PDF version - Chesapeake Community Modeling Program
Transcription
PDF version - Chesapeake Community Modeling Program
Models of the Chesapeake Basin Kevin Sellner Executive Director Chesapeake Research Consortium www.chesapeake.org G. Brush Class, JHU 10/22/09 Outline • • • • • • • • General Air Watersheds Bay Physics Biogeochemistry Food web Nowcasts/Forecasts Climate Change? Hydrologic Budget 3 Need data to describe hydrologic processes in watershed B. Benham, VA Tech; brbenham@exchange.vt.edu Watershed Characterization Hydrography Soils Land Use Topography Pollutants Management Practices Elevation Watershed Boundaries Canopy Cover Surface Roughness Ponds Percent Impervious Erodibility Streams Cross-Sections Slope B. Benham, VA Tech; brbenham@exchange.vt.edu Rainfall 487 daily stations 192 hourly stations Monthly Regression of Latitude Longitude Altitude Daily Intercept 1984-2005 5 Lauren Hay – USGS NRP CMAQ Model The Community Multiscale Air Quality Model (CMAQ) has a domain that covers the North American continent at a 36 km x 36 km grid scale and is nested at a finer 12 km x 12 km grid scale over the Chesapeake watershed and Bay. 6 Atmospheric Deposition Estimates Combining a regression model of wetfall deposition... Jim Lynch, Jeff Grimm Penn State …with CMAQ estimates of dry deposition for the base… Robin Dennis, EPA/ NOAA …and using the power of the CMAQ model for 7 scenarios. 8 Important Considerations: Modeling Accuracy B. Benham, VA Tech; brbenham@exchange.vt.edu 9 Model Application Considerations B. Benham, VA Tech; brbenham@exchange.vt.edu SPARROW (SPAtially Referenced Regression on Watershed Attributes) Smith et al., Water Resour. Res., 1997 Monitoring Data Aquatic Landscape Terrestrial Landscape Predicted Flux, Concentration and Yield: Origin and Fate Spatial arrangement of land and water properties matters! Steve Preston [spreston@usgs.gov] Incremental Reach Watershed Use reach network to relate watershed data to monitored loads Load leaving the reach = Load generated within upstream reaches and transported to the reach via the stream network Load originating within the reach’s incremental watershed + and delivered to the reach segment SPARROW - Example Applications Identification of Spatial Distribution of Source Areas for Understanding Impacts on Receiving Water Bodies Chesapeake Bay SPARROW Model HOT SPOT ID! •uses a nonlinear regression model approach •annual output •source variables include point sources, urban area, fertilizer application, manure generation, and atmospheric deposition • other variables include soil permeability and instream-loss rates for four stream-reach classes HSPF (Hydrologic Simulation Program Fortran) Watershed Model Snapshot: Hourly output is summed over 10 years of hydrology to compare against other management scenarios Hourly Values: Rainfall Snowfall Temperature Evapotranspiration Wind Solar Radiation Dewpoint Cloud Cover HSPF Land Use Acreage BMPs Fertilizer Manure Atmospheric Deposition Point Sources Septic Loads 1991-2000 “Average Annual Flow-Adjusted Loads” 13 G. Shenk, gshenk@chesapeakebay.net Each segment consists of separately-modeled land uses • • • • • • • • High Density Pervious Urban High Density Impervious Urban Low Density Pervious Urban Low Density Impervious Urban Construction Extractive Wooded Disturbed Forest Plus Point Source and Septic • Corn/Soy/Wheat rotation (high till) • Corn/Soy/Wheat rotation (low till) • Other Crops • Alfalfa • Nursery • Pasture • Degraded Riparian Pasture • Animal Feeding Operations • Fertilized Hay • Unfertilized Hay – Nutrient management versions of the above 14 A software solution was devised that directs the appropriate water, nutrients, and sediment from each land use type within each land segment to each river segment External Transfer Module N, P, Sediment load Each land use type simulation is completely independent. Each river simulation is dependent on the local land use type simulations and the upstream river simulations. 15 • Goal Chesapeake Bay Watershed Modeling at ESSIC/UMCP Model water quality and quantity of the Chesapeake Bay Watershed (CBW) Provide river discharge, sediment, and nutrient forecast for the prominent tributaries in CBW • Approach Adopt a hydrologic and water quantity model – Soil and Water Assessment Tool (SWAT) for modeling CBW An independent SWAT model for each major river basin and some secondary river basins on the MD shores Pilot river basin: Rappahannock River basin Huan.Meng@noaa.gov • Rappahannock SWAT Model initial set-up requires information on topography, land use, soil, point and non-point nutrient sources, etc. Model calibration and validation using USGS observations of flow, sediment, and nutrient Model driving force: weather (solar radiation, temperature, precipitation, wind speed, relative humidity) Rappahannock SWAT has been successfully calibrated and validated for river discharge, sediment load, and nutrient loads. • Application – A component of the Chesapeake Bay Forecast System (CBFS) project at ESSIC/ UMCP - CBFS is an integrated atmosphere/ land/ocean (Bay) earth system – Provides 14-day forecast of river flow, sediment load , and nutrient loads for Rappahannock River on a daily basis. – The model output is used in the CBFS ocean model as input from the Rappahannock River basin. Huan.Meng@noaa.gov Penn State Integrated Hydrologic Model: PIHM Qu and Duffy 2007 • Modeling water movement • Soil saturation • Surface & Groundwater flows • Evapotranspiration • Output • Water Budget •Nutrient Loads •Nutrient Concentrations C. Duffy, cxd11@psu.edu Irregular Mesh & Stream Network Elevations Soil Classes Land Cover The Hydrochemical Model GWLF • Simple approach to landscape modeling – Lumped parameter model, monthly time step – Visual basic (fast, efficient on a PC) – Sufficiently detailed to mimic important processes • Publications – Lee et al. 2000. Biogeochem. 49: 143-173 – Lee et al. 2001. Biogeochem. 56: 311-348 – Fisher et al. 2006. Limnol. Oceanogr. 51: 435-447 – Used by others as well (Penn State, Howarth, etc.) • Capable of predicting annual export T. Fisher, UMCES-HPL; fisher@umces.edu GWLF validation %$ %&,"', %$,"(, (%,"'), !&$$$ Historical Export from Tilghman Island GWLF Output " &!* # %!!* " #!!$!!* # &!!'!!* $ #!!!* T. Fisher, UMCES-HPL AnnAGNPS: Uses NRCS Standards Annualized Agricultural Non-Point Source Pollution computer model. Processes Databases • Weather Generation - GEM • Soils - NASIS • Runoff – SCS Curve Number • Crops and Operations – Set by NRCS State Agronomists • Peak Runoff – TR-55 • Erosion - RUSLE • Sediment Delivery - HUSLE • HUWQ Databases – Fertilizer, Pesticides, Animal Wastes, etc. AnnAGNPS: the pollutant loading model Loadings by source of pollutant: • Cells (land areas)water, sediment, & chemicals. • Feedlotssoluble nutrients. • Gulliessediment and chemicals. • Point Sourceswater and chemicals. • Reachessediment yield & chemical transport. • Impoundmentssediment deposition. AnnAGNPS: source accounting 10% of outlet sediment from gully gully cell A cell B 80% of outlet pest X from cell C cell C feedlot 25% of outlet nitrogen from feedlot cell D cell F cell E watershed outlet SERC Empirical Models Regressions of Land Use & Nutrients " ! ! ! !! T. Jordan & D. Weller, SERC; jordant@si.edu, wellerd@si.edu Maximum average restoration benefit Piedmont 29% reduction 0.98 mg/l Coastal Plain 54% reduction 0.56 mg/l Current buffer Appalachian Mountains 15% reduction 0.29 mg/l Complete buffer Chesapeake Bay Program Decision Support System Land Use Change Model Management Actions Watershed Model Bay Model Criteria Assessment Procedures Scenario Builder Airshed Model Sparrow Effects Allocations 29 G. Shenk, gshenk@chesapeakebay.net Meta-Modeling with WOOOMM WOOOMM is an Online Web-based, multi-user environment Object-Oriented Robert.Burgholzer@deq.virginia.gov Model components are linked in object-oriented fashion, similar to Stella, PowerSim, or Simulink Meta-Model Has typical primitives: equations, graphs, and tables; and has special components for coupling environmental models Watershed Model Habitat Model Q, N Water Supply Q-W Stream Model Q, V, DO WOOOMM Details WOOOMM is Open Source Detailed Components And also Made of 100% Open Source Real-time data from USGS and NOAA Loose and Tight Coupling with HSPF Summarize data and Geo-process using PostgreSQL/PostGIS Statistical Analysis with R Currently Used For: Water Supply planning model for Commonwealth of Virginia Links CBP-HSPF model, USGS Gages, Habitat Suitability models from IFIM studies Multi-user capability allows planners, modelers and permit writers to develop and share model scenarios GW: ParFlow Conceptual Model Atmospheric Forcing Land Surface Flow Divide LSM LSM LSM LSM LSM Air Root Zone LSM Vadose Zone Water Table Vegetation LSM LSM Rout ed Wate r LSM LSM Flow s Line •Water table location is an output, driven by topography • Streams form where land is saturated Groundwater C. Welty, UMBC; welty@umbc.edu Source: R M Maxwell Inputs and Outputs • Inputs – Climate data – Soil and rock permeability, porosity – Land surface elevations • Outputs – Pressure head and degree of saturation (surface water, vadose zone, groundwater) – Derived outputs • Volumetric flow rate in streams • Water table delineation • Subsurface flow paths (using particle tracking code) Goal is not calibration but rather process understanding Kollet and Maxwell 2008 Watershed Model Input Data Chesapeake Bay Land Change Model (CBLCM) County Population Projections Growth Allocation Model Future Urban Area Sewer Service Areas Sewer Model Slope, Protected lands, Zoning, Priority Funding Areas Land Use/ Land Cover (NLCD) Calibration Metrics Cellular Automata Model 1990 and 2000 Impervious Surface P. Claggett, USGS; PClagget@chesapeakebay.net Sewer outflows Septic loads Proportions of Urban infill, New urban growth Forest loss, Farmland loss, Growth on septic, Growth on sewer Determining Proportions of Farmland and Forest Loss: Using SLEUTH Results: Salisbury, MD 2030Protected 1990 2000 Urban Land Impervious growth areas Cover Forecasted Urban Growth (2000 to 2030) Farmland and Forest Land Loss (2000 to 2030) Forecasted Population Growth: Sewer vs. Septic (2000 to 2030) Bay Physics Coupled hydrodynamic and water quality models J. Fitzpatrick, HydroQual; jfitzpatrick@hydroqual.com Bathymetry, Ocean Boundary, Tides, Winds, Ri ver Discharge all Critical H. Wang, VIMS; wang@vims.edu CH3D Circulation Model freshet dry Tracers are released from Susquehanna River 1 month ! ")( !"'*%!" & ' &% 4 month Typical events: 3 month longer for dissolved substances coming from SR enter BH during freshet event than during dry event Tracers are released from mid-bay B. Hong, VIMS; wang@vims.edu Tide Comparison (ELCIRC) Tidal elevation (m) red: NOAA tide Tolchester Beach, MD blue: model Lewisetta, VA CBBT, VA Eulerian-Lagrangian CIRCulation model J. Cho, VIMS; wang@vims.edu Semi-implicit Eulerian-Lagrangian Finite-Element model Salinity - Floyd (SELFE) M5 M3 Courtesy of Valle-Levinson J. Cho, VIMS; wang@vims.edu ANIMATION QUODDY Velocity Field Animation Velocity fields are used to visualize transport of materials, track & spread contaminants and aid navigation. Exchange of estuarine waters with the ocean is through the current patterns at the mouth of the bay. The flushing rate of the Bay determines its ability to discharge or retain pollutants. T. Gross, UNESCO; t.gross@unesco.org QUODDY Flapping Bay W. Long, UMCES; wenlong@umces.edu Temperature and salinity (2003) • CB5.3 Mid Bay Temperature simulation excellent Salinity simulation not as good R. Hood, UMCES; rhood@umces.edu NOAA CSDL CBOFS2 Set-up and Validation • Motivation : to upgrade present 2D barotropic (pressure alone determines gradients) NOS CBOFS implementation and produce water levels and fully 3D current, T, S fields for the public including ecologists • Validation sequence : (i) Constant density 3-D baroclinic (pressure, T, and sal determine gradient) simulation to validate tides - water levels and currents against predictions, (ii)Synoptic hindcast simulation (June 2003- September 2005) with full suite of forcings to validate water levels, currents, T and S against observed data; initialization and boundary T, S fields from climatology, river forcing from USGS and Chesapeake Bay Program, non-tidal open boundary water level forcing from NOS/CO-OPS observations and meteorological forcing from NARR blended with NOAA/NDBC buoy data for winds, air T, air P and DP/RH. (iii) real-time simulations to monitor model performance and validate against observations (in preparation for being fully operational); currently running x 4 daily and outputs being monitored and will become fully operational later in FY09 L. Lanerolle, NOA/NOS/CSDL; lyon.lanerolle@noaa.gov Chesapeake POM Guo & Valle-Levinson 2008 Princeton Ocean Model: International ocean modeling software, available through the web. Used throughout the open ocean community http://www.aos.princeton.edu/WWWPUBLIC/htdocs.pom/ ADCIRC • Vertically-integrated continuity equation for water surface elevation • Models tides and wind driven circulation, analysis of hurricane storm surge and flooding, dredging feasibility and material disposal studies, larval transport studies, & near shore marine operations. • Used for FEMA CB flood mapping NOAA CSDL uses ADCIRC for inundation/storm surge, the VDATUM project (tidal datums), and for generating tidal databases containing harmonic constituents for water levels and currents. R. Luettich, UNC; rick_luettich@unc.edu CHIMP: ROMS-based Circulation Webpage http://www.d.umn.edu/~jaustin/CHIMP.html Fig. 3. Plan views. (a) Water height levels displayed via color mapping. (b) Velocity direction vectors with color mapped magnitudes (Crouch et al. 2008). M. Dinniman, ODU-CCPO; msd@ccpo.odu.edu Tidal Biogeochemistry Plankton & Impact of Hurricane Isabel Unusual fall bloom observed after Isabel The model can be used to understand how hurricanes affect plankton production and hypoxia in Chesapeake Bay. M. Li, UMCES; mingli@umces.edu ChesROMS Biogeochemical Model - Work in Progress (chief architect Dr. Jerry Wiggert) Based on Fennel et al. core model bundled with ROMS NPZD type model with oxic sediment denitrification Adding DON ISS loading Atmospheric N deposition Anoxic benthic denitrification Water column denitrification R. Hood, UMCES; rhood@umces.edu Longitudinal Section of Patuxent Showing Region Boundaries and Fluxes Between Regions (Boxes) Upper Estuary Middle Estuary Lower Estuary (Hagy et al. 2000) Use flow balance and salt balance equations for each box to compute unknown values for water flows (Q’s) and mixing rates (E’s), given salinities and FW inputs; also used for main Bay (Hagy et al. 2000). J. Testa, UMCES; jtesta@umces.edu Box Model computations of net production of oxygen, inorganic nitrogen, and dissolved silicate in both the surface and bottom layer of the Patuxent estuary “Sweet spot Model” Apply classical engineering river model (Streeter-Phelps DO-sag) Scavia et al. pers. comm.; scavia@umich.edu Application to ChesBay Susquehanna Load ~N Pycnocline Diffusion v v B D Advection Organic matter decay Model Calibration R2=0.73 Model Data Although some think of SOD and nutrient fluxes as internal sources in a estuary or lake mass balance, they originate from external sources. The sediment flux model (SFM) considers the deposition of organic matter (phytoplankton & detritus), its subsequent decomposition (diagenesis), burial of the more refractory components to the deep sediment bed, and the flux of resulting end-products back to the overlying water column. J. Fitzpatrick, HydroQual; jfitzpatrick@hydroqual.com Importance of overlying water column dissolved oxygen on sediment phosphate release … under conditions of low dissolved oxygen PO4 fluxes increase. 1985-1988 SONE data (Boynton, Kemp, et.al. (circles) vs. model computations Tidal Food Web Coupled hydrodynamic and water quality models J. Fitzpatrick, HydroQual; jfitzpatrick@hydroqual.com Coupling of nutrients (N, P, Si), primary production, and dissolved oxygen in the water column and the sediment bed Chesapeake Bay Regional Estuarine Ecology Model (CBREEM) • • • Purpose – Generate historical patterns in primary productivity for EwE Introduction & Methods – Two layer, simple hydrographic model (monthly time steps for 50+ years) – Use wind, rainfall, gage inflow, and relative loading as inputs – Solve for equilibrium velocity fields on Richardson grids and make chemical mass-balanced calculations (Wright et al. 1986, Hunter and Hearn 1991) Results – Chla (used as nutrient loading forcing function for EwE) NOAA Chesapeake Bay Office Inorganic Carbon Three Algal Groups Respiration Respiration Microzooplankton Respiration Dissolved Organic Carbon External Loads Labile Particulate Organic Carbon In CBP WQ Model: Carbon Cycle Refractory Particulate Organic Carbon C. Cerco, ACE; Carl.F.Cerco@usace.army.mil Mesozooplankton Sediments Larval Pools Oyster Larvae Pelagic Prey Fish Bay Anchovy Anchovy Larvae Ctenophore Larvae Atlantic Menhaden Gelatinous Zooplankton Zooplankton Sea Nettles Ctenophores Acartia tonsa Microzoopnktn Phytoplankton < 2 microns Pelagic Bacteria 2-4 microns HNAN 4-10 microns Non reef fish Benthic Reef-assoc. fish Oysters 10-50 mic On-reef inverts 50-100 mic Off-reef inverts > 100 microns Benthic Bacteria DOC N P POC Si Detritus Pools Fulford et al. Ecological Applications in press Algal Speciation Model of the Potomac Portion of the Chesapeake Bay WQSTM • Study Objectives – Refine and improve representation of dominant algal groups in the Potomac – Calibrate the revised Potomac portion of the WQSTM • Study Tasks – Literature review – Data analysis – Development of new algal speciation sub-model – Calibrate Potomac portion of WQSTM Amanda Flynn, LimnoTech; aflynn@limno.com Algal Speciation Model (continued) • Revised model contains five (5) algal groups • Calibration runs were used to: – understand the physical, chemical, and biological dynamics represented in the model – determine model sensitivity to changes in key input parameters – improve representation of algal groups • Model Output – algal biomass, primary production – water quality (N, P, DO, Light) • Revised Potomac model can/will be used to address local WQ issues and serve as a research testbed for the WQSTM SAV Model Interactions CH3D Hydrodynamic Model WindGenerated Wave Model Bottom Shear Stress, Currents SAV Model Optical Model ROMS Bed Model Resuspension ICM Eutrophication Model SAV Unit Model •Computes SAV density (mass/unit area) as a function of irradiance and nutrients. •Irradiance and epiphytes calculated separately. •Interacts with water column and bed sediments. SAV Sub-Grid •SAV is computed on a sub-grid independent of the hydrodynamic grid. •Sub-grid areas are based on observed SAV beds rather than arbitrary computational elements. •Sub-grid elements permit refined depth increments for computation of available light. Oysters excretion Filter Feeders feeding Particulate Organic Matter Deposit Feeders diagenesis excretion diagenesis Dissolved Oxygen sediment-oxygen demand biodeposits Sediments Dissolved Nutrients sediment-water exchange Water Column respiration filtration settling Particulate Organic Matter Dissolved Nutrients Oxygen Demand respiration Diagenesis Model with Benthos System-Wide Summary Larval Pools Oyster Larvae Pelagic Prey Fish Bay Anchovy Anchovy Larvae Ctenophore Larvae Atlantic Menhaden Gelatinous Zooplankton Zooplankton Sea Nettles Ctenophores Acartia tonsa Microzoopnktn Phytoplankton < 2 microns Pelagic Bacteria 2-4 microns HNAN 4-10 microns Non reef fish Benthic Reef-assoc. fish Oysters 10-50 mic On-reef inverts 50-100 mic Off-reef inverts > 100 microns Benthic Bacteria DOC N P POC Si Detritus Pools Fulford et al. Ecological Applications in press R. Fulford, USM; Richard.Fulford@usm.edu Menhaden: Prey, Predation Story Menhaden Menhaden Controls X. Zhang, NOAA Oxford Lab; xinsheng.zhang@noaa.gov Menhaden There do seem to be positive relationships between menhaden recruitment and bottom-up (i.e. prey quality & quantity) variables. Chlorophyll seems be better than annually integrated primary production in most of fits to date. Data from Houde, Harding, and Annis 2006. Progress Report to NOAA Chesapeake Bay Office Menhaden Modeling in Chesapeake Bay • Research Question: Are Atlantic menhaden (a planktivorous fish) impacting water quality? • Model formulation: Monte Carlo simulation of seasonal migration in and out of bay • Consumption (algae, zooplankton, detritus) based on food availability (in water quality), fish output based on bioenergetics formulation • Conserve mass (fish outputs, mass from mortality returns to water quality model) S. Daylander, ACE; Patricia.A.Dalyander@usace.army.mil Menhaden Modeling in Chesapeake Bay Effects of Fish Population • Algae, fish both seasonally variant • More fish, more nutrient recycling (like fertilizer), faster algal growth, BUT at some point they eat so much algae can’t “keep up” • Also has annual menhaden school movements animated as well as annual changes in menhaden-induced longitudinal chlorophyll in the bay Multi-Layered Bioenergetics Modeling Fully Coupled NPZ-Menhaden-Piscivore Model M.J. Brush, VIMS; brush@vims.edu Time to filter the … Entire bay Source: M.J. Brush Surface mixed layer Effect of Juvenile + Adult Consumption with Juvenile + Adult N excretion Source: M.J. Brush Fig. 7. As for Figs. 4-6, but effect of adding juvenile and adult menhaden consumption and both juvenile and adult N excretion to the NPZ model. )2*/)$#/! ( ( 1! + %* /444*0../, !+!*/443*0../, + , (+, (+,% & ( +, * & (%( %# ) ! $! )! # " ) #! !"+!", !+, ! &+#(,! ) Source: M.J. Brush Next steps: Climate Influence X. Zhang, NOAA Oxford Lab; xinsheng.zhang@noaa.gov Modeling Striped Bass Habitat Suitability Habitat Requirements Habitat Suitability Index Response Scale Functions Combine into habitat suitability index Decision rules Interpolated Data Response functions based on Bain & Bain , 1982 X. Zhang, NOAA Oxford Lab; xinsheng.zhang@noaa.gov High temperature-low oxygen habitat squeeze during July 1985 1997 1986 1998 1987 1999 1988 2000 1989 2001 1990 2002 1991 2003 1992 2004 1993 2005 1994 2006 1995 1996 Responses to high summer surface temperature & low bottom DO vary among striped bass & menhaden { { Habitat Good July 1996 Greater Habitat Overlap { July 1999 Smaller Habitat Overlap Bad { S. Brandt, OSU; Stephen.Brandt@oregonstate.edu SB Larvae Habitat Suitability at Peak Spawning Season May 1996 May 1999 X. Zhang, NOAA Oxford Lab; xinsheng.zhang@noaa.gov Larval Pools Oyster Larvae Pelagic Prey Fish Bay Anchovy Anchovy Larvae Ctenophore Larvae Atlantic Menhaden Gelatinous Zooplankton Zooplankton Sea Nettles Ctenophores Acartia tonsa Microzoopnktn Phytoplankton < 2 microns Pelagic Bacteria 2-4 microns HNAN 4-10 microns Non reef fish Benthic Reef-assoc. fish Oysters 10-50 mic On-reef inverts 50-100 mic Off-reef inverts > 100 microns Benthic Bacteria DOC N P POC Si Detritus Pools Fulford et al. Ecological Applications in press R. Fulford, USM; Richard.Fulford@usm.edu Oyster Impact on Summer Primary Production Oyster Impact on Summer Secondary Production CB Fisheries Ecosystem Model • Developed in cooperation between NOAA CBO, CRC, UBC with support from many bay researchers using Ecopath with Ecosim software (code base) • A companion to the CB Fisheries Ecosystem Plan • Technical report (230 p) completed/in press • Chesapeake Bay tidal waters • 45 functional groups • Replicates ecosystem history 1950 – present H. Townsend, NCBO; howard.townsend@noaa.gov NOAA Chesapeake Bay Office Bacteria/Pathogen Modeling: Models from Other Regions Background • Meteorological, physical, and environmental data • Useful for predictive model development H. Kelsey, UMCES; Heath.Kelsey@noaa.gov Nowcasts/Forecasts ChesROMS Operational Modeling NCBO stable prediction of Sea nettles, K. veneficum, & V. cholerae since January 2008 http://155.206.18.162/cbay_hab/; http://155.206.18.162/vvul/; W. Long, UMCES; wenlong@umces.edu Ecological Forecasts (Sea Nettles): T and S strongly constrain sea nettle distributions Estimate T and S using ChesROMS Provides input to an empirical logit model that predicts probability of sea nettle occurrence R. Hood, UMCES; rhood@umces.edu A fuzzy logic approach to predicting Prorocentrum minimum blooms • • Fuzzy logic: Fuzzy logic addresses the uncertainty of descriptive categories (e.g. “hot”, “cold”, “warm”, or “low”, “medium”, “high”) by allowing a parameter’s value to exist in multiple categories simultaneously, and to varying degrees. Objectives: – Examine the success of a hybrid fuzzy logic and decision tree model for predicting harmful algal blooms – Provide better predictive success than traditional statistics – Provide a reliable, robust forecasting tool for managers J. Anderson, MSU; jon.anderson@morgan.edu Fuzzy Decision Tree for Station CB3.3C Environmental conditions are 1.5 months prior to prediction time-point J. Goodwin, UMCES; jgoodwin@hpl.umces.edu Oyster Habitat Model Input/Output • Monthly spatial maps of modeled oyster growth (from 1985 – current) based on salinity, dissolved oxygen and temperature (see http://chesapeakebay.noaa.gov/IEA/Oyster.htm) for individual plots) • Trends over time in changes of modeled oyster growth (Figure A) Potential • Compare growth model trends with other factors important to oyster survival and growth (e.g., disease, substrate) using a GIS platform (e.g. Historical Yates Bars/substrate w/ model output on Figure B) LTRANS (Lagrangian Transport Model) Iteratively (repeatedly) solve for the new position at each time step to develop trajectories of particle motion t=2 t=1 t=3 t=4 t=5 E. North, UMCES; enorth@umces.edu Percent settled particles 0 – 10% 10 – 20 20 – 30 30 – 40 40 – 50 50 – 60 60 – 70 70 – 80 80 – 90 90 - 100 There are clear spatial patterns in transport success When combined with demographics, they could be used to guide restoration and management • To design, develop, test, and implement an agent-based simulation model in a multiplayer “game” format. – The research objective was to model and simulate a complex environmental system composed of many interrelated subsystems, with no central coordinating authority, with many independent decisionmaking entities (agents), and where the behavior of the complex system displays unpredictable, emergent outcomes – The agent-based simulation had to instantiate a large number software agents and a smaller number of “live” agents – game players – in various decision-making roles – http://www.virginia.edu/vpr/baygame.html © Copyright 2009 The Rector and Visitors of the University of Virginia G. Learmonth, jl5c@virginia.edu 105 Game Model © Copyright 2009 The Rector and Visitors of the University of Virginia 106 Crab fishery © Copyright 2009 The Rector and Visitors of the University of Virginia 107 April Test Result • The test of the Bay Game presented a realistic scenario for a 20-year planning horizon • The Bay Health Index responded correctly to the decisions made by the “live” agent/ players and the software agents – they “saved the Bay” • Players were not constrained to behave in their own best interests, e.g., some went bankrupt! © Copyright 2009 The Rector and Visitors of the University of Virginia 108 Chesapeake Inundation Prediction System (CIPS) – Part of the Chesapeake Bay Observing System • High resolution visualization of storm surge and rive flooding of specific areas • Much finer resolution than current NOAA operational model SLOSH • Goal: Another flooding tool for NOAA’s storm ‘tool box’ B. Stamey, Noblis, Inc.; barry.stamey@noblis.org The intersection of Union and King Streets in Old Town Alexandria, VA )$$" % /31140-%"&#", %# )&&$* /0- #$%&# %"%! *% %"&#")%$*'#" 25&&$ ) ! %&#&#! $ & &$ # !/0 - . Stre et #!# King Lee S treet $!!"*-,,.+(%!' ##! #!& !"# "" #!" *+ ( ) ) +4 !#"!%!#+ %# % %# % (## $153362/* #. !!"%143%#"*!#,! %#$!(&! 2 !#$%!$ $ / 0 Hurricane Isabel SLR=0.0m Hurricane Isabel SLR=0.5m Hurricane Isabel SLR=1.0m Hurricane Isabel SLR=1.5m Flood Depth vs Sea Level Rise National Airport 2 200cm 150cm 1.8 100cm 50cm 1.6 0 1.4 Meters 1.2 1 0.8 0.6 0.4 0.2 0 0 20 40 60 Hours 80 100 120 Thank you! Kevin Sellner Chesapeake Research Consortium www.chesapeake.org sellnerk@si.edu