2004 Progress Report: Ecological Classification of Rivers for Environmental Assessment: Demonstration, Validation, and Application to Regional Risk Assessment Across Illinois, Michigan, and WisconsinEPA Grant Number: R830596
Title: Ecological Classification of Rivers for Environmental Assessment: Demonstration, Validation, and Application to Regional Risk Assessment Across Illinois, Michigan, and Wisconsin
Investigators: Seelbach, Paul W. , Clark, Richard D. , Wehrly, Kevin E. , Wiley, Michael J. , Stevenson, R. Jan , Holtrop, Ann , Cooper, Arthur , Pijanowski, Bryan , Allan, David , Day, David , Baker, Edward , Lyons, John , Hinz, Leon , Wang, Lizhu , Mitro, Matt , DePhilip, Michelle , Brenden, Travis , Zorn, Troy
Current Investigators: Seelbach, Paul W. , Clark, Richard D. , Wehrly, Kevin E. , Wiley, Michael J. , Stevenson, R. Jan , Aichele, Stephen S. , Holtrop, Ann , Cooper, Arthur , Pijanowski, Bryan , Bissell, Ed , Stewart, Jana , Lyons, John , Hinz, Leon , Wang, Lizhu , Mitro, Matt , Steen, Paul , Brenden, Travis , Zorn, Troy
Institution: University of Michigan , Purdue University , Wisconsin Department of Natural Resources , Illinois Department of Natural Resources , Nature Conservancy, The , Michigan Department of Natural Resources , Michigan State University
Current Institution: Michigan State University , Illinois Department of Natural Resources , Illinois Natural History Survey , Michigan Department of Natural Resources , Purdue University , United States Geological Survey , University of Michigan , Wisconsin Department of Natural Resources
EPA Project Officer: Packard, Benjamin H
Project Period: December 1, 2002 through December 31, 2006
Project Period Covered by this Report: December 1, 2003 through December 31, 2004
Project Amount: $842,547
RFA: Development of Watershed Classification Systems for Diagnosis of Biological Impairment in Watersheds (2002) RFA Text | Recipients Lists
Research Category: Watersheds , Water
The objective of this research project is to couple landscape-based modeling from large, regional data sets and regional land transformation models (LTMs) with a valley segment ecological classification approach already being employed in several Midwestern states. Specific objectives of the research project include the completion of a GIS-based river segment classification and provision of a comprehensive status and risk assessment of river systems across the upper Midwestern states of Illinois, Michigan, and Wisconsin.
We completed the GIS and landscape databases for each of the states of Illinois, Michigan, and Wisconsin and believe that our approach is broad enough to be adopted as a national standard for regional river studies as described in Brenden, et al. (in press).
Delineating Valley Segments
We are attempting to establish general methods that will provide a stronger statistical basis for defining valley segments in our three-state area or for use elsewhere. We developed algorithms and statistical analyses that will allow clustering of the smaller river reaches into valley segments based on the landscape and modeled variables in our GIS. The primary method being used at present is based on the Cluster Affinity Search Technique (CAST). Essentially, this approach will guarantee a predetermined level of homogeneity for valley segments. We tested this method with dummy variables; modeled variables for temperature, flow, and fish must be created before a definitive analysis takes place. We then made an initial assessment to determine which of the many variables in our GIS will be most important. A more detailed account of this analysis is available at http://sitemaker.umich.edu/riverclassproject/files/report_from_travis.doc Exit .
Stepwise multiple linear regression analysis was used to develop models predicting mean July stream temperatures in Michigan streams as a function of various landscape-scale data. These models explained from 50 to 70 percent of the variation in mean July temperatures. A number of additional methods were explored in an effort to improve the predictive ability of stream temperature models.
Multiple linear regression was used to develop predictive models of annual exceedence flows (i.e., 5%, 10%, 25%, 50%, 75%, 90%, and 95%) and high and low flows throughout the study area. Model development began with a simple hydraulic geometry equation that included the catchment area, precipitation, and valley slope. Additional predictors were added to this base model in a stepwise fashion beginning with surficial geology summaries. When the base model with surficial geology had been developed, agricultural, urban, and other land covers of interest were added. These models explained between 72 and 99 percent of the variation in flows. The flow models recently were used to predict flows for all stream reaches in our three-state GIS, which completes the flow modeling segment of this study.
Preliminary fish modeling work was conducted in all three states. In Michigan, we evaluated four methods (i.e., multiple linear regression, logistic regression, neural networks, and classification trees) for predicting the presence or absence of brook trout. The logistic regression model predicted with the least error, followed by multiple regression, classification trees, and neural networks (Steen, et al., in press). In Wisconsin, we used Classification and Regression Tree (CART) models to predict the occurrence, abundance, and standing crop of fish assemblages and individual species based on 30 landscape variables. A detailed account of this analysis is available at http://www.sitemaker.umich.edu/riverclassproject/files/2004_wifishmodels_lyons.doc Exit . Fish modeling efforts also are underway for Illinois.
Land Transformation Modeling
Development of land transformation models (LTMs) is nearing completion. Data for the three-state forecasts are now assembled, and the projected and land-use classification system was standardized across the study region. Michigan forecasts are now complete to year 2040 with 5-year time steps. Maps of predicted land-use changes in Michigan are available at http://www.sitemaker.umich.edu/riverclassproject/files/pijanowski_et_al_epa_2005_report.doc Exit . Improvement over past forecasts of the state include increased accuracy based on historical hind-cast calibration metrics. Also, multiple transitions of use incorporated into the model were completed to ensure that major transition types occurring in the region (e.g., agriculture to forest) are accounted for adequately in the coupled models. New land-use data for the Milwaukee metropolitan area have been examined in detail in our final preparation to use this dataset as the primary training set for the Wisconsin projections. Analysis of this portion of the study area shows that patterns and rates of change are similar to those occurring across the three-state region. A draft of a paper that outlines a new method incorporating entropy measures in the assessment of model performance is being developed.
We evaluated the relative effects of human disturbance levels on the influence of catchment, network riparian, reach riparian, and instream variables on fish assemblages (Wang, et al., in press). We found that in largely undisturbed catchments, fish assemblages were predominantly influenced by local factors; however, as disturbance increased in catchments and riparian areas, the relative importance of local factors declined and that of catchment increased. Instream and riparian habitat improvements, therefore, would be most effective in catchments that are largely undisturbed and catchment scale land-use management would be more effective for improving stream quality in degraded catchments.
During Year 3 of the project, we will complete: (1) statistical analyses to define valley segments; (2) a regional algal database and LTM; (3) temperature, flow, and fish modeling work; and (4) classifications of valley segments.
Journal Articles:No journal articles submitted with this report: View all 34 publications for this project
Supplemental Keywords:watershed, geographic information system, GIS, landscape metrics, Upper Midwest region, Illinois, Michigan, Wisconsin, IL, MI, WI, stream valley segment definition and validation, ecological classification of streams, modeling stream flows, modeling stream temperatures, modeling fish distribution and abundance, modeling regional land transformation, stream status and risk assessment, state resource agency survey databases, predicting stream reference conditions from landscape variables, hydrologic regime, thermal regime, bioassessment, fish assemblage, hierarchy, monitoring,, RFA, Scientific Discipline, INTERNATIONAL COOPERATION, ECOSYSTEMS, Water, Geographic Area, Waste, Ecosystem Protection/Environmental Exposure & Risk, Bioavailability, Aquatic Ecosystems & Estuarine Research, Water & Watershed, State, Aquatic Ecosystem, Water Quality Monitoring, Environmental Monitoring, Terrestrial Ecosystems, Ecology and Ecosystems, Watersheds, anthropogenic processes, fate and transport, model, nutrient transport, anthropogenic stress, bioassessment, watershed classification, biodiversity, watershed management, ecosystem monitoring, conservation, diagnostic indicators, ecosystem indicators, Illinois (IL), aquatic ecosystems, water quality, Wisconsin (WI), bioindicators, watershed sustainablility, biological indicators, ecosystem stress, watershed assessment, transport modeling, nitrogen uptake, conservation planning, bioavailable phosphorus, agricultural community, aquatic biota, land use, restoration planning, watershed restoration, Michigan (MI), ecosystem response