Development of Aquatic Ecosystem Models Lizhu Wang, Shaw Lacy, Paul Seebach, Mike Wiley Institute for Fisheries Research MDNR and U of M
Project Objectives Develop statewide models for predicting stream flows and summer water temperature; Develop models describing relationships among base flow, water temperature, and trout population; Test model sensitivity by incorporating site- specific data from Augusta Creek; Evaluate effectiveness of catchment BMPs in offsetting ground water withdraw.
TROUT FLOW TEMPERATURE The Key Step: Describe relationships among flow, temperature, and trout
Build Statewide GIS Databases Identify stream measuring unit
Michigan 38,000 inter-confluence stream reaches 1:100,000 NHD
Build Statewide GIS Databases Delineate catchment boundaries for each unit
Reaches, Watersheds, Riparian Zones, and Upstream Catchments
Upstream Catchment Context
Build Statewide GIS Databases Synthesize landscape & instream characteristics for each stream unit
Reach of Interest
Synthesize Landscape Data Land use/land cover Surficial geology – texture & formation Soil permeability Bedrock depth & geology Average annual growing degree days Average annual precipitation Air temperature Ground water delivery potential – Darcy Land use transformation model output
Synthesize Network & Instream Data Arc gradient Catchment gradient Arc sinuosity Arc stream order, linkage number Arc fragmentation by dams and linkages with lakes and large rivers
Variables and Scales The database contain about 300 variables. The data are organized into 5 scales Channel Riparian (arc) Riparian (entire) Watershed (arc) Watershed (entire) From Paul Steen, 2004
Build Sampled Databases Link sampled flow, temperature, and trout data with each stream unit
Develop Summer Water Temperature Model Collected continuous summer water temperature from about 500 sites throughout Michigan. Developed geo-statistical and generalized additive regression models for July mean ---- Temperature kriging + residual modeling (Mallow’s Cp, Min AIC). Model explains 77% variance.
Michigan Stream Temp. Predictions 26.0º C 9.0º C
Flow Models Multiple regression models for predicting annual and August 5%, 10%, 25%, 50%, 75%, 90%, 95% exceeding flows; Models explain 78 – 97% variance. Based on 83 sites that have >20 years flow data
Annual Flow Predictions 90% Flow Yield
Stream Sites Sampled for Fish Community Based on fish data from about 1500 sites
July mean water temperature o C Link flow, Temperature, and Trout (hypothetical model) Cold Marginal Typical
ColdTypicalMarginal
July mean water temperature o C Link flow, Temperature, and Trout (hypothetical model) Cold Marginal Typical
ColdTypicalMarginal
Augusta Creek Watershed Max July Temp Difference 1)± 4°C (± 7.2°F) 2)± 7°C (± 12.6°F) 3)± 4°C (± 7.2°F) 4)± 11.5°C (± 20.7°F) 5)± 5°C (± 9°F) 6)± 10.5°C (± 18.9°F) 7)± 6.5°C (± 11.7°F) 8)± 8°C (± 14.4°F) 9)± 7.5°C (± 13.5°F) (trib) Gull Lake N
Summary Identified stream measuring unit; Delineated catchment boundaries of each unit; Synthesized landscape & instream characteristics of each unit; Linked sampled data for trout, flow, and tempertaure with each stream unit. Developed models for prediction of flow and temperature. Partially built models linking trout, flow, and temperatur. Collected temperature and fish data from Augusta Creek.