Solutions for Hydrologic Analysis, Real-time Prediction, and Management of Stormwater
Baxter E. Vieux Ph.D., P.E. Professor, School of Civil Engineering and Environmental Science Director, Natural Hazards and Disaster Research Center National Weather Center University of Oklahoma Norman, OK 73072 -- [email protected]
1 Overview
Technological innovation is creating unprecedented opportunities to apply information technology, gain new data from advanced sensor systems, and develop more effective urban infrastructure. Radar detection of precipitation and distributed modeling is a rapidly evolving technology that holds promise in making real-time predictions and for solving stormwater and drainage problems. Two key areas continue to present challenges: 1) representative precipitation over watershed areas in real-time; and 2) spatially distributed hydrologic prediction of stage and flow rates.
2 Hydrologic Prediction
3 Continental NEXRAD coverage
4 Rain gauge and radar
Radar rainfall— Radar rainfall measured by reflectivity depends on the number of drops and distribution of sizes Z=300R1.2 Z=300R1.4
, Z Rain gauge amounts depend ity Rainfall on time rate of arrival of each ctiv fle drop and its size Re Rate, R
Bias correction with quality controlled gauges 5 Combining strengths
Radar
Rain Gauge
Better rainfall estimates than either system alone
6 Quantitative Point Measurement Map of Intensities Radar considerations
Rain gauges – Supply gauge only products Gauge correction for bias Gauges must be quality controlled Radar affected by – Updrafts/downdrafts Partial beam filling and blockages Bright banding Anomalous propagation Clutter suppression Low precision
7 Vflo™ Physics-Based Approach
h uh)( IR t x
Rainfall Distributed rainfall input Runon Channel/overland Hydraulics Drainage network Runon Soil Infiltration/Impervious Land Cover Runoff Antecedent Moisture Water Control Structures Runon Recharge/channel losses Infiltration
Hydraulics=Hydrology 8 Scour Critical Bridges ODOT/FHWA
Which bridge are at risk of scour? 9 Radar Precipitation Remnants TD Erin - August18-19, 2007
10 Radar Rainfall Totals Remnants TD Erin - August18-19, 2007
Daily total for 2007-08-18 CDT
Daily total for 2007-08-19 CDT Mean Bias Correction = 1.748 Average Difference = ± 11%
Without correction: 1 inch
would be 0.56 inches11 Real-time Monitoring of Scour Risk
12 Infrastructure Operations and Management
Simultaneous simulation at 134 bridge locations with Real-time Vflo™ Accounts for bridge geometry and channel profile tracking
13 Recent Activity
14 NWS Headwaters Routing Remnants TD Erin - August18-19, 2007
Real-time Vflo (9202 cells @ 300x300 meters) Radar precipitation input
Distributed model with interior watch points
Kingfisher @ US 81 (320 sq. mi.)
Kingfisher Cr. @ SH 33 (Two Locations)
Winter Camp Cr @ SH33
15 Rainfall Totals Remnants TD Erin - August18-19, 2007
0.5 6
0.45 Rain (in) 5 0.4 Rainfall Over Kingfisher Accum (in) 0.35 4 0.3 0.25 3
0.2 2 0.15
0.1 1 0.05
0 0 0:00:00 1:00:00 2:00:00 3:00:00 4:00:00 5:00:00 6:00:00 7:00:00 8:00:00 9:00:00 10:00:00 11:00:00 12:00:00 Gage = 5.66 in
Rainfall Depth (in) Value High : 15 5 Miles Low : 1 Radar Storm Total Kingfisher @ SH 33 Aug 19 00:00-12:00 UTC 16 Radar Average = 8.28 in NWS Headwaters Forecast March 4, 2004
Reported Crest: 23.5 feet
17 Urban Flood Forecasting
Background Brays Bayou subject to intense and prolonged precipitation from tropical storms, continental fronts and coastal convergence Relies on radar-based flood forecasting since 1997 to take flood protection measures Texas Medical Center/Rice Hydrologic lead-time University, Houston Texas provides critical information for emergency management actions
18 Brays Bayou Model Setup
Brays Bayou Vflo Model 120x120m Area =100mi2 (260 km2)
19 Hydraulics
Action Level
JuneJune 8, 2004,2004, observed observed and and Vflo™ Vflo™ simulated simulated hydrographs hydrographs at the atHarris the HarrisGully outlet Gully outlet • 1x1 km NEXRAD 4.5 4.5 4.0 4.0 • Vflo 3.5 3.5 40x40m 3.0 3.0 2.52.5 • Harris Stage (ft) Stage Stage (ft)2.02.0
1.51.5 Gully 1.0 1.0 2 0.5 10 km 0.5 0.0 0.0 20
7:12 9:36 12:00 14:24 16:48 19:12 21:36 7:12 9:36 12:00Observed Simulated 14:24 16:48 19:12 21:36 Observed Simulated Time (CDT) Time (CDT) Radar Input TS Allison
Scatter Plot of Calibrated RG Pairs
400.0
Ri* = 1.00 * Gi R2 = 0.97 300.0
10 mm 500 200.0 Radar (mm) 100.0
0.0 0.0 100.0 200.0 300.0 400.0 Gauge (mm)
Allison 3rd
21 Streamflow Validation of QPE
Brays Bayou is highly urbanized with impervious Radar to Stream Gauge Volume 200 areas and clay soils
180 Adjusted Infiltration averages 7% of 160 y = 1.076x 2 140 R = 0.9646 rainfall ) 120 Streamflow Validation 100
80 Adjusted: Slope 1.076 2 60 Unadjusted (+) R =0.96 Radar (mm Rainfall Volume y = 1.1003x 40 2 R = 0.2129 Unadjusted: slope 1.10, 20 0 50 100 150 200 2 0 Stream Gauge Volume (mm) R =0.2129
Observed Streamflow and Radar Rainfall input comparison (R2=0.96)
Vieux, B.E., and P.B. Bedient, 2004. Assessing urban hydrologic prediction accuracy through22 event reconstruction. J. of Hydrology. 299(3-4), pp. 217-236. Special Issue on Urban Hydrology. Real-time Stage Prediction
23 Time Series of Bias Correction Factors, F
4.5 Mean 4 field bias 3.5 factor, F 3 for each 2.5 6-hr 2 moving 1.5 window.
Mean Field Bias Field Mean 1 0.5 Gi 0 FMR Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan Ri 2005 2006 2007 24 25 26 27 Rainfall Distribution
28 29 30 Flood Early Warning System City of Austin, TX
Radar and rain gauge integration Observed and simulated stage displayed Cell-phone notification system based on user requested thresholds
31 Vflo Shoal Creek Austin, TX
32 Continuous Model Results
16 14 Observed 12 Simulated 10 8 6
Stage (feet) 4 2 0 Jan-13 Mar-12 Mar-13 May-3 Events 2007
33 Notifications
Audible reception of Text/email messages Icons and Blinking Icons with mouse-over messages Threshold with color change and action list Hydrographs Inundation maps
34 Vflo™ Predicted Inundation in GoogleEarth
35 December 2, 2008 35 Maximum Predicted Inundation
36 36 37 38 39 40 41 42 Summary
Radar detection of precipitation and distributed modeling is a rapidly evolving technology that holds promise in making real-time predictions, and for solving Claudette in Vflo Interface stormwater and drainage problems.
43