Tampa Bay Water Piloting Utility Modeling Applications Alison - PowerPoint PPT Presentation

Tampa Bay Water Piloting Utility Modeling Applications Alison Adams, Ph.D., P.E. Jeff Geurink, Ph.D., P.E. Tirusew Asefa, Ph.D., P.E. Workshop One December 1-3, 2010 San Francisco 1

Tampa Bay Water - Public Water Supplier for the Tampa Bay Region 2.4 Million Residents Served 220-250 mgd public supply annual average 2

Tampa Bay Water Multiple Types of Raw Water Sources Desal Water Ground Water Surface Water 15 BG Reservoir

Tampa Bay Water has Developed an Integrated and Diverse Water Supply System An integrated, flexible and diverse system that produces a su stainable and reliable water supply 4

Why Climate Variability is Important to Tampa Bay Water Seasonal Rainfall Pattern 9 8 7 Monthly Rainfall, inches 60% of the 6 annual rainfall in 4 5 months 4 3 2 1000 1 900 Hillsborough River Alafia River 0 Monthly Mean flow, cfs 800 JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC 700 600 500 400 300 200 100 0 5

Decision Making Actions Influenced by Climate Factors l Long range Planning (5 Years and beyond) – Demand forecasting / supply availability – Vulnerability assessments (reliability) – Long range water supply needs l Operational (Weekly to Annual) – Weekly forecasting demands and supply – Monthly / seasonal supply allocation – Annual budgeting process 6

Challenges and Issues l Acceptance by Board of Directors of vulnerability to climate changes l Climate ready regulations and regulators l Making good decisions with uncertainty l Embracing an adaptive management style of decision making l Customer acceptance of the agency’s efforts regarding water supply vulnerability 7

Integrated Hydrologic Model (IHM) Hydrologic Processes IHM Sequential Integration Read control file and data base, Start Launch processes Runoff, Surface/Vadose ET, Recharge, HSPF PET for Vadose/GW, Soil Moisture (Land Segments) Route land flows through reaches; Streamflow and Stage in reaches HSPF (Reaches) Cell values: SY, recharge & GW PET; Update RIV stages; Write EVT, RCH, H2M RIV packages, SY array MODFLOW Groundwater head and ET, Baseflow Land Segment values: LZS, LZSN, LZETP, INFILT(for saturation-excess); (HSPF) M2H Reach values: Baseflow, PET Coeff; Cell values: Mass Balance Flux HSPF (Reaches) Optional Second Reach Routing Time End End (MODFLOW) 8

Integrated Hydrologic Model Simulated Processes Surface-Water Processes Ground-Water Processes Non-Integrated Rainfall and potential ET input Lateral flow Abstraction storages Inter-aquifer leakage Interflow storage Well pumping Percolation Confined ground-water storage Interflow Spring flow Abstraction evapotranspiration Irrigation flux Surface-water diversions Level-pool reach routing Integrated Vadose zone storage Recharge Infiltration & redistribution of infiltration Flow exchange water bodies ↔ ground water Overland flow Unconfined ground-water storage Vadose zone evapotranspiration Ground-water evapotranspiration Reach evapotranspiration 9

Integrated Hydrologic Model Surface & Ground Water Interaction Legend Basin Boundary Grass/Pasture Grid Cell Forested Open Water Land Use Urban Wetlands Irrigated Mined/Disturbed 10

Integrated Northern Tampa Bay Model Surface-Water Component (HSPF) Convective Rainfall (4 months) l – 60% volume / 75% events – 1.25-mile event spatial scale 65% of basins with 2 mile radius l Rain input: 300 gauges, 15-min. l Potential Evapotranspiration l – minor spatial variation – 5x seasonal variation Average Annual Budget 1989-98 Flux Budget Term Percent (in/yr) Evap. & Transp. 69 38.0 Stream & Spring Q 21 11.0 Well Pumping 5 3.0 GW Flow to Gulf 3 1.5 SW Pumping 1 0.5 Other GW Outflows 1 0.5 11 Total 100 54.5

Integrated Northern Tampa Bay Model Ground-Water Component (MODFLOW) Ground-Water Component l – 95,000 active nodes – ¼ to 1-mile cell dimension – 85,000 water-body units – 8700 production wells Ground-Water Temporal Scale l – Sub-daily computation – Daily stress changes 12

INTB Calibrated Model & Climate Data Informed Decisions CYC-TMR-4d Land Surface Elevation 70 Observed l Regulations Simulated Scaled Simulated 65 60 Head, Feet l Operations 55 50 45 l Estimate ground water safe yield 40 1/1/89 1/1/91 1/1/93 1/1/95 1/1/97 1/1/99 1/1/01 1/1/03 1/1/05 Date – Climate and pumping variability – Regulation, infrastructure, mgmt constraints l Understand & compare uncertainty l Adaptive management strategies l New source assessments 13

Support Regulatory Requirements Define Drawdown & Assess Impacts Protection of other well owners l Single regulatory scenario for l rainfall & pumping (worst case) Regulations & regulators not ready l for climate variability assessments Historical well mitigation has l depended on rainfall magnitude Dry Years Wet Years 14

Optimize Well Pumping Distribution Unit Drawdown Response (Per 1 MGD) Drawdown response for 1 MGD well rate l Temporal and spatial convolution defines l total drawdown over time & space Historical climatic variability captured with l 1000 rainfall realizations Ensemble median drawdown response l 15

Ground Water Safe Yield Variability in Climate and Well Pumping Regulatory protection metrics 1000 rainfall realizations, 20 yrs l l – Wetlands & lakes (levels) Uncertainty in levels and flows l – Streams & springs (flow) Water-supply system reliability l Variability in climate and pumping l Well-High Variance Rainfall Well-Low Variance Streamflow 16

Tampa Bay Water’s Statistical Models Weekly Ensemble Stream Flow Forecast Models – Artificial Neural Network Based on Generalize Likelihood Estimation (GLUE) – Driven by recent weather, river flows, and groundwater levels – Input supply availability to Weekly Operations Model 0.8 Dry Normal Wet Dry Normal 0.7 Wet Seasonal Stream flow Models Resultant Dry 0.43 0.34 0.23 0.6 Probability Density 0.5 Normal 0.14 0.70 0.16 0.4 Wet 0.72 0.18 0.10 0.3 – Multivariate regression 0.2 Jan – Rainfall based on HMM 5 0.1 Normal Cum. from Mean (inches) Dry 0 – Three month to annual 0 2 4 6 8 10 12 14 16 18 20 0 Wet Rainfall, inches – Conditioned on recent weather -5 -10 -15 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

Central Florida Rainfall: “a Multi-scale Climatic Signal Measurement Device” 1 3 ENSO JFM La Nina Neutral El Nino M arProb Wet 0 45 55 21 Normal 17 48 35 45 Dry 61 36 3 34 2 M arProb 28 43 28 100 b) rainfall Wavelet Power Spectrum c) Global Wavelet Spectrum 2 AMO Filter 4 Period (years) 8 16 Cold Phase Warm Phase 32 64 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 0 20 40 Time (year) Power (mm 2 ) La Nina Neutral El Nino M arProb La Nina Neutral El Nino M arProb d) 4-8 yr Scale-average Time Series Wet 0 22 78 12 8 Wet 0 58 42 11 2 ) 10 Scale Av. Power ( mm Normal 23 41 36 21 8 Normal 12 56 32 24 6 Dry 52 48 0 24 Dry 82 9 9 10 4 2 M arProb 15 11 13 39 0 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 M arProb 12 31 15 58/ 59 Time (year) 18

Tampa Bay Water’s Seasonal Outlook Below Above Contingency Table Normal Normal Normal DJF 65 35 0 Climate Outlook JFM 85 12 3 & Real time observation Conditional Markov Rainfall Model Rainfall/ Runoff Model

Long-range Rainfall/ Runoff Simulation Models Three rainfall stations l – Two 106 years one 30 years, some 30 miles apart, 1095 square mile watershed Same model structure but two l parameter set – (October through May, and June through September) Lag 0 through 3 rainfall as wells l as lag 4, 12 month cumulative Non-parametric residual resample l to account process uncertainty Monthly to Daily disaggregation l – Conserve volume, intra and inter month daily flow continuity 20

Operational Modeling System Daily Ensemble Flow Simulations 21

Resilience, Reliability, and Vulnerability (RRV) Analysis Currently uses seasonal Hidden Markov rainfall l models (based on 106 years of data) – Future climate scenarios could replace this Analyze scenarios based on specific reliability and l vulnerability measures For each scenario, 1000 ensembles of 300yrs long l daily simulations Uses distributed computing over 64 Computer l cluster – a 2.5 day cluster run would have taken over 120 days on a single 8GB PC

Thank you Questions? 23