Investigators

Witold F. Krajewski¹, James A. Smith², Mary-Lynn Baeck², and Anton Kruger¹, Brian R. Nelson¹ and Remko Uijlehoft²

Affiliations

¹IIHR, The University of Iowa. 

²Department of Civil Engineering and Operations Research, Princeton University.

Sponsoring Agency

National Oceanic and Atmospheric Administration, Office of Global Programs

Animation of 15-minute NEXRAD reflectivities.

Background

The principal objective of the development of archival precipitation products for the GEWEX Continental-Scale International Project (GCIP) region (the Mississippi River basin). We are developing this precipitation data set for the period 1995-2000 for hourly accumulations with a spatial resolution of 4×4 km. We have proposed to use the national, 15-minute reflectivity composite (see image below) and hourly rain gauge data as the principal inputs for the legacy precipitation data set. Other inputs will come from the National Center for Environmental Prediction (NCEP) hourly rainfall composite, the WSR-88D level II reflectivity data from selected locations, and data from the GOES-8 imagery.

Data Format Development

A major hurdle in the development of these precipitation data is the formulation of an infrastructure for the storage, management, analysis, and distribution of such large data sets.  We have developed a fast, efficient, and highly compressed algorithm for storing the precipitation data.  We have also developed a data browser based in Geographic Information Systems (GIS) software that will read precipitation maps and allow for browsing the data on a sub-basin level.

Our algorithm involves storing the data in a run-length encoded (RLE) manner.  The savings in storage space compared to the original Hierarchical Data Format (HDF) is almost 3 times.  Also, the time it takes to access the data for display or analysis is almost ˝ the time as for the HDF data.  You can see a distribution of the size of the data sets in each format.  The total size of the data set in HDF format is 24 GB.  The total size of the data set in our format is less than 8 GB

HDF versus RLE file size.

Data Browser Development

We have developed a data browser based in a commercial GIS software for visualization of the precipitation data.  We created an extension to the ArcView® 3.2 GIS software that allows for easy manipulation of high-resolution spatial data, and our extension to the software allows for more facilitation of the temporal data.  The creation of this extension involved the development of a Dynamically Linked Library (DLL) to connect our RLE formatted data to the GIS software.  The flow chart shows how the DLL uncompresses the data and converts it to a Band Interleaved by Pixel (BIP) format that is ready for import to ArcView.

The browser allows for viewing of the data from large-scale hydrologic basins to smaller watersheds.  We have incorporated tools for the user to select specific rainfall maps based on the date, the file size, and the percentage of echoes in the entire map.  You can see one 15-minute rainfall-reflectivity map in the figure.  We will also build in capabilities for a user to see animation of certain rainfall patterns.  Another figure shows one of these animations.

Screenshot of ArcView Data Browser

In addition to these methodological issues, there are more pragmatic questions on the form of the final data set. To help us develop a useful product, we have developed a short, easy, check-box type questionnaire to gather input from the hydrology, hydrometeorology, and atmospheric science communities of potential users. If you are a potential user of this data set, please click on the link and submit the survey. It takes less than 5 minutes to complete. 

Publications

Conference Presentations

Conference Posters