Improving Hydrologic Analysis and Applications through the Use of - - PowerPoint PPT Presentation

Improving Hydrologic Analysis and Applications through the Use of Quality Controlled Radar Data and the Storm Precipitation Analysis System

Douglas M. Hultstrand, Metstat, Inc., Windsor, CO Beth Clarke, Weather Decision Technologies, Inc., Norman, OK Tye W. Parzybok, Metstat, Inc., Windsor, CO Edward M. Tomlinson, Ph.D., Applied Weather Associates, LLC, Monument, CO Bill D. Kappel, Applied Weather Associates, LLC, Monument, CO National Hydrologic Warning Council May 18-21, 2009 Vail, CO

Outline Background Storm Precipitation Analysis System (SPAS) –SPAS –SPAS-NEXRAD SPAS Output

Why Are Spatial Precipitation Estimates Important?

Rain Gauges Inadequate

• Spatial/Temporal

Crucial for hydrologic Modeling

• Calibration
• Validation

Traditional Techniques

not Representative

• Thiessen Polygon
• Inverse Distance Square
• Geostatisical Techniques

Total Rainfall Comparison

13.12” 6.17” 14.59” 7.98” 7.53” 6.51” 16.63” 10.07”

Inverse distance weighting (no radar) Default ZR relationship & no bias adjustment NWS radar-estimated rainfall SPAS-NEXRAD

Streamflow Comparison Peak Discharge (Nov. 6-8, 2006)

• Observed:

16,700 cfs

• SPAS-NEXRAD:

16,792 cfs (+ 92 cfs)

• IDW:

17,121 cfs (+ 421 cfs)

• Default ZR:

18,588 cfs (+ 1,888 cfs)

2000 4000 6000 8000 10000 12000 14000 16000 18000 20000 22000 24000 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100105110 Time (30-min Intervals) Streamflow (cfs) 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 Incremental Precipitation (in) Precipitation Observed SPAS-NEXRAD Default ZR IDW

• A comprehensive, state-of-the-science gridded

precipitation analysis software program

• Developed and operated by meteorologists and

hydometeorologists since 2002

• Historically a post-storm analysis program, but is

evolving into a real-time tool

• Skilled in analyzing extreme storm events
• Generates a plethora of output used for hydrologic

applications

• Has the unique capability to compute storm

centered depth-area-duration (DAD) tables

Storm Precipitation Analysis System

SPAS Modes SPAS operates in two modes

• SPAS (pre-NEXRAD storms)
• Utilizes a “basemap” for interpolating hourly storm
• precipitation. Basemap options include:
• Precipitation Frequency grids (e.g. 100-year 24-hour) - NOAA Atlas

14, TP-40, NOAA Atlas 2, etc.)

• Elevation - Digital Elevation Model (DEM)
• Mean (1971-2000) monthly precipitation - Parameter-elevation

Regressions on Independent Slopes Model (PRISM)

• PRISM Mean (1971-2000) annual precipitation
• PRISM Total monthly precipitation (e.g. July 1935)
• No basemap
• SPAS-NEXRAD
• Requires SPAS general to be run first
• Uses calibrated radar data for interpolating hourly precipitation
• SPAS-NEXRAD Real-Time
• In development stage

Hourly gage data Daily gage data

• Supp. gage data

Reformat & QA/QC Pooled hourly gage data Convert to hourly Convert to hourly Basemap Compute % of Basemap (“isopercental”) at gages Spatially interpolate gage Isopercentals to a grid Isopercental * Basemap = hourly precip grid

Repeat each hour

• Prelim. total storm grid

Final total storm grid QA/QC QA/QC QA/QC Raw gage precip. data QA/QC DAD results

Repeat (if necessary)

Storm center(s) mass curve (timing information)

SPAS Flowchart

Depth-Area-Duration Analysis Other (GIS files, etc.)

SPAS-NEXRAD

QA/QC Radar?

Yes No Hourly precip. Grids

SPAS utilizes a variety of precipitation data to achieve the highest

level of spatial and temporal detail possible.

• Hourly data
• In-house National Climatic Data Center (NCDC) database
• Automated Local Evaluation in Real Time (ALERT) networks,

Remote Automated Weather Stations (RAWS) stations, NWS’s Automated Surface Observing Systems (ASOS), municipal networks, flood control districts.

• Daily data
• In-house National Climatic Data Center (NCDC) database
• Municipal networks, etc
• Supplemental data
• Storm total’s from “bucket survey’s”, public reports NWS Storm

Data, etc.

SPAS Precipitation Input

Hourly gage data Daily gage data

• Supp. gage data

Reformat & QA/QC Pooled hourly gage data Convert to hourly Convert to hourly Basemap Compute % of Basemap (“isopercental”) at gages Spatially interpolate gage Isopercentals to a grid Isopercental * Basemap = hourly precip grid

Repeat each hour

• Prelim. total storm grid

Final total storm grid QA/QC QA/QC QA/QC Raw gage precip. data QA/QC DAD results

Repeat (if necessary)

Storm center(s) mass curve (timing information)

SPAS Flowchart

Depth-Area-Duration Analysis Other (GIS files, etc.)

SPAS-NEXRAD

QA/QC Radar?

Yes No Hourly precip. Grids

To achieve an hourly time step at ALL stations, its necessary to convert

daily & supplemental stations into estimated hourly stations.

In the past, timing of daily measured data was accomplished by

associating each daily station with a single nearby hourly station.

SPAS, however, uses several hourly stations to time each of the daily

stations, thereby allowing the hourly precipitation distribution to be unique at each daily station.

SPAS Methodology

Daily to Hourly Precipitation

Daily data



Estimated hourly data

2 4 6 8 1 2 3

D ay In.



0.0 0.2 0.4 0.6 0.8 1 .0 1 .2 1 1 3 25 37 49 61 73 85 H o ur

In.

This provides more representative spatial and temporal detail.

Hourly gage data Daily gage data

• Supp. gage data

Reformat & QA/QC Pooled hourly gage data Convert to hourly Convert to hourly Basemap Compute % of Basemap (“isopercental”) at gages Spatially interpolate gage Isopercentals to a grid Isopercental * Basemap = hourly precip grid

Repeat each hour

• Prelim. total storm grid

Final total storm grid QA/QC QA/QC QA/QC Raw gage precip. data QA/QC DAD results

Repeat (if necessary)

Storm center(s) mass curve (timing information)

SPAS Flowchart

Depth-Area-Duration Analysis Other (GIS files, etc.)

SPAS-NEXRAD

QA/QC Radar?

Yes No Hourly precip. Grids

The base map helps interpolate values at ungauged locations in

complex terrain.

Methodology

Base Map Concept

Without base map With base map (Mean Monthly Precipitation)

Hourly gage data Daily gage data

• Supp. gage data

Reformat & QA/QC Pooled hourly gage data Convert to hourly Convert to hourly Basemap Compute % of Basemap (“isopercental”) at gages Spatially interpolate gage Isopercentals to a grid Isopercental * Basemap = hourly precip grid

Repeat each hour

• Prelim. total storm grid

Final total storm grid QA/QC QA/QC QA/QC Raw gage precip. data QA/QC DAD results

Repeat (if necessary)

Storm center(s) mass curve (timing information)

SPAS Flowchart

Depth-Area-Duration Analysis Other (GIS files, etc.)

SPAS-NEXRAD

QA/QC Radar?

Yes No Hourly precip. Grids

The hourly

precipitation grids serve as the basis for much of the output statistics Hourly precipitation ending at 1600 GMT

• Nov. 6, 2006

Methodology

Hourly Precipitation Grids

Hourly gage data Daily gage data

• Supp. gage data

Reformat & QA/QC Pooled hourly gage data Convert to hourly Convert to hourly Basemap Compute % of Basemap (“isopercental”) at gages Spatially interpolate gage Isopercentals to a grid Isopercental * Basemap = hourly precip grid

Repeat each hour

• Prelim. total storm grid

Final total storm grid QA/QC QA/QC QA/QC Raw gage precip. data QA/QC DAD results

Repeat (if necessary)

Storm center(s) mass curve (timing information)

SPAS Flowchart

Depth-Area-Duration Analysis Other (GIS files, etc.)

SPAS-NEXRAD

QA/QC Radar?

Yes No Hourly precip. Grids

Pooled hourly gage data (R)

SPAS-NEXRAD Flowchart

Hourly NEXRAD Reflectivity (Z) QA/QC Relate, optimize & QC ZR relationship Compute initial precip grid using ZR algorithm Compute gage residual (Robs – Rcalc) Spatially interpolate gage isoresidual to grid Isoresidual * basemap = Bias correction grid Compute residual as %

• f basemap (“isoresidual”)

Bias correction grid + initial precip grid = final precip grid Repeat each hour QA/QC Basemap

NEXRAD data Provided by Weather Decision Technologies (WDT) WDT uses advanced algorithms for mosaicing Z from

multiple radar sites and overcoming common radar errors (blockage, clutter, etc.)

SPAS-NEXRAD imposes further QC on the WDT grids

SPAS-NEXRAD

NEXRAD Reflectivity (Z)

RAW Z QC’ed Z

Pooled hourly gage data (R)

SPAS-NEXRAD Flowchart

Hourly NEXRAD Reflectivity (Z) QA/QC Relate, optimize & QC ZR relationship Compute initial precip grid using ZR algorithm Compute gage residual (Robs – Rcalc) Spatially interpolate gage isoresidual to grid Isoresidual * basemap = Bias correction grid Compute residual as %

• f basemap (“isoresidual”)

Bias correction grid + initial precip grid = final precip grid Repeat each hour QA/QC Basemap

Reflectivity-rainfall (ZR) relationships are computed using a

complex set of thresholds, rules and algorithms to compute rainfall rates from radar reflectivity

Outliers are identified and QCed. The non-linear ZR relationship is optimized based on observed and

default ZR precipitation.

Instead of adopting a standard (e.g. 300^1.4) ZR relationship, SPAS

computes, optimizes, QCs and applies a ZR relationship each hour

** use default ZR relationship when insufficient data available

SPAS-NEXRAD

ZR Relationship

20 40 60 80 100 10 20 30 40 50 60 Reflectivity (dbz) Precipitation (mm)

SPAS-NEXRAD 300R^1 .4 ObservedPrecipitation 20.4/0.80 5.4/0.21

Pooled hourly gage data (R)

SPAS-NEXRAD Flowchart

Hourly NEXRAD Reflectivity (Z) QA/QC Relate, optimize & QC ZR relationship Compute initial precip grid using ZR algorithm Compute gage residual (Robs – Rcalc) Spatially interpolate gage isoresidual to grid Isoresidual * basemap = Bias correction grid Compute residual as %

• f basemap (“isoresidual”)

Bias correction grid + initial precip grid = final precip grid Repeat each hour QA/QC Basemap

SPAS-NEXRAD

Apply ZR to QC’ed Z grid (initial grid) Compute and interpolate “isoresiduals” at gages Add initial to residuals to create final grid

Utilizing a basemap to “normalize” the ZR residuals is a powerful

technique because …

It infuses the influence of orographics into the spatial

interpolation/patterns.

During radar outages, this technique mimics the SPAS

basemap approach, which is the next best spatial interpolation method.

Allows areas with poor or no radar data (e.g. in complex

terrain) to be interpolated using a basemap approach.

SPAS-NEXRAD

Basemap

SPAS-NEXRAD basemap SPAS basemap, no radar SPAS-NEXRAD no basemap Radar reflectivity

The value of radar The value of a basemap

SPAS-NEXRAD blends the best of SPAS and radar data

Hourly gage data Daily gage data

• Supp. gage data

Reformat & QA/QC Pooled hourly gage data Convert to hourly Convert to hourly Basemap Compute % of Basemap (“isopercental”) at gages Spatially interpolate gage Isopercentals to a grid Isopercental * Basemap = hourly precip grid

Repeat each hour

• Prelim. total storm grid

Final total storm grid QA/QC QA/QC QA/QC Raw gage precip. data QA/QC DAD results

Repeat (if necessary)

Storm center(s) mass curve (timing information)

SPAS Flowchart

Depth-Area-Duration Analysis Other (GIS files, etc.)

SPAS-NEXRAD

QA/QC Radar?

Yes No Hourly precip. Grids

Unique capability to calculate

storm-centered D-A-D

Three-dimensional characterization Quantify rainfall event Warning criteria

SPAS

Depth-Area-Duration Analysis

Hourly gage data Daily gage data

• Supp. gage data

Reformat & QA/QC Pooled hourly gage data Convert to hourly Convert to hourly Basemap Compute % of Basemap (“isopercental”) at gages Spatially interpolate gage Isopercentals to a grid Isopercental * Basemap = hourly precip grid

Repeat each hour

• Prelim. total storm grid

Final total storm grid QA/QC QA/QC QA/QC Raw gage precip. data QA/QC DAD results

Repeat (if necessary)

Storm center(s) mass curve (timing information)

SPAS Flowchart

Depth-Area-Duration Analysis Other (GIS files, etc.)

SPAS-NEXRAD

QA/QC Radar?

Yes No Hourly precip. Grids

SPAS Output

Total Storm Map

SPAS-NEXRAD: Magma, AZ August 29, 2000 SPAS: Vilas County, WI, August 21-24, 1978

Mass curves for any location

SPAS Output

Mass Curves

SPAS Output

Watershed Total Storm and Mass Curves

Complete station list and summary

– Including station density stats

Hourly precipitation grids (importable into a GIS)

• At a user-defined resolution

Hourly precipitation isopluvials (importable into a GIS)

• At a user-defined interval

Other customized output (e.g. storm animations)

SPAS Output

Other

The use of radar data increases the spatial and temporal

detail of rainfall information vs. use of rain gauges only.

The blend of radar- and basemap-based approaches allows

for consistent precipitation patterns across varied terrain, during times of radar outages.

SPAS-NEXRAD demonstrates the advantages of computing

and optimizing ZR relationships each hour rather than adopting a default.

Quality rainfall input to hydrologic models helps to achieve:

• Improved flood prediction
• Improved design of hydrologic structures
• Improved model calibration and verification

Real-time SPAS-NEXRAD capability being developed

Summary

Take Home Message

Questions

Applied Weather Associates:

Email: awaadmin@comcast.net Website: http://www.appliedweatherassociates.com

Metstat:

Email: info@metstat.com Website: http://www.metstat.com

Portland, OR

Precipitation (in/hr)

Portland, OR

Precipitation (in/hr)

• The DAD

functionality of SPAS was subjected to extensive testing to make sure…

• 1. It was correctly

computing the analytical truth AND…

Methodology

DAD Verification

Depth-Area Curves for 10-hr Storm "Pyramidville" - 39.5N 104.5W

0.0100 0.1000 1.0000 10.0000 100.0000 0.2 0.4 0.6 0.8 1 1.2 Maximum Average Precipitation Depth (inches) Area (sq. mi.) DAD Software Analytical truth Interpolated standard area sizes

Unique capability to calculate

storm-centered D-A-D

three-dimensional characterization

Methodology

Depth-Area-Duration Analysis

• The DAD results

compared favorably to previously analyzed storms, including:

• 1. Westfield, MA, storm
• f August 17-20,

1955

• 2. Ritter, IA storm of

June 7, 1953

Methodology

DAD Verification (cont.)

SPAS Sq-Miles 6-hour 12-hour 24-hour 36-hour 48-hour 60-hour Total 10 7.96 11.48 16.40 19.10 19.11 19.47 19.70 100 7.22 10.72 15.20 17.77 17.76 18.23 18.47 200 6.99 10.27 14.28 16.91 16.84 17.39 17.54 1000 5.97 9.06 12.55 14.97 15.08 15.40 15.95 5000 4.14 6.45 9.25 11.70 12.02 12.35 13.05 10000 3.23 5.46 7.63 9.60 9.91 10.26 10.86 20000 2.24 4.03 5.91 7.66 7.97 8.22 8.77 Weather Bureau Sq-Miles 6-hour 12-hour 24-hour 36-hour 48-hour 60-hour Total 10 7.80 11.10 16.40 18.90 19.40 19.40 19.40 100 7.60 10.50 14.60 18.10 18.80 19.00 19.00 200 7.40 10.20 14.20 17.60 18.20 18.40 18.40 1000 6.20 9.20 12.40 15.90 16.20 16.40 16.40 5000 4.00 6.30 9.50 12.10 12.60 13.00 13.00 10000 3.10 5.00 8.00 10.00 10.60 10.80 10.80 20000 2.10 3.60 6.30 7.90 8.30 8.50 8.50 Percent Difference Sq-Miles 6-hour 12-hour 24-hour 36-hour 48-hour 60-hour Total 10 2.1% 3.4% 0.0% 1.1%

• 1.5%

0.4% 1.5% 100

• 5.0%

2.1% 4.1%

• 1.8%
• 5.5%
• 4.1%
• 2.8%

200

• 5.5%

0.7% 0.6%

• 3.9%
• 7.5%
• 5.5%
• 4.7%

1000

• 3.7%
• 1.5%

1.2%

• 5.8%
• 6.9%
• 6.1%
• 2.7%

5000 3.5% 2.4%

• 2.6%
• 3.3%
• 4.6%
• 5.0%

0.4% 10000 4.2% 9.2%

• 4.6%
• 4.0%
• 6.5%
• 5.0%

0.6% 20000 6.7% 11.9%

• 6.2%
• 3.0%
• 4.0%
• 3.3%

3.2%

Generally within +/- 5% !!