Grantee Research Project Results
Final Report: Integrating Planning, Forecasting, and Watershed Level Ecological Risk Assessment Techniques: A Test in the Eastern Cornbelt Plains Ecoregion
EPA Grant Number: R824769Title: Integrating Planning, Forecasting, and Watershed Level Ecological Risk Assessment Techniques: A Test in the Eastern Cornbelt Plains Ecoregion
Investigators: Gordon, Steven I. , White, Dale A , Ward, Andy
Institution: The Ohio State University
EPA Project Officer: Packard, Benjamin H
Project Period: October 1, 1995 through September 1, 1998
Project Amount: $445,000
RFA: Water and Watersheds (1995) RFA Text | Recipients Lists
Research Category: Watersheds , Water
Objective:
The objectives of this research are to: (1) test the relationships between biological conditions of streams and the nature and distribution of human activities on the watershed; (2) demonstrate methods for linking physical models of urban and agricultural impacts on runoff volume and runoff quality; (3) define the relationships between physical model forecasts and the biological conditions of streams; and (4) integrate all of the findings into an expert system to be used by planners.For the empirical models, the objectives were to find a set of regional scale indicator variables that provide a statistically significant explanation of the differences in watershed biological quality. Based on such a model, the original intent was to create a screening tool that planners could use to forecast the impacts of land use and related changes.
For the physical modeling, the main goals of the study were to: (1) develop a watershed methodology that linked three physically based models (ADAPT, SWMM, and REMM); and (2) to find the integrated impacts of agricultural, urban, and riparian zone land use changes on the hydrology of a mixed land use watershed. The philosophy of this study was to link existing programs that simulate individual components of a watershed system together to obtain a more complete simulation of the hydrology of the total system. The benefits of this modeling approach are: (1) advantage is made of previous efforts to simulate complex hydrologic processes; and (2) algorithms that have been accepted by the user community, or have been thoroughly tested, can be utilized and there is less need to conduct extensive validation studies on each component of the linked methodology.
Summary/Accomplishments (Outputs/Outcomes):
Using a comprehensive regional database, we tested watershed scale relationships between biological measures of water quality and watershed characteristics. These have been derived in the form of regression models where the dependent variable is our measure of biological quality - the Index of Biotic Integrity, a measure of the diversity of the fish community. The independent variables are watershed land use distribution variables, stream habitat condition as measured by several components of the Qualitative Habitat Evaluation Index, a proxy for point source pollution based on selected point source discharge variables, stream order, and several soils variables. The best results were obtained by subdividing the sample into two subpopulations - one with watersheds with more urban land uses and one with little (mostly agricultural) land uses.The watershed scale modeling and data analysis were accomplished primarily as the work of Sarada Majumder, a Ph.D. candidate in City and Regional Planning at The Ohio State University. In Year 2, we were able to assemble a large, regional database for the Eastern Cornbelt Plains ecoregion. The stream GIS files were used as the basis to divide the region into two distinct sets of watersheds. Data were assembled on 25 coarse scale watersheds. These were further subdivided into 137 finer scale watersheds.
For the region as a whole, LANDSAT imagery was utilized to define the major land use categories across the region. This was superimposed on the two watershed coverages to extract the land use distribution by watershed for each set of watersheds. A number of other data were compiled using the same geographic base. First, we assembled the dataset for the IBI (Index of Biologic Integrity), the metric for the health of the fish community measured by the Ohio Environmental Protection Agency between 1990 and 1994. The data were collected at 522 locations throughout the watersheds.
We have been able to define a set of empirical models for the Eastern Cornbelt Plains Ecoregion that explain a significant amount of the variation in IBI levels across watersheds. The models explain from 53 percent to 66 percent of the variance in IBI and are all statistically significant at the .01 or .05 level. We have found significant spatial autocorrelation in the models indicating that there is some spatial regularity to the unexplained variance. Based on this, we further subdivided the region into two groups of watersheds - rural and urban - and reformulated the models for watersheds classified as urban and rural. For the urban models the spatial autocorrelation is no longer significant. We suspect that further refinement of the agricultural land use categories might help to further subdivide the region and result in a better level of explanation.
Because the statistical models contain information on the ratings of stream habitat conditions, it was not possible to directly derive a screening model that planners could use directly. Instead, we have begun to investigate the potential of a two-stage model where land use decisions in the riparian zone can be used to project habitat conditions and then fit into the regional models. These models are being derived as a part of a continuing cooperative agreement.
The physical modeling study used a stepwise method to evaluate the effectiveness and accuracy of using a simple modeling approach (based on outputs from a single field scale model) compared to considering progressively more complex modeling approaches that incorporated several models, subdivision of a watershed into subwatersheds, and channel routing. For the first step, the ADAPT model was used to predict the hydrologic responses of each subwatershed, then the results were combined to obtain total discharge, sediment, and nitrate predictions for the whole watershed. For step two, the REMM model was used to model the effect of riparian areas. Several tests were conducted to find the respective impacts of geometry, vegetative structure, and association between upland and riparian zones. For step three, the SWMM model was used to predict the hydrologic responses of the urban areas, ADAPT was used for predicting hydrologic responses of rural areas, and REMM was used to simulate the riparian zones on the subwatershed.
The study was conducted on the Darby Creek Watershed in central Ohio. This 534 square mile watershed is in the Eastern Corn Belt Plains Ecoregion and has smooth plains, a predominance of cropland, soil parent material of limestone and glacial drift, and potential natural vegetation of beech-maple forest. The watershed drains into two main streams: Big Darby Creek and Little Darby Creek, which are national scenic rivers and contain a wide diversity of fish and mussels. Currently, the aquatic ecosystem is being threatened by land use changes, which include intensive development along parts of it eastern boundary and losses of riparian zones. Observed daily flow information for the period 1990 to 1995 was used to evaluate the performance of the various modeling strategies. In some cases, data for the first 3 years were used to calibrate the models. Observed sediment and nitrate data were available for part of the period and were used to evaluate the ability of the models to predict these flow constituents.
Simulation Modeling Accomplishments:
- A prototype modeling approach that combines three existing models was developed and tested.
- The rural land use model (ADAPT) was enhanced and evaluated on a 534 square mile watershed. A comparison between observed and predicted total discharges was not quite as good as those obtained on a 37 square mile watershed in an earlier study with this model. However, the sediment and nitrate predictions were substantially better. A calibration and input parameter sensitivity analysis was performed with seven of the model inputs and observed data for the period 1990-1992. After calibration, the predicted total discharges went from being underpredicted by 14.0 percent to an overprediction of only 0.09 percent. However, most of the other statistics did not improve and even declined slightly. For the whole 6-year period (1990-1995), total discharges went from being underpredicted by 20.9 percent before calibration to being underpredicted by 8.8 percent after calibration.
- Only limited observed sediment load data were available. As an input parameter, the ADAPT model requires a sediment delivery ratio for the watershed. Procedures for determining the delivery ratio are empirical and rather approximate. Ideally, the sediment delivery ratio should be obtained from a calibration analysis. Due to the lack of available data, we estimated a value of 0.1 and then conducted an analysis using values of 0.08, 0.10, 0.12, and 0.45. A delivery ratio of 0.12 give an underprediction of 0.7 percent.
- Linking the rural land use model to a riparian zone model (REMM) resulted in a capability to evaluate the influence of riparian zones on the hydrologic response of watersheds. Developing the linkage proved more complex than anticipated and not all modeling issues have been resolved. The percentage of each subwatershed that are currently riparian zones varies from 0.5 percent to 2.1 percent. Based on these current natural settings, the upland flow that discharges into the riparian areas was varied from 0, 10, 15, 20, and 25 percent to evaluate their potential impacts on runoff and sediment. In a second analysis, the effect of increasing or decreasing the land base that are riparian zones was evaluated. The riparian area percentages evaluated were 0, the natural setting, and 5 percent. Again, the upland flow that discharges into the riparian areas was varied from 0, 10, 15, 20, 2 percent. As the percentage of the upland flow that discharges through the riparian areas was increased to 25 percent, there was an 11 percent decrease in the total discharge. Total sediment discharges showed an almost linear decrease of about 1 percent for each percentage increase in the upland flow that discharged through the riparian zones. When an evaluation was made of the benefit of increasing riparian zones to 5 percent of the total watershed area, there was an unexpected initial increase in the total discharge. This anomaly is probably due to a minor error in the code of the prototype methodology that we have developed. As the percentage of the total flow that discharged through a riparian zone increased, the total discharge at the outlet of the watershed decreased. When the total discharge into the riparian zone increased from 10 to 25 percent, the percentage reduction in the total flow at the outlet was 11.2 percent for the natural setting riparian zone percentages compared to 20.6 percent when 5 percent of the watershed was riparian zones. As expected, there was less sediment discharge for the various scenarios with the riparian zone percentage increased to 5 percent. Typically, for each scenario, the sediment discharges were 5 to 6 percent less than those predicted for the natural riparian zone setting compared to increasing the riparian zones to 5 percent of the watershed.
- A preliminary study was performed using the SWMM urban model to model urban areas and the rural ADAPT model to model other land uses. The purpose of this study was to evaluate the change in prediction accuracy associated with using a simple modeling approach (ADAPT without routing) compared to complex modeling approaches (ADAPT + SWMMs with routing) to estimate monthly flows and water quality at the outlet of a small watershed with mixed land uses. Statistics for the single model and the combined model simulations were almost identical (e.g., R2 values for both approaches of 0.62 for runoff, 0.90 for nitrate discharges, and 0.75 for soil losses). For the parameters considered, there was little benefit in using the urban model because urban land uses were less than 3 percent of the watershed area. In the next phase of our study, we will assess the influence of potential urban growth on the hydrology of this watershed. This work is scheduled to be completed by the middle of the year 2000.
- A study was conducted to evaluate if there is a need to use finer resolution Natural Resources Conservation Service MUIR data versus their STATSGO data to predict flow, sediment, and nutrient loads using the ADAPT model. The results showed that at a watershed scale, there was no significant difference between flow and sediment results obtained by using the two databases. For nutrients, there was a small (< 5%), but significant difference. The fact that the differences were small, and generally not significant, is important because there is a considerable reduction in time and resources needed to conduct studies of this nature if STATSGO data is used rather than MUIR data.
Journal Articles on this Report : 2 Displayed | Download in RIS Format
Other project views: | All 14 publications | 2 publications in selected types | All 2 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Gordon SI, Majumder S. Empirical stressor-response relationships for prospective risk analysis in the Eastern Cornbelt Plains ecoregion. Environmental Toxicology and Chemistry 2000;19(4):1106-1112. |
R824769 (1999) R824769 (Final) |
not available |
|
Jones AL, Gordon SI. From plan to practice: Implementing watershed-based strategies into local, state, and federal policy. Environmental Toxicology and Chemistry 2000;19(4 Pt 2):1136-1142. |
R824769 (1999) R824769 (Final) |
not available |
Supplemental Keywords:
ecological modeling, risk assessment, regional watershed models, riparian zone modeling, watershed biological diversity, water quality modeling, MUIR, STATSGO, sediment, ADAPT, SWMM., RFA, Scientific Discipline, Toxics, Geographic Area, Water, Ecosystem Protection/Environmental Exposure & Risk, Hydrology, Nutrients, Ecology, Water & Watershed, exploratory research environmental biology, Environmental Chemistry, Ecosystem/Assessment/Indicators, Chemical Mixtures - Environmental Exposure & Risk, Ecosystem Protection, Chemistry, pesticides, State, Ecological Effects - Environmental Exposure & Risk, Ecological Effects - Human Health, Wet Weather Flows, Biology, EPA Region, Watersheds, Ecological Indicators, aquatic ecosystem, agriculturally impacted watershed, anthropogenic stress, environmental monitoring, nutrient supply, nutrient transport, remote sensing, risk assessment, ecological risk assessment, riparian zone conditions, biological condition, field scale process model, human activities, physical model forecasts, streams, runoff, agricultural watershed, urban runoff, sediment, Georgia (GA), Cornbelt Plains Ecoregion, plant growth, statistical model, Ohio (OH), aquatic ecosystems, Index of Biological Integrity (IBI), remotely sensed data, Region 5, land use, stream ecosystem, agriculture , decision analysis tool, forecasting, land managementProgress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.