J36.1 The Weather Information Statistical Post-Processing System (WISPS): Project Update

Wednesday, 10 January 2018: 8:30 AM
Room 19AB (ACC) (Austin, Texas)
Jason J. Levit, CIRA, Fort Collins, CO; and M. Peroutka, R. Conroy, E. Engle, and I. Stajner

The Meteorological Development Laboratory (MDL) of the National Weather Service (NWS) is developing the Weather Information Statistical Post-Processing System (WISPS), a community-based scientific software system for post-processing numerical weather prediction data. Started in earnest during early 2016, WISPS has evolved into functional software components, and is moving towards an initial operating capability (IOC) of reproducing the output from MDL’s Model Output Statistics software (MOS-2000). As WISPS continues to grow as a community-based system, we encourage the global weather enterprise to join our development effort. We are actively pursuing linkages to post-processing data from NOAA’s evolving Next Generation Global Prediction System (NGGPS), and members of our development staff also serve on the NGGPS Post-processing Strategic Implementation Plan Working Group.

The main purpose for developing WISPS is to create a system that many organizations can use for statistical post-processing (StatPP) efforts. To facilitate this idea, WISPS is employing the use of modern software design, data storage techniques, and data standards to ensure community adoption. The WISPS primarily uses an object-oriented design language, Python, and NetCDF to store data, is compliant with Climate and Forecast conventions (NetCDF-CF), and contains metadata related to StatPP processes, data, and concepts that are described at a code registry website (codes.nws.noaa.gov). We have proposed additional metadata standards that can improve the interoperability of post-processed NWP output.

As a modular software system, the WISPS intends to be as much “plug and play” as possible, allowing for StatPP techniques to be rapidly coded into the system. The NetCDF data storage solution, using linked data techniques, is the backbone of WISPS and allows users to search data with metadata key tags in order to deliver particular data to a StatPP algorithm. At first, and to meet IOC, the WISPS will use the scientific and mathematical workflow behind MOS-2000 and create MOS much as it is produced today, using quality controlled weather observations, NWP output, and least squares regression for weather stations or model grids. As WISPS grows, we will support the various multi-model ensemble techniques that currently support NOAA’s National Blend of Models (NBM). We encourage the global weather enterprise to add their own algorithms to the suite of StatPP algorithms, and generate different techniques for creating post-processed output.

The WISPS package is currently managed through NOAA’s Virtual Laboratory (VLab) and uses Git for version control, as well as Jenkins and Gerrit for testing and software review. In addition, the VLab site contains documentation on the project and provides a platform for discussion forums.

An overview of the current status of the project, emerging linkages to the NGGPS, and future direction will be presented.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner