1998 Spotlight on Graduate and Undergraduate Research at ESF
Environmental Resource and Forest Engineering Abstracts

METHODS OF ACCURACY ASSESSMENT FOR SPATIALLY EXPLICIT MODELS.  Jennifer L. Barber, William M. Stiteler, and Paul F. Hopkins.

ENGINEERING ANALYSIS OF GALLEY SYSTEMS AS A SUBSURFACE WASTEWATER DISPOSAL OPTION.  Kelly A. Bazukiewicz and James M. Hassett.

ASSESSING PRECISION OF A TOPOGRAPHIC SURVEY.  Amy Becker and Soojeong Myeong.

PRODUCING A TOPOGRAPHIC MAP.  Amy Becker, Kevin Riordan and Jessica Ross.

PHYTOREMEDIATION OF LEAD CONTAMINATED SOIL IN RESIDENTIAL AREAS.  Marie Bowser, Douglas Daley and James M. Hassett.

AN ECONOMIC ANALYSIS OF DISINFECTION ALTERNATIVES AT THE ITHACA WASTEWATER TREATMENT PLANT:  CHLORINATION VS. ULTRAVIOLET LIGHT.  Nate Cheal, and Chuck Kroll, Faculty of Environmental Resources and Forest Engineering, Bray Hall, SUNY College of Environmental Science and Forestry, Syracuse, NY 13210.

INFRASTRUCTURE REHABILITATION AT CAMP GOODWILL, CHITTENANGO, NY.  Nathan Cheal, Kim Heller, Jeff Ryan, Janine Sleeper, and James M. Hassett.

AN ANALYSIS OF DISTRIBUTIONAL FIT TO LOW STREAMFLOW SERIES THROUGHOUT THE UNITED STATES.  Michael Conway and Chuck Kroll.

WHY IS DRY LAKE SO WET?? SEDIMENTATION PROBLEMS AT DRY LAKE, CLARK RESERVATION STATE PARK, JAMESVILLE, NY. Christopher Craner, Ronald Ferris, Daniel Gilbert, Thomas Parmiter, Trevor Thomas and James M. Hassett.

COMPARISON OF MANNING’S EQUATION TO REGRESSION EQUATIONS FOR FLOW ESTIMATES IN OPEN CHANNELS. Andrew Gerardi, Matthew Stokes, Bradley Kubiak and James M. Hassett.

A NEW PROBABILISTIC METHODOLOGY FOR SELECTING REMEDIATION GOALS: A CASE STUDY OF A LEAD RISK ASSESSMENT FOR A SUPERFUNDSITE IN MURRAY, UT. Philip Goodrum, James M. Hassett, Gary Diamond, and Susan Griffin.

STREAM POWER AND FROUDE NUMBER: NEW WAYS TO DESCRIBE FISH-FLOW RELATIONSHIPS.   Kevin Hartman and James M. Hassett.

APPLICATIONS OF RESIDUAL BIO-TECH SOIL (RBSÔ )IN VEGETATED BIOFILTRATION SYSTEMS.  Brian A. Jerose, Aaron D. Richardson, and Richard J. McClimans.

IMPROVING ACCESS AND USE OF SPATIAL DATA VIA THE INTERNET AND GEOGRAPHIC INFORMATION SYSTEMS: CASE PROJECT - USE OF THE ENVIRONMENTAL PROTECTION AGENCY'S INDEX OF WATERSHED INDICATORS FOR ACCESS AND USE OF WATER QUALITY DATA FOR WATERSHEDS. Michael Plastino and Paul Hopkins.

AN INVESTIGATION INTO THE EFFECT OF SPATIAL DISTRIBUITION OF SPECIES IN MIXED FOREST ON THE SPECTRAL RESPONSE OF TM IMAGERY. Lester G. Power and Paul F. Hopkins.

DEMYSTIFYING KAPPA; A STATISTICAL MEASURE OF MAP ACCURACY. Lester G. Power and David L. Szymanski.

MULTIMODAL MEDICAL IMAGE REGISTRATION.  Lindi Quackenbush and Paul Hopkins..

COMPUTER SIMULATION OF HYDROGEN SULFIDE REMOVAL FROM DIGESTOR OFF-GASSES BY USE OF A FIXED FILM BIOSCRUBBER.   Jan Salzman, Kirsten Kenty, James M. Hassett and Siddharth G. Chatterjee.

THE ESTIMATION AND USAGE OF BASEFLOW RECESSION CONSTANTS IN LOW STREAMFLOW MODELING. Frank Sidari and Charles N. Kroll.

DEVELOPING AN INTERNET-ACCESIBLE MODELING PACKAGE. William M. Stiteler, Jennifer L. Barber, and Paul F. Hopkins.

GIS DESIGN AND IMPLEMENTATION FOR RESTORATION OF THE NAVAJO COMMERCIAL FOREST. David L. Szymanski, Bruce D. Willett, Ivan Joe, Russell D. Briggs and Paul F. Hopkins.

A STRATEGY TO IMPROVE FOREST COVER CLASSIFICATION ACCURACY IN NEW YORK USING LANDSAT AND ANCILLARY DATA. David L. Szymanski and Paul F. Hopkins.

DYNAMIC SIMULATION MODELING OF THE LAND USE, ECONOMY AND ENVIRONMENT IN CHIANG MAI, THAILAND USING GIS AND REMOTE SENSING. Siripun Taweesuk and Charles A.S. Hall.

APPLICATION OF SPATIAL MODELS TO DESCRIBE LEAD CONCENTRATIONS AT SKEET AND TRAP RANGES FOR USE IN HUMAN HEALTH RISK ASSESSMENTS.  William Thayer and James M. Hassett.

DETECTION OF POTENTIAL EMBANKMENT BREECHES ALONG THE ERIE CANAL USING INFRARED REMOTE SENSING.  George Washburn.

ASSESSING THE ROLE OF THE CAPILLARY FRINGE IN STREAMFLOW GENERATION: A LABORATORY HILLSLOPE STUDY.   Daniel L. Welsch, K. Niclas Hjerdt, Jeffrey J. McDonnell and Charles N. Kroll..

     Methods used to assess the accuracy of conventional maps may not be appropriate for assessing the accuracy of spatial model output because the intended purposes of maps and models are different. The goal of map creation and analysis is to show exactly what exists at a given location. The goal of a spatial model is often to simulate what could exist in a given location, or show spatial variables in general terms. Error matrices are a useful and commonly accepted method for assessing satellite land cover classifications, but a standard error matrix is inappropriate for many spatial models, since it assumes a level of spatial precision that is difficult to attain and that is not intended. Error assessment for spatial models is complicated by the fact that error may arise from assumptions in the model, or from errors in the spatial data layers used in the model.
     We present a multiple-step method of accuracy assessment for spatially explicit models. Error propagation is used to assesses total error resulting from input data by cascading error from each individual layer through the model. An error matrix is then used to asses total error in the model’s output. An error matrix can then be modified to factor out errors in data layers and determine what portion of model error is due to flaws in the model assumptions.

ENGINEERING ANALYSIS OF GALLEY SYSTEMS AS A SUBSURFACE WASTEWATER DISPOSAL OPTION.  Kelly A. Bazukiewicz and James M. Hassett, Faculty of Environmental Resources and Forest Engineering, 312 Bray Hall, SUNY - College of Environmental Science and Forestry, Syracuse, NY 13210.

     A study of galley systems as an alternative wastewater disposal option has been proposed by the U.S. Environmental Protection Agency (EPA), the New York City Department of Environmental Protection (NYCDEP), the New York State Department of Health (NYSDOH) and the State University of New York College of Environmental Science and Forestry (SUNY ESF). The purpose of this research is to asses the effectiveness of galley systems in treating domestic wastewater as compared to conventional absorption trench systems in a surface water supply watershed. Galley systems are constructed of cement chambers that are open on the bottom and also have small openings on the sidewalls. The difference between galley systems and conventional wastewater absorption trenches is that galley systems are designed to handle highly variable flows and have a structure that allows them to be placed under paved surfaces. The four sites chosen for the study are located in Putnam County, New York (just north of New York City). The flows vary by the day of the week and also by site. At each site one upgradient well along with three monitoring wells located 10 ft from the end of the absorption fields along the flow path of the groundwater were installed. The well water will be analyzed for the following parameters: BOD5, Total Suspended Solids, Dissolved Oxygen, Temperature, Nitrate/Nitrite, Ammonia, Total Kjehdahl Nitrogen, Total Phosphorous, Total Dissolved Phosphorous, Total Coliform, and Fecal Coliform. Sampling will continue for one year after which it will be determined if the galley systems treat the wastewater to acceptable levels.

     Making measurements and subsequent computations and analyses using them, are a fundamental task of surveyors. The process requires a combination of skill and mechanical equipment applied with the utmost judgment. However, no matter how carefully measurements are made, they are never exact and will always contain error. It is important to be able to assess the magnitudes of errors in measurements so that either their acceptability can be verified or, if necessary, new ones taken.
     Accuracy and precision of field data must be established in order to determine the reliability of results. Accuracy refers to the relative nearness a measured value is in relation to the true value while precision refers to the degree of consistency of repetitive measurements. Several repetitions are required and variability, through a series of computations, is usually defined in terms of a standard deviation. For the example provided, overall variability was determined by calculating standard deviations and the root mean square error.
     Data for this example provided came from a study conducted by S.U.N.Y College of Environmental Science and Forestry in conjunction with the New York State Department of Health and the New York City Department of Environmental Protection.

PRODUCING A TOPOGRAPHIC MAP.  Amy Becker, Kevin Riordan and Jessica Ross, Department of Environmental Science and Resource Engineering, SUNY College of Environmental Science and Forestry, One Forestry Drive, Syracuse, NY 13210.

     ERE 371, Surveying for Engineers, is designed to train students in environmental science, engineering science and engineering design for the purposes of providing solutions to problems associated with managing rural or forest land resources. The basic premise being that the forest engineer or manager must be able to locate and quantify the resource being managed or problems being solved. Also, engineers involved with design and construction of structures must acquire some knowledge of construction survey principles and practices.
     Surveying fieldwork requires surveying crews to execute field data collection procedures properly. Laboratory exercises provide experience in making field measurements and applying surveying theory to actual problems. These exercises worked toward preparing an accurate topographic map of an assigned area or traverse, a series of consecutive lines whose lengths and directions have been determined from systematic field measurements. In order to determine ground elevation, automatic levels, theodolites were used to perform techniques such as differential leveling and stadia. Horizontal angles and distances were determined by using transits and total stations.

PHYTOREMEDIATION OF LEAD CONTAMINATED SOIL IN RESIDENTIAL AREAS.  Marie Bowser, Douglas Daley and James M. Hassett, Senior and Professors, Faculty of Environmental Resources and Forest Engineering, 312 Bray Hall, SUNY - College of Environmental Science and Forestry, Syracuse, NY 13210.

     Urban soils frequently have high soil lead concentrations, the source of which is historic deposition of combustion products of leaded gasoline and weathering of lead-based exterior and interior paints. Children are exposed to soil lead by their play activities and hand to mouth behaviors. Children so exposed frequently exhibit elevated ( > 10 m g/dL) blood lead levels; thus soil lead is a public health issue. Phytoremediation, the use of plants to extract contaminants from the soil, is a possible low cost, inexpensive method to reduce soil lead concentrations from shallow contaminated soils. The purpose of this study is to establish criteria by which to evaluate candidate plant species, and to determine methods for phytoremediation of lead-contaminated soils in residential areas.

     Many wastewater treatment plants throughout the United States use chlorination/dechlorination for disinfection of their effluent waters. The handling and storage of chlorine poses a potential health risk to workers at these plants. In addition, the presence of chlorine in the wastewater effluent can form trihalomethanes, which are harmful to aquatic life. New changes in the Clean Air Act are requiring many wastewater treatment plants to upgrade their current chlorine handling and storage procedures. Of interest was how the current costs of chlorination/dechlorination compare to those associated with replacing this system with an alternative disinfection procedure, specifically ultraviolet (UV) light. The Ithaca Wastewater Treatment Plant located in Ithaca, New York was used as a case study for this analysis. Two ultraviolet light disinfection systems were compared to a conventional chlorination/ dechlorination system at the plant using engineering economic comparison techniques. For this case study, the costs of chlorination/dechlorination were significantly less than those associated with retrofitting the plant with a new UV disinfection system. This was due to the relatively large capital costs associated with UV disinfection systems. A sensitivity analysis which analyzed the impact of fluctuation in future costs and the discount rate was performed. This analysis showed that if past increases in chlorination costs continue in the future, the costs of the UV systems become competitive with those of chlorination. In addition, the costs associated with UV systems are strongly influenced by power costs, while fluctuations in the discount rate had relatively little impact on the analysis.

INFRASTRUCTURE REHABILITATION AT CAMP GOODWILL, CHITTENANGO, NY.  Nathan Cheal, Kim Heller, Jeff Ryan, Janine Sleeper, and James M. Hassett, Seniors and Professor, Faculty of Environmental Resources and Forest Engineering, 312 Bray Hall, SUNY - College of Environmental Science and Forestry, Syracuse, NY 13210.

     Camp Goodwill is a summer recreation area for physically challenged children. The camp grounds include a stream that flows into a small (approximately 1 acre) trout pond which is currently full of sediment. The pond and stream are in need of improvements in order to reduce the amount of sediment in the stream and pond and to reduce future sediment transport from the watershed to the pond. To accomplish these goals, a rainfall-runoff model was used to determine storm water flow rates based on a chosen design storm. The model results provided the flowrates for an improved channel cross section and an improved pond outlet structure. Manning’s equation was employed to redesign the channel upstream of the pond so as to prevent sedimentation in the channel section. A pond design methodology was employed to evaluate the performance of the current outlet structure and to consider alternatives for redesign of the pond outflow control device. The final design included an inlet channel that prevents sedimentation and that is lined to prevent erosion. In addition, a new outlet structure is sized to convey the flow of the chosen design storm. The infrastructure improvements will improve the camp grounds and enhance the recreational opportunities for the camp’s clients.

     Low streamflow statistics are used to determine waste-load allocations, site treatment plants and landfills, determine interbasin withdrawals and transfers, and plan water supply, irrigation and hydropower systems. While the need for low streamflow statistics is clear, the methodology to obtain such statistics is not. At gauged river sites, a frequency analysis is typically performed to obtain estimates of low streamflow statistics. Unfortunately, there is no general consensus as to the appropriate probability distribution to describe low streamflow series. Using data from the Hydro-Climatic Data Network (HCDN), a USGS database of streamflows from over 1500 unregulated river sites throughout the United States, various regional distributional hypotheses were performed. Initial results indicate simple 2-parameter probability distributions, such as the lognormal, Weibull, or Gumbel distributions, are not adequate in describing low streamflow series in the United States. L-moment diagrams are developed for various regions and the entire United States for 1-day, 7-day, and 30-day annual minimum streamflow series. These diagrams expose the problems with selecting an appropriate regional probability distribution, especially the difficulty in distinguishing between the fit of competing distributions. The impact of intermittent streamflows on L-moment diagrams is also explored.

WHY IS DRY LAKE SO WET?? SEDIMENTATION PROBLEMS AT DRY LAKE, CLARK RESERVATION STATE PARK, JAMESVILLE, NY. Christopher Craner, Ronald Ferris, Daniel Gilbert, Thomas Parmiter, Trevor Thomas and James M. Hassett, Seniors and Professor, Faculty of Environmental Resources and Forest Engineering, 312 Bray Hall, SUNY - College of Environmental Science and Forestry, Syracuse, NY 13210.

     Dry Lake is a geologic feature located in Clark Reservation State Park, Jamesville, NY. The one acre depression fills with spring runoff and after heavy rains. Park officials have noticed that, beginning about 1991, the depression retains more water for a longer time than was the case in earlier years. The increased level and duration of inundation has caused changes in the plant community surrounding the depression, degrading the lake’s unique character.
     The increased water levels in Dry Lake were observed at about the same time as a housing development was under construction in the watershed draining into Dry Lake. It was hypothesized that the construction activities, which ultimately occurred over 40% of the watershed area, contributed an increased sediment load to Dry Lake, thus decreasing the permeability of the soils in the depression and increasing the levels and period of inundation.
     We investigated the watershed and in particular the storm water management system constructed as part of the housing development. We focused on the storm water detention basin and suggest a redesign to better capture soil particles eroded from the watershed. We also investigated the soils within Dry Lake itself and suggest remedial actions to restore the drainage characteristics to pre-development values.
     The combination of improvements to the storm water retention basin and the rehabilitation of Dry Lake itself will return Dry Lake to pre-development conditions.

     Manning’s equation, first formulated in the 1890’s, is widely used for open channel flow analysis and design. The equation relates channel slope, geometry, and an empirically-derived friction factor (Manning’s n) to flow. Manning’s n has proven problematic in that there exist no universally accepted methods to estimate values for it; the friction factor is therefore the largest source of error in estimating open channel flow values. In this study, we measured flows at three channels (Meadowbrook near Nottingham High School, Harbor Brook at Grand Avenue and Onondaga Creek at Dorwin Avenue) and compared our measured flows to flows obtained from the USGS, and flows calculated from two forms of Manning’s equation and two recently proposed regression-based flow equations. We compared the estimates of flows derived from the different methods in an attempt to identify the major sources of error. We did not find a consistent pattern in that some equations seemed to be more accurate in some channels but not others. Overall, Manning’s equation performed as well as any other method evaluated to calculate open channel flow values.

A NEW PROBABILISTIC METHODOLOGY FOR SELECTING REMEDIATION GOALS: A CASE STUDY OF A LEAD RISK ASSESSMENT FOR A SUPERFUNDSITE IN MURRAY, UT. Philip Goodrum and James M. Hassett, Faculty of Environmental Resources and Forest Engineering, 312 Bray Hall, SUNY - College of Environmental Science and Forestry, Syracuse, NY 13210, Gary Diamond, Syracuse Research Corporation, 6225 Running Ridge Road, North Syracuse, NY 13212, and Susan Griffin, USEPA Region 8, Denver CO.

     We describe an uncertainty analysis of childhood lead (Pb) exposures for the Murray Smelter Superfund site. Our probabilistic methodology combines information from site sampling data with estimates of health risk (specifically predicted blood-Pb concentrations) estimated from different post-remediation exposure scenarios. A Monte Carlo exposure model linked to the USEPA Integrated Exposure, Uptake and Biokinetic model for lead was run iteratively, with each simulation using a different distribution of post-remediation soil Pb concentrations. A ‘decision matrix’ was generated which shows site managers how the distribution of blood-Pb concentrations vary as a function of both the arithmetic mean and the expected spatial variability in Pb soil concentrations after site remediation. The results are compared with simulations using other probabilistic methods, including 1) inputting a fixed risk level in the back-calculation of Preliminary Remediation Goals and 2) iteratively truncating the distribution of soil Pb concentrations in Monte Carlo simulations until the probability of blood Pb levels exceeding 10 m g/dL is below 5%. The proposed methodology can incorporate geostatistical information from sites with limited sampling data, and can be applied to risk assessments of other contaminants of concern in soil.

     Fish habitat in stream systems is affected by the velocity and depth of flow and the composition of the channel substrate. We are investigating the use of two hydraulic measures to describe habitat: stream power and the Froude number. Stream power is the rate at which work is done by flowing water and can be expressed as w = g QS where w is the stream power, g is the specific weight of water, Q is the discharge, and S is the stream gradient. The Froude number is a dimensionless variable used by hydraulic engineers to describe flow at a channel cross section and can be expressed as where B is channel width at the water surface, g is acceleration due to gravity, A is cross sectional area through which the flow occurs, and Q is again the discharge. Fr incorporates both morphological (area, top width, depth) and hydraulic (discharge) features of the channel cross section. Stream power is thus a measure over a channel reach identified by a discharge and slope; the Froude number is a measure of channel attributes at a particular cross section.
     We are relating the two hydraulic measures to the fish community in a small (116 km2) stream system in central New York. We are comparing, by regression techniques, fish density estimates (in kg m-2) for six fish species at fifteen different sites. Our results suggest that either hydraulic measure can explain some of the variability in fish community structure, and that either or both can be incorporated into Habitat Suitability Indices to describe stream habitats.

     Residual Bio-Tech Soil (RBSÔ ) consists of municipal incinerator ash combined with yard waste compost at varying mixtures. The structure of the ash fraction of RBS provides the substrate with excellent soil aeration in the form of large percentages of pore spaces (10-15% macropore, 35-40% micropore). Meanwhile the compost fraction provides a high organic matter content to RBSÔ . The mixed substrate has a high field capacity and is well suited for supporting vegetation. Previous studies have shown that this substrate is also capable of supporting large populations of earthworms. Two separate but related projects are currently under way. The OFFRSÔ (On Farm Fertilization Recovery System) is investigating the possibility of using RBSÔ with the earthworm Eisenia foetida to treat dairy wastes. The CMADSÔ (Continuous Mesofaunal Agitation, Digestion, & Stabilization) is investigating the possibility of treating municipal wastewater sludge with this system. In both projects, the RBSÔ – earthworm combination acts as a biofilter, removing unwanted biosolids and eluting clear, clean water. The effectiveness of the system as a biofilter is due to enhanced microbial decomposition as a result of the combination of ash particles and earthworm populations. The high specific surface area of the ash particles provides a greater number of sites for microbial attack that would normally be found in native soils or filter using glass beads or sand as a substrate. Earthworms present in a soil have been observed to increase microbial activity up to 600%, but generally from 50-200% (1). The total microbial population remains the same, but the selective feeding of earthworms on microorganisms provides as opportunity for microorganisms to re-colonize earthworm feeding sites and burrows. Further research on ammonia and salinity stress, as well as the ultimate fate of pathogens is underway.

1. Satchell, J.E. 1983. Earthworm Ecology from Darwin to Vermiculture. Chapman and Hall, New York.

IMPROVING ACCESS AND USE OF SPATIAL DATA VIA THE INTERNET AND GEOGRAPHIC INFORMATION SYSTEMS: CASE PROJECT - USE OF THE ENVIRONMENTAL PROTECTION AGENCY'S INDEX OF WATERSHED INDICATORS FOR ACCESS AND USE OF WATER QUALITY DATA FOR WATERSHEDS. Michael Plastino, Faculty of Environmental Resources and Forest Engineering, 411 Bray Hall, SUNY-College of Environmental Science and Forestry, Syracuse, NY 13210. Faculty Advisor: Dr. Paul Hopkins.

     Water pollution control has traditionally relied on broadly applied national programs to reduce water pollution from individual sources and from polluted runoff. The Clean Water Initiative, recently released by President Clinton, emphasizes broadening of this scope from individual sources to watersheds. Such an approach is becoming increasingly feasible due to advances in internet and geographic information systems (GIS) technology. The Environmental Protection Agency (EPA) has GIS Reach Files which allow access and use of water quality data for any watershed in the United States. Watershed data sets collected by various tribal, state, and federal agencies are available in 15 different categories, which are collectively referred to as the "Index of Watershed Indicators" (IWI). Access to the IWI is available via the EPA web site.
      The goal of this project is to assist the EPA in improving water quality analysis capability at the watershed scale. Following are the primary project objectives for achieving this goal:

• Write and modify programs in Avenue (ArcView GIS Programming Language) and HTML (Internet Programming Language) to improve the user interface with the IWI data sets. Programs will facilitate the downloading or linking of data sets corresponding to the 15 IWI indicators.
• Perform spatial analyses on data sets to assess the validity of watershed data. The EPA is currently working on upgrading the capability to provide water quality data at the watershed level so that information on individual watershed components, such as rivers and lakes, could also be obtained using the EPA "Surf Your Watershed" web site. Spatial analyses will be performed to evaluate the accuracy of this upgrading process.
• Perform network analyses to assess the efficiency of the Reach Files' hydrologic network, which allows users to navigate upstream or downstream from any point of interest in a watershed. Upgrades in the hydrologic network will be suggested and implemented as necessary.

     Mixed forest classification in upstate New York is problematic at best; this is partly due to the definitions used as to what constitutes mixed forest, its spatial and species characteristics, and the confusion resulting from the delineation of forested wetlands.
     There are two main components of mixed forest, the variety of species present and how they mix spatially. To be simplistic, species can mix either homogeneously or in discrete clumps . This ongoing investigation aims to see if species mix in a predictable manner for specific types of mixed forest. In order to determine this, a selection of continual forest inventory plots located in ESF forest properties have been mapped to determine the stem position (horizontal mixing component), crown position (vertical mixing component), species, height and DBH of all the trees in the plot. By using differentially corrected GPS data to locate the plot centers, the plots can be located on the relevant satellite images. The spectral response of the selected plots will be analyzed using existing Thematic Mapper imagery held in the Mapping Sciences Laboratory. Ultimately this research aims to see if there is a predictable spectral response for the different types of spatial mixing encountered in mixed forest.

     The Kappa coefficient has been used since the mid 80's as the standard assessment of accuracy for digitally derived maps. Kappa has proved to be a useful tool, but there is now a need to reexamine its utility as the standard accuracy assessment for digital mapping purposes.
     This poster, which is being done to meet the course requirements for Steve Stehman's APM 696 Map Accuracy course, does not seek to provide a replacement for Kappa, but to introduce discussion as to its validity and use for basic mapping purposes. We seek to raise questions in the minds of the producers of digital maps as to how best to express the accuracy of their maps.

     In recent years, digital imaging has become increasingly common in a wide range of disciplines. As a result of this the application of image processing techniques frequently used in the mapping sciences has expanded. This project looks at image registration being carried out in medical imaging. Image registration in medicine enables doctors to combine images from a variety of sources, each providing different and often complementary views of the anatomy. The images used for this project are a set of Magnetic Resonance images (MRI) and a set of Single Photon Emission Computed Tomographic (SPECT) images of a human brain. The MR images provide anatomical detail; the SPECT images provide information about brain function through blood flow
     The algorithms typically used in the mapping sciences are not appropriate for the medical image set available. Low resolution in the SPECT data set limits the ability to define features that are common to both image sets, therefore a surface matching technique must be used to register the images. The technique used for the registration is called a "head and hat" approach. This procedure carries out a least squares solution to match two three dimensional surfaces. In this case the layers of the high resolution MRI form a "head" while points on the surface of the lower resolution SPECT creates a "hat". The "hat" is then matched to the "head" through a three dimensional translation, rotation and scaling.

COMPUTER SIMULATION OF HYDROGEN SULFIDE REMOVAL FROM DIGESTOR OFF-GASSES BY USE OF A FIXED FILM BIOSCRUBBER.   Jan Salzman, Kirsten Kenty and James M. Hassett, Faculty of Environmental Resources and Forest Engineering, 312 Bray Hall, and Siddharth G. Chatterjee, Faculty of Paper Science and Engineering, 205 Walters Hall, SUNY-ESF, Syracuse, NY 13210.

     Anaerobic digestion processes frequently generate hydrogen sulfide (H2S) gas as an unintended byproduct. H2S is toxic at low concentrations and, by virtue of its obnoxious odor, represents an environmental insult to people in the vicinity of the reactor. We are examining a computer model to examine H2S removal in digestor off-gasses by means of a fixed film bioscrubber. The model is being adapted from a computer simulation of volatile organic carbon (VOC) treatment. Specific enhancements to the VOC model to better simulate H2S treatment include: incorporation of a novel process stoichiometry describing both cell synthesis and H2S oxidation to sulfate, addition of improved descriptions of gas-liquid and liquid-liquid mass transfer processes, and the inclusion of H2S equilibrium chemistry in describing gas-liquid mass transfers. Model results will be compared to data from a pilot plant designed and built to treat H2S generated from the anaerobic digestor at a local industrial facility. In addition, the use of the model to investigate H2S treatment options (recycle rates, superficial gas velocities and reactor residence time, co-current versus counter-current gas and liquid flows) will be illustrated.

THE ESTIMATION AND USAGE OF BASEFLOW RECESSION CONSTANTS IN LOW STREAMFLOW MODELING. Frank Sidari, Faculty ofEnvironmental Resources and Forest Engineering, Bray Hall, Charles N. Kroll, Faculty of Environmental Resources and Forest Engineering, Bray Hall, SUNY College of Environmental Science and Forestry, Syracuse, NY 13210.

     Low stream flow estimates are required for planning water supply, hydropower and irrigation systems, determining wasteload allocations to rivers, and siting treatment plants and sanitary landfills. A common method to estimate low streamflow statistics at ungauged river sites is to develop a regional model of low streamflow statistics as a function of easily measured watershed parameters such as drainage area, average basin slope, and indices of precipitation and soil. Since low streamflows are usually the result of groundwater outflow during periods of little or no rainfall, the hydrogeologic properties of the watershed should be strongly correlated with low streamflow characteristics. Unfortunately, indices describing hydrogeology have not been employed in regional low streamflow models because they are either difficult to obtain or highly uncertain. This research examined the estimation of baseflow recession constants (Kb), a hydrogeologic index which describes how streamflows recede during periods of no recharge. Of interest is whether Kb is a significant descriptive variable in low streamflow models. A case study from the Northeastern United States is presented.

DEVELOPING AN INTERNET-ACCESIBLE MODELING PACKAGE. William M. Stiteler, Jennifer L. Barber, and Paul F. Hopkins, Faculty of Environmental and Resource Engineering, 312 Bray Hall, SUNY College of Environmental Science and Forestry, Syracuse, NY 13210.

     Ecological and environmental models are generally created by those whose expertise lies in a particular field. Models from different fields often share similar components and processes, but also include unique approaches, strategies, and tools. Scientists and others who wish to use models are often not experts at modeling, and could benefit from a modeling package that simplifies the process. We are creating an Internet-accessible modeling package that brings together techniques from various fields, and that features a Graphical User Interface (GUI). A GUI which presents the user with a set of modeling functions and which aids in setting up interactions will allow the user to concentrate on the modeling task at hand, rather than on the details of modeling. The first implementation of this modeling package will use Northern Forest Lands Inventory (NFLI) data to model hunting pressures as they vary with land ownership in Lewis County, New York.

     Within the Navajo Nation Territory, Dine’ Citizens Against Ruining Our Environment (CARE) has been working to restore traditional values in land management decisions. A geographic information system (GIS) is being constructed to help organize and understand the complex information needed for this purpose. Currently, a restoration effort is underway within the borders of the Sanostee Chapter (New Mexico) commercial forest that has a history of heavy logging and fires. The objectives for using GIS in this restoration project include: restoring the forest to support traditional subsistence uses; remove the aspen/scrub oak remnants of the extensive 1956 fire; and abate the severe erosion caused by the neglected logging roads.
     With a bedroom as an office and a small budget, we began the effort by acquiring the requisite computer hardware, software, and digital data. The database consists of forest stands, elevation, watershed boundaries, and soil information. A Landsat-TM image also is to be used to provide a current documentation of the forest cover. These layers will be registered to a common datum and queried to define the best course of restoration planning.
     The database is still under construction and the planning process for restoration is being defined. The Sanostee people, especially the youth, are eager to use modern technology, like GIS, to address traditional concerns about their natural resources

A STRATEGY TO IMPROVE FOREST COVER CLASSIFICATION ACCURACY IN NEW YORK USING LANDSAT AND ANCILLARY DATA. David L. Szymanski and Paul F. Hopkins, Faculty of Environmental Resource and Forest Engineering, 312 Bray Hall, SUNY College of Environmental Science and Forestry, Syracuse, NY 13210.

     Forest cover maps derived from Landsat Thematic Mapper (TM) imagery have become a popular layer in geographic information systems (GIS). A common classification scheme, the Anderson Level II Deciduous Forest, Evergreen Forest, and Mixed Forest cover classes provide a sufficient documentation of the current inventory on the ground. Using TM imagery in a classification technique (like maximum likelihood) with only spectral information from one date, the three Anderson classes often have degraded accuracy (< 80% correctly classified). The Mixed Forest, with its complex, heterogeneous canopy, seems to be incorrectly classed with more frequency than either of the other two classes that have a more homogeneous canopy.
     A simplified method is developed and tested for its efficacy to improve the Anderson Mixed Forest classification accuracy. Huntington Forest and Heiberg Forest will be used as test sites. A spectrally classified image is augmented with digital data representing slope, aspect, elevation, and soil drainage. A spatial model which relates these site factors to forest development, by way of a plausibility index, is implemented such that image pixels whose forest cover class is sufficiently implausible (lowest index score), given the topoedaphic constraints, is assigned to a more plausible forest cover class (highest index score). Image processing is performed using the Erdas Imagine package while the spatial model algorithm is designed with C++ computer language.
     Results indicate a small, yet statistically insignificant, increase in classification accuracy. With Anderson Level II, species with varying site tolerances will be included in the same broad class. A simplified plausibility index may not be enough to overcome this within-class variability. This spatial model may be inappropriate for areas which have undergone management practices that create a forest with some departure from natural growing conditions.

     Land use is changing rapidly in Chiang Mai, Northern Thailand. Population growth is clearly an important driver. However, the linkage between population growth and change in the forest area, cultivated land or built-up area is modified by a number of environmental, economic, political, and social factors that vary with space and time. Several research projects previously have quantified and also predicted land use change in different areas of Chiang Mai. More recently new technologies of science, including remote sensing, GIS and simulation modeling, have been integrated to understand and simulate land use and land use change patterns. These models can simplify complex environmental problems for the decision-maker and lead to a better understanding of complex ecosystem. Thus the emerging synthesis of digital image processing, GIS technology and modeling is an interdisciplinary attempt to meet the challenge of predicting how and where people will use different land resources in the future.
     The purpose of my proposed research is to demonstrate that the advanced technologies available in computerized geographic information systems using raster format are capable of combining remote sensing data and simulation modeling to analyze land use change and soil erosion. The use of historical trends, socioeconomic trends and statistical analysis of economic drivers will be examined to enhance the accuracy of land use images.

     Risk assessments are performed on sites contaminated with hazardous substances to determine if a risk to humans or the environment exists and if so, to aid in the management of remedial actions taken at the site. One of the critical inputs to the risk equation is the exposure point concentration (EPC). EPA risk assessment guidance requires the use of the 95th-percentile upper confidence limit (UCL) of the arithmetic mean of the site contaminant data as the EPC. The 95th UCL of the mean is used to account for the variability and uncertainty inherent in estimating the EPC. The deterministic, non-spatial EPA method produces an estimated EPC that is often greater than the highest concentration observed in the site data and implicitly assumes that each data point represents an equal area of the site. The use of spatial models to estimate the EPC allows for the explicit treatment of uncertainty and variability in the risk assessment process that provides an opportunity for an iterative approach to risk management. The effect of using spatial models to estimate exposure concentrations at skeet and trap ranges will be investigated. The use of the estimated concentrations to formulate risk-based approaches to site remediation and management will also be explored. Estimates of the 95th UCL by the EPA method (non-spatial, deterministic method), the bootstrap method (non-spatial, stochastic method), Thiessen polygons with bootstrapping (spatial, stochastic method) and geostatistics (spatial, stochastic method) will be used to determine risk-based soil cleanup goals for lead, arsenic and antimony. A risk-based approach to perform additional sampling and generate remedial alternatives for a generic skeet and trap range will be developed.

DETECTION OF POTENTIAL EMBANKMENT BREECHES ALONG THE ERIE CANAL USING INFRARED REMOTE SENSING.  George Washburn, Faculty of Environmental Resource and Forest Engineering, 410A Bray Hall, SUNY-College of Environmental Science and Forestry, Syracuse, NY 13210.

     Thermal infrared remote sensing technology has been utilized for finding water or wet areas over the past few decades. These sensors can be used on the ground, from airplanes, and from spaceborne satellites. It has long been understood that water has a much greater thermal capacity than metals, rocks and soil. Because of this property, water maintains a much more constant temperature than these other types of material, and therefore has a more consistent diurnal radiant temperature. Likewise, wet soil has a greater thermal capacity than dry soil and therefore maintains a more constant temperature.
     Thermal infrared remote sensors designed to measure radiated electromagnetic heat (8 - 14 um wavelength) can be used to take advantage of water’s thermal properties, in order to find water bodies, standing water and wet soil from their surrounding background. Because of water’s greater thermal capacity, it heats up more slowly during daylight hours, and cools down more slowly after nightfall. Therefore, standing water and wet soil will tend to show as light (or warm) areas in thermal infrared imagery collected during mid night-time hours, and as dark or cool areas on the same type of imagery collected at mid-day.
     Whereas thermal infrared radiance measured at an airborne sensor is mainly emitted energy from the Earth’s surface and self radiance of the atmosphere, near infrared radiance (0.7 - 1.0 um wavelength) consists mainly of energy from the Sun reflected off the Earth’s surface or particles in the Earth’s atmosphere. Because of this property, near infrared radiance is of a much greater amplitude during daylight hours than during the night. Water absorbs most of the near infrared energy incident upon it, and likewise, wet soil reflects less near infrared energy than dry soil. Therefore, water and wet soil appear as dark areas on near infrared, gray-scale imagery, where water bodies appear to be black and wet soil appears dark gray.
     Taking all of the aforementioned considerations into account, water and wet soil can best be ascertained from a combination of near and thermal infrared remote sensing performed by an airborne sensor flown during mid-day hours on a sunny day following a cold night. By mid-day, the surrounding dry areas on the ground have heated up substantially and therefore radiate strongly in the thermal band pass, objects on the ground reflect the greatest amount of energy in the near infrared band pass due to steep Sun angles, and thermal and near infrared shadows caused by sunlight zenith angles are minimal.

ASSESSING THE ROLE OF THE CAPILLARY FRINGE IN STREAMFLOW GENERATION: A LABORATORY HILLSLOPE STUDY.   Daniel L. Welsch, K. Niclas Hjerdt, Jeffrey J. McDonnell and Charles N. Kroll, Faculty of Environmental and Resource Engineering, Bray Hall, SUNY College of Environmental Science and Forestry, Syracuse, NY 13210.

     The role of the capillary fringe in streamflow generation has become firmly entrenched in the literature as a mechanism to explain large contributions of pre-event water to the stream during precipitation events. However, the narrow physical constraints necessary for substantial contributions of capillary fringe-induced subsurface water discharge prohibits this process from being a generally accepted streamflow-generating mechanism. We present results from a laboratory tracer-hydrometric study that explores several assumptions associated with the capillary fringe phenomenon. Specifically, we address the following questions: (1) Do differing fringe heights have an effect on the proportion of event / pre-event water in hillslope subsurface flow? (2) Where and to what extent does the mixing between event and pre-event water occur under varying fringe heights? (3) How are mixing mechanisms related to fringe height and rapid water table response? A Plexiglas tank on a 12 degree slope, measuring 3 x 0.3 x 0.3 meters, was filled with homogeneous sand and instrumented with 15 recording tensiometers and six suction lysimeters. The sands used in the three runs had grain sizes of 0.21 mm, 0.42 mm and 0.84 mm. The capillary fringe heights for these sands were 0.39 m, 0.30 m, and 0.15 m, respectively. A bromide tracer was applied in simulated rainfall events of 30 mm/hr for both 10 and 20 minutes. Water samples from the outflow and the lysimeters were analyzed and a hydrograph separation was made, using a simple two-component mixing model. Preliminary results from this investigation suggest that the fraction of event / pre-event water is similar irrespective of capillary fringe heights and grain size characteristics. Tensiometer information will be used to identify mechanisms that are responsible for mobilization of pre-event water during laboratory rainfall events.