 |
||||
|
||||
|
|
|
 |
||
|
||||
      |
GLERL 2003 Milestone Reports
GLERL 2003 Milestone Home
GOAL 1: Protect, Restore and Manage Use of Ocean and Coastal Resources Through Ecosystem Management Approaches
OBJECTIVE: Increase ocean and coastal areas explored, mapped, characterized and inventoried.
Specific Strategy: Develop watershed models in the southeast and Great Lakes Region.
Milestone: Apply GLERL's Large Basin Runoff Model in a distributed-parameter fashion for the Kalamazoo River Basin and calibrate and adjust the model to account for difference in soil characteristics.
Scientist: Thomas E. Croley II, GLERL
Purpose: Since macro scale operational models of watershed hydrology link the research community with water management policy and decision support institutions, the Great Lakes Environmental Research Laboratory (GLERL) identified issues associated with applying a macro scale model to the micro scale by looking at a specific case study, the Kalamazoo River watershed.
Efforts: GLERL modified their Large Basin Runoff Model (LBRM) continuity equations to allow up-stream inflow when the model is applied to a single cell within a watershed. They found the modifications in terms of corrector equations to be applied to the original equation solution. They also changed the LBRM to use independent actual and potential evapotranspiration (appropriate for micro scale use) instead of complementary actual and potential (appropriate for the macro scale). GLERL found that linear reservoir coefficients for moisture storages of the upper and lower soil zones and the groundwater zone can be used across scales in initial approximations. However, initial approximations of surface storage coefficients for a network of cells are related to a macro scale coefficient through consideration of a cascade of linear reservoirs and characteristic travel times.
GLERL organized LBRM applications to constituent watershed cells in a flow network by identifying the network flow cascade and then automatically arranging the cell computations accordingly. They identified required characteristics of any flow network map and designed system checks to guarantee them. These characteristics include the presence of a unique watershed outlet cell and the absence of flow loops within the watershed. They devised a network cell numbering and coding scheme for these checks and for subsequently ordering LBRM computations and routing flows throughout the watershed. Finally, they outlined the computer implementation for the micro scale distributed LBRM and refined the implementation by application to the 5,612 one square kilometer cells of the Kalamazoo River watershed. GLERL divided the watershed into a grid of one square kilometer cells and assembled data for each cell on elevation, slope, land cover, flow roughness, upper soil zone depth, upper soil permeability, lower soil zone depth, and lower soil permeability; see Figure 1. They used the data to estimate spatial variability of model parameters as they calibrated the (spatial) mean values by matching observed Kalamazoo River flows.
| Elevation (m) | Slope (%) |
 |
 |
| Land Cover | Manning's n |
 |
 |
| USZ Depth (m) | LSZ Depth (m) |
 |
 |
| USZ Permeability (m) | LSZ Permeability (m) |
 |
 |
Figure 1. Kalamazoo River Watershed Parameters (one square kilometer resolution)
Customers: Engineering firms, government agencies, water managers, hydrological forecasters, and other decision-makers, will find the distributed LBRM useful in analyzing spatial storm impacts, movements of conservative substances from the watershed surface into the waterways, and changes in land use/cover and soils. The model is currently not appropriate for flooding predictions as insufficient data exist to use the model at smaller time increments than a day.
Significance: The best-fit distributed-parameter model uses independent evapotranspiration and potential evapotranspiration (as opposed to complementary), which is appropriate for consideration of small-scale processes. It also uses a linear surface storage coefficient proportional to the square root of cell slope and inversely proportional to Manning's roughness. Upper soil zone percolation to the lower soil zone and lower soil zone interflow to the surface are described with linear reservoir coefficients proportional to upper soil permeability. Lower soil zone deep percolation to the groundwater zone and groundwater flow to the surface are described with linear reservoir coefficients proportional to lower soil zone permeability. Distributed-parameter model calibrations yield higher root mean square errors between observed and modeled Kalamazoo River flows than do lumped-parameter model applications. However, inspection of hydrographs reveals that the distributed-parameter model did a better job than the lumped-parameter model in matching variations in hydrograph recessions; see Figure 2.
Figure 2. Selected 1950-1951 Model Comparisons With Actual Kalamazoo River Outflows.
Next steps: This work provides the foundation for applications to other areas and continued model development. Application to the Maumee River is currently underway. Several model extensions appear logical and include spatial routing of groundwater, interflow, or upper soil zone flows.
Last updated: September 2, 2003 mbl