Rainfall Runoff Prediction

1
Rainfall Runoff Prediction

• Designed and presented by George Limpert
• in association with CARES and Chris Barnett
• Mentor Dr. Neil Fox

2
Project Goals

• The prediction of rainfall runoff has useful
  applications to agriculture
• Prediction depends on terrain, soil type, and
  rainfall amounts
• Terrain and soil type information is already
  available via existing GIS systems
• Provide rainfall amounts in a form that can be
  read by existing GIS software
• Create a product that can be used to assist in
  the prediction of runoff

3
Precipitation Estimates

• Precipitation estimates are provided by radar
  estimates
• Over 100 WSR-88D doppler radar sites throughout
  the country
• Precipitation estimation algorithm takes into
  account terrain and reflectivity to produce an
  accurate estimate
• Several precipitation estimates are distributed
  as level III data

4
Level III Data

• Four precipitation estimate products
• Digital precipitation array
• One hour precipitation estimate
• Three hour precipitation estimate
• Storm total precipitation estimate
• Tradeoff between various products

5
One Hour Precipitation Estimate

• Radial data
• Resolution of 1 degree x 2 kilometers
• 16 levels
• Good resolution but poor precision
• Provides a range for possible precipitation total

6
Digital Precipitation Array

• Provides precipitation totals for the past hour
• 256 levels
• Approximately 4 km resolution
• Gridded data
• Good precision but mediocre resolution
• Provides an actual estimate for total
  precipitation

7
Level III Data

• Freely available from NWS ftp server
• Available only in a poorly documented format
• Few third party tools for decoding level III data
• NWS ftp server has limited resources and provides
  slow downloads

8
Requirements

• Cover the entire state of Missouri if possible
• Nine radar sites needed
• Saint Louis
• Pleasant Hill
• Springfield
• Paducah
• Memphis
• Tulsa
• Omaha
• Quad Cities
• Des Moines

9
Requirements

• Produce an accurate decoding of the data
• Convert the data into a format which is readable
  by GIS systems
• Provide the data at reasonable intervals
• System should run automatically if possible
• Some faults should be tolerated, particularly
  radar downtime and network outages

10
System Structure

• Three modules organized in a linear fashion
• First module downloads appropriate data for radar
  sites from NWS ftp servers
• Second module decodes the data and converts it to
  a format which can be read by GIS systems
• Third module uploads data to a computer with GIS
  software installed

11
System Structure

• Second module is the most complex and can be
  further broken down
• Organized into input modules, a core, and output
  modules
• Input modules responsible for reading in level
  III data
• Core selects an appropriate input and output
  module
• Output module produces readable data in a
  different format

12
Development Tools

• Cygwin provides a UNIX-like environment on a
  computer running Windows
• Compiler of choice is gcc
• Output graphics produced by libpng
• Output to shapefiles produced by shapelib

13
Demonstration

• March 22, 2005
• One hour precipitation total
• Tallahassee, FL

14
Demonstration

• March 22, 2005
• One hour precipitation total
• Tallahassee, FL

15
Demonstration

• March 22, 2005
• One hour precipitation total
• Tallahassee, FL

16
Demonstration

• March 22, 2005
• One hour precipitation total
• Tallahassee, FL

17
Demonstration

• March 22, 2005
• One hour precipitation total
• Tallahassee, FL

18
Demonstration

• March 22, 2005
• One hour precipitation total
• Tallahassee, FL

19
Demonstration

• March 22, 2005
• One hour precipitation total
• Tallahassee, FL

20
Demonstration

• March 22, 2005
• One hour precipitation total
• Tallahassee, FL

21
Demonstration

• March 22, 2005
• One hour precipitation total
• Tallahassee, FL

22
Demonstration

• March 22, 2005
• One hour precipitation total
• Tallahassee, FL

23
Demonstration

• March 22, 2005
• One hour precipitation total
• Tallahassee, FL

24
Demonstration

• March 22, 2005
• One hour precipitation total
• Tallahassee, FL

25
Demonstration

• March 22, 2005
• One hour precipitation total
• Tallahassee, FL

26
Demonstration

• March 8, 2005
• Near Montgomery, AL
• One hour precipitation total

27
Verification

• March 8, 2005
• Near Montgomery, AL
• One hour precipitation total
• Image generated by the NWS

28
Demonstration

• March 9, 2005
• Melbourne, FL
• Base reflectivity

29
Verification

• March 9, 2005
• Melbourne, FL
• Base reflectivity
• Image generated by the NWS

30
Licensing

• Support free and open exchange of data and ideas
• Private weather industry wants to restrict the
  ability of NOAA to disseminate data to the public
• Open source software, by definition, cannot place
  licensing restrictions to prevent commercial use
• Therefore, this product will not be released as
  open source software

31
Development Status

• Core functionality of decoding level III data is
  completed and has been tested
• Converting data to GIS readable format has not
  been tested
• Automatic downloading, conversion, and uploading
  of data has not been tested
• Data from this project has not yet been
  integrated into the CARES website

32
Future Development

• Radar coverage may be improved if NWS places a
  WSR-88D in Mid-Missouri or north central Missouri
• Make use of actual observed rainfall totals to
  calibrate precipitation estimates
• Generate precipitation estimates from raw level
  II data
• Make use of less advanced radar to improve
  estimates in areas with poor radar coverage