; subsurface flow; Typha augustifolia; onsite wastewater treatment; water quality.)     

A 396-YEAR RECONSTRUCTION OF PRECIPITATION IN SOUTHERN JORDAN1

ABSTRACT: Water resources are the lifeblood of the Near East region. Careful planning and management of water resources in dry land regions requires information on the likelihood of extreme events, especially prolonged drought. It is essential to understand the variability of climate on time scales of decades to centuries to assign reasonable probabilities to such events. Tree-ring analysis is one way to increase our knowledge of the climate variability beyond the short period covered by the instrumental data. In this paper, we reconstruct October-May precipitation from a Juniperus phoenicia tree-ring chronology in southern Jordan to gain a long-term (A.D. 1600–1995) perspective on runs of dry years and on time series fluctuations in precipitation averaged over several years. The reconstruction equation derived by regression of log-transformed precipitation on tree-ring indices explains 44 percent of the variance of observed precipitation. The longest reconstructed drought, as defined by consecutive years below a threshold of 217.4 mm, was four years, compared with three years for the 1946–95 instrumental data. A Monte Carlo analysis designed to account for uncertainty in the reconstruction indicates a lower than 50 percent chance that the region has experienced drought longer than five years in the past 400 years.     

CONTROLLING PHOSPHORUS IN RUNOFF FROM LONG TERM DAIRY WASTE APPLICATION FIELDS1

ABSTRACT: Phosphorus (P) in runoff from long term animal waste application fields can contribute to accelerated eutrophication of surface waters. Manure when applied at nitrogen (N) agronomic rates generally increases soil P concentrations, which can increase runoff of soluble P. Along the North Bosque River in central Texas, dairy waste application fields are identified as the most controllable nonpoint source of soluble P in a total maximum daily load. To evaluate P reduction practices for fields high in soil extractable P, edge‐of‐field runoff was measured from paired plots of Coastal bermudagrass (Cynodon dactylon) and sorghum (Sorghum bicolor)/ winter wheat (Triticum spp.). Plots (about 0.4 ha) received manure at P agronomic rates following Texas permit guidelines and commercial N during the pretreatment period. During the post‐treatment period, control plots continued to receive manure at P agronomic rates and commercial N. Treatment plots received only commercial N during the post‐treatment period. Use of only commercial N on soils with high extractable P levels significantly decreased P loadings in edge‐of‐field runoff by at least 40 percent, but runoff concentrations sometimes increased. No notable changes in extractable soil P concentrations were observed after five years of monitoring due to drought conditions limiting forage uptake and removal.     

STRATIFICATION OF VARIABILITY IN RUNOFF AND SEDIMENT YIELD BASED ON VEGETATION CHARACTERISTICS1

ABSTRACT: Runoff and sediment yield were collected from 100 plots during simulated rainfalls (100 mm/hr for 15 minutes) at antecedent soil moisture conditions. A clustering technique was used to stratify the variability of a single data set within a sagebrush‐grass community into four groups based on vegetation life form and amount of cover. The four cluster groups were grass, grass/shrub, shrub, and forb/grass and were found to be significantly different in plant height, surface roughness, soil bulk density, and soil organic matter. Stepwise multiple regression analyses were performed on the single data set and each cluster group. Results for individual groups resulted in more robust predictive equations for runoff (r²= 0.65–0.73) and sediment yield (r²= 0.37–0.91) than for equations developed from the single data set (r²= 0.56 for runoff and r²= 0.27 for sediment yield). The standard errors of the cluster group regression equations were also improved in three of the four group equations for both runoff and sediment yield compared to the single data set. Runoff was found to be significantly less (p >0.01) in the forb/grass group compared with other vegetation cluster groups, but this was influenced by four plots that produced little or no runoff. Sediment yield was not found to be significantly different among any cluster groups. Discriminant analysis was then used to identify important variables and develop a model to classify plots into one of the four cluster groups. The discriminant model could be incorporated into rangeland hydrology and erosion models. The percentage cover of grasses, shrubs, litter, and bare ground effectively stratified about 12 percent of the variation observed in runoff and 26 percent of the variability for sediment yield as determined by r².     

DIFFERENCES IN WARM-SEASON,RAINSTORM-GENERATED STORMFLOWS FOR NORTHEASTERN ILLINOIS URBANIZED BASINS1

ABSTRACT: Precipitation, streamflow, and population data were analyzed over the 1941–1990 period to determine whether changes in stormflows and net (post. minus pre-rainstorm) stormflows, associated with warm-season large rainstorms, were similar for two urbanized northeastern Illinois basins. Warm season large rainstorms were defined as April through October rainfall events in which ? 5.1 cm occurred in a 48-hour period over the basin. To minimize differences associated with varying large rainstorm amounts over time, the net sthrmflow for each event was divided by the large rainstorm amount. This ratio, Ui, indicated that the two urbanized basins experienced significant, yet different, increases (102 percent and 49 percent) in flow amount per centimeter of rainfall from 1941–1965 to 1966–1990. Results of a regression analysis between Ui and population showed that the increase in U_i per 100,000 increase in population ranged from 0.59 to 0.67 m³s^-1 per cm of rainfall for the two basins. These results demonstrate the varying degree of change that urban planners can expect in stormflows associated with large warm season rainstorms for areas undergoing urbanization.     

Threatened biotas: "Hot spots" in tropical forests

Summary The mass-extinction episode underway is largely centred on tropical forests, insofar as they contain at least half of all Earth's species and they are being depleted faster than any other biome. But species distributions and depletion patterns are anything but uniform throughout the biome. This paper identifies 10 areas that, a) are characterised by exceptional concentrations of species with high levels of endemism and b) are experiencing unusually rapid rates of depletion. While these "hotspot" areas comprise less than 3.5 percent of remaining primary forests, they harbour over 34 000 endemic plant species (27 percent of all plant species in tropical forests and 13 percent of all plant species worldwide). They also feature 700 000 endemic animal species and possibly several times more. Unfortunately, they appear likely to lose 90 percent of their forest cover as soon as the end of the century or shortly thereafter, causing the extinction of almost 7 percent of Earth's plant species and at least a similar proportion of animal species, this occurring in only 0.2 percent of Earth's land surface. By concentrating on such areas where needs are greatestand where the pay-off from safeguard measures would also be greatest, conservationists can engage in a more systematised response to the challenge of largescale extinctions impending in tropical forests.Professor Norman Myers is a consultant in environment and development. He is author of many important environmental books, an Advisory Board Member, and a regular contributor toThe Environmentalist.     

Relative Influence of Streamflows in Assessing Temporal Variability in Stream Habitat1

Abstract: The effects of streamflows on temporal variation in stream habitat were analyzed from the data collected 6‐11 years apart at 38 sites across the United States. Multiple linear regression was used to assess the variation in habitat caused by streamflow at the time of sampling and high flows between sampling. In addition to flow variables, the model also contained geomorphic and land use factors. The regression model was statistically significant (p < 0.05; R² = 0.31‐0.46) for 5 of 14 habitat variables: mean wetted stream depth, mean bankfull depth, mean wetted stream width, coefficient of variation of wetted stream width, and the percent frequency of bank erosion. High flows between samples accounted for about 16% of the total variation in the frequency of bank erosion. Streamflow at the time of sampling was the main source of variation in mean stream depth and contributed to the variation in mean stream width and the frequency of bank erosion. Urban land use (population change) accounted for over 20% of the total variation in mean bankfull depth, 15% of the total variation in the coefficient of variation of stream width, and about 10% of the variation in mean stream width.     

Mapping soil textural fractions across a large watershed in north-east Florida

Assessment of regional scale soil spatial variation and mapping their distribution is constrained by sparse data which are collected using field surveys that are labor intensive and cost prohibitive. We explored geostatistical (ordinary kriging-OK), regression (Regression Tree-RT), and hybrid methods (RT plus residual Sequential Gaussian Simulation-SGS) to map soil textural fractions across the Santa Fe River Watershed (3585 km²) in north-east Florida. Soil samples collected from four depths (L1: 0–30 cm, L2: 30–60 cm, L3: 60–120 cm, and L4: 120–180 cm) at 141 locations were analyzed for soil textural fractions (sand, silt and clay contents), and combined with textural data (15 profiles) assembled under the Florida Soil Characterization program. Textural fractions in L1 and L2 were autocorrelated, and spatially mapped across the watershed. OK performance was poor, which may be attributed to the sparse sampling. RT model structure varied among textural fractions, and the model explained variations ranged from 25% for L1 silt to 61% for L2 clay content. Regression residuals were simulated using SGS, and the average of simulated residuals were used to approximate regression residual distribution map, which were added to regression trend maps. Independent validation of the prediction maps showed that regression models performed slightly better than OK, and regression combined with average of simulated regression residuals improved predictions beyond the regression model. Sand content >90% in both 0–30 and 30–60 cm covered 80.6% of the watershed area.