Date post: | 25-Dec-2015 |
Category: |
Documents |
Upload: | barbara-joseph |
View: | 219 times |
Download: | 3 times |
Using LiDAR, “WATER”, and TOPMODEL
TOPO-DRIVEN HYDROLOGY
Carey Johnson, KY Division of Water State CTP Program Manager Has led Kentucky through MapMod for all
120 counties in the Commonwealth
Trevor Timberlake, URS Water Resources Engineer Hydrology/Hydraulics Modeling, Project
Management FEMA, Dams, Transportation
INTRODUCTIONS
Kentucky’s Approach to RiskMAPWhat is “WATER”?
The WATER Software TOPMODEL The Variable Source Area Concept
Our Study Objectives Methods
Results & Conclusions Results of Analysis Conclusions Direction for Moving Forward
OVERVIEW
Better Data, Strong Partnerships, and Better Answers
KENTUCKY’S APPROACH TO
RISKMAP
Better Input = Better OutputPart of the RiskMAP Vision
“High-quality elevation data form the foundation for increasing the quality of the flood maps…” – FEMA’s RiskMAP Report to Congress (3/15/11)
Acquisition of LiDAR Acquired for portions of Kentucky Has been a catalyst for new Statewide LiDAR acquisition
BETTER DATA
LIDAR ACQUISITION
Find“win-win” relationshipsUSGS and KDOW
Development of WATER tool, funding from KDOW
Gage DataNWS and KDOW
KDOW’s contribution to the Ohio River Community Model
Receipt of NexRAD data for entire OHRFC area
STRONG PARTNERSHIPS
Credibility The Floodplain Managers, City Engineers, etc. The Public
Should regression equations be the default answer? What about ungaged, unmodeled areas?What about a software package:
…developed by a partnership between KDOW and USGS …utilizing NWS data ...incorporating a model “driven” by topography …tested in an area for which we have LiDAR data …that can yield “model-based” results …effi ciently?
BETTER ANSWERS
“WATER”, TOPMODEL, and the VSA Concept
WHAT IS “WATER”?
Developed by USGS in conjunction with KDOW
The Water Availability Tool for Environmental Resources
A “User-Friendly Decision Support System”
THE “WATER” SOFTWARE
Original ly Intended for Water Budget Modeling “A Water-Budge Modeling Approach for
Managing Water-Supply Resources” Based on daily inputs of precipitation,
evapotranspiration, withdrawals, and other data; used to estimate shortages
Phase 1 Initial Software Development Calibrated for Non-Karst Areas Includes basin delineation tool Uses TOPMODEL for simulation
Phase 2 (under review) Calibration updated to include Non-Karst
Areas Construction of Estimated Flow-Duration and
Load-Duration Curves Has seen widespread application in
USGS Water availability estimation in KY and AL Load-Duration Curves for TMDLs Flood Assessments in IN
THE “WATER” SOFTWARE
Phase I
Phase II
SCREENSHOTS OF WATER
Developed by Keith Beven and Mike Kirkby in 1979
Has been used in more than 30 countries worldwide
Simulates the Variable Source Area concept
Topographically-driven
Semi-distributed
Note: There are many “implementations” of TOPMODEL
TOPMODEL
Infiltration-Excess:Developed by Horton (1933)“Typical” method
THE VARIABLE SOURCE AREA CONCEPT
Figure from Wolock, 1993
Variable Source Area Initial concepts developed by Hursh
John D. Hewlitt coined the phrase Early career:
Mountainous watersheds in the southern Appalachians
Struggled with developing model that incorporated VSA concept
THE VARIABLE SOURCE AREA CONCEPT
Figure from Wolock, 1993
ILLUSTRATION OF THE VARIABLE SOURCE AREA CONCEPTAdapted from Bul let in 164, Loganathan, et al (1989)
3 types of flow are computed Direct Flow Return Flow Subsurface Flow
Soil parameters that are typically used Hydraulic Conductivity Available Water
Capacity Transmissivity ... and others.
SUBSURFACE FLOW
“Dunne” Overland Flow Total
Discharge
Topographic Wetness Index (TWI)
High values of TWI = High potential for saturation
Low values of TWI = Low potential for saturation
TOPOGRAPHICALLY “DRIVEN”
TWI HistogramTypically, 30 binsMean Fraction of total watershed area
Groups areas that are hydrologically similar
SEMI-DISTRIBUTED
Figures from USGS
Objectives and Methods
OUR STUDY
How does the WATER software need to be modified to fi t FEMA needs? Need: Computation of peak discharges
Annual peaks Recurrence interval events (1%-AEP, aka 100-year)
Need: Effi cient methodology Placement of pour points Delineation of watersheds
Need: “Better” answers Better than regression equations? As good as other rainfall-runoff models? How does LiDAR affect the answer?
Paper by Bhaskar indicated need for using more resolute elevation data
THE CHALLENGE
Two Tests1. LiDAR and Topographic Wetness Index (TWI)
WATER results (baseline)LiDAR re-sampled at same resolution as WATER DEM (30’ x 30’ DEM)
LiDAR at “native” resolution (5’ x 5’ DEM)2. Daily vs Hourly Timestep
Mimic WATER results using TOPMODEL Replace precipitation time series with hour-based storm event
OBJECTIVES
STUDY AREALevisa Fork Watershed
7 Gages – from 0.8 to 56 square miles in drainage area
Levisa Fork
TEST 1: LIDAR-BASED TWI
1. Check TWI-grid creation by comparison with WATER results
Same elevation source (30-foot DEM, NED-based) Substitute DEM file in the WATER database and run
2. Re-sample LiDAR DEM to same resolution as original DEM
5’ x 5’ DEM to 30’ x 30’ DEM Better estimate of elevation values (averaging 36
cells into one) Substitute DEM file in the WATER database and run
3. Use LiDAR DEM at native resolution 5’ x 5’ DEM Substitute DEM file in the WATER database and run
TEST 2: HOURLY PRECIPITATION
Create baseline run Run TOPMODEL (R Version) using same daily
precipitation and evapotranspiration values as WATER
Compare results and calibrate as necessarySubstitute daily precipitation values
In TOPMODEL R Select events
Design storm Historic storms
Check Q’s versus gage values
What we discovered
RESULTS & CONCLUSIONS
Graph of TWI-results (% error vs drainage area)
LIDAR-BASED TWI
HOURLY PRECIPITATION - BASELINE RUN
Baseline Runs (i.e. WATER with daily precipitation input)
BASELINE RUN
1. Clear indications that hourly data will lead to improvement
2. Computation of precipitation time series:a) Hourly precipitation time series have been created for
each gage using NOAA Atlas 14 data for precipitation depths, and NRCS Type II 24-hour rainfall distribution (for the 10- and 100-year events)
b) Processing is underway for NexRAD datai. Completed for Grapevine Creek (6 square miles)ii. Begun for Johns Creek (56 square miles)iii. Process the storm for each annual peak flood in the USGS
gage record that is within the period of record of the NexRAD data (1997=2010)
iv. NexRAD provides insight into the use of continuous simulation for estimation of peakflow
STATUS OF HOURLY TEST
1. LiDAR-based TWI grids did not have a signifi cant eff ect, however:
Other parameters should be re-calibrated when resolution changes The effect of the # of histogram bins was not tested At least one other study in the region has indicated better
resolution = better result
2. We anticipate improvement of peak fl ow estimation with the implementation of hourly data
3. Collaboration with USGS could lead to further development of WATER, and an effi cient, accurate means of estimating recurrence interval peak fl ows – for ungaged locations.
4. Any improvement in discharge estimation will have to be weighed against two factors:
a) % Errorb) Cost
CONCLUSIONS
REFERENCES/CREDITS
Thanks to Jeremy Newsom, Mike Griffi n, and Pete Cinotto, USGS KY Science Center!
Thanks to Tom Adams, NWS Ohio River Forecast Center Sources:
Williamson, et al. The Water Availability Tool for Environmental Resources (WATER). Phase I. US Geological Survey Scientific Investigations Report 2009-5248, 34 p.
Wolock, David M. Simulating the Variable-Source-Area Concept of Streamflow Generation with the Watershed Model TOPMODEL. US Geological Survey Water-Resources Investigations Report 93-4124, 39 p.
Loganathan, GV, et al. Variable Source Area Concept for Identifying Critical Runoff-Generating Areas in a Watershed: Bulletin 164. Virginia Polytechnical Institute and State College, May 1989, 125 p. Online at http://vwrrc.vt.edu/pdfs/bulletins/bulletin164.pdf
Questions?