OVERVIEW
I report here on the
development and implementation of meteorological analyses and forecasts
for the Local Analysis and Prediction System within the Forecast Applications
Branch (FAB).
Background information about LAPS is available on the World Wide Web at
http://laps.noaa.gov/. Other GSD projects include Science on a Sphere.
This report is organized with a section on LAPS improvements followed by
a project highlights section. One key to our long term success is the ability
to leverage similar work done for multiple projects into LAPS improvements,
especially since only a small portion of our funding is
specifically earmarked for such software development.
Conversely, specific LAPS improvements often benefit multiple projects.
The number of collaborative projects that
I work on has become relatively large and these may not be fully reflected
in the original CIRA statement of work due to the fluid nature of new contracts
being agreed to at various times.
To me this all lends credence to the notion of
amalgamating some of the project work into common themes of software development
for improved organization in this report.
LAPS/STMAS ANALYSIS IMPROVEMENTS
I have been working to increase the versatility
of LAPS analyses (both the traditional version and the newer STMAS version).
They can handle more types of data in a greater variety of
situations to provide more accurate analyses. The analyses are now better
suited for forecaster interpretation and model initialization. I'm also working
on a few aspects of the model initialization and post-processing.
First Guess Interpolation
Software now runs faster on large domains based on improvements to the bilinear
and binomial interpolation routines.
The program now just does vertical interpolation in a box that is a subset
of the large scale model to save more time.
Support for the HRRR model first guess was
improved.
The first guess interpolation now reads in GRIB data such as (HRRR) more
efficiently, taking 25% less time. I've talked with Paula McCaslin about
other changes that can take another 25% off, meaning the end result
will be about twice as fast.
Observational Data Sets
Improvements were made in LAPS to analyze observations from new types of instruments
and new data formats - thus expanding the envelope of meteorological data environments
that we can operate in with our ever growing set of users. This is outlined in detail
below for surface and upper air observations.
The surface observation blacklist was updated with a station that
regularly reports erroneous wind directions.
Sub-hourly boundary layer RASS data can now be ingested, though the
software no longer waits around for such data, helping to speed up
our high-frequency runs.
Surface Analysis
New QC routines produce statistics using the 24-hour history of
surface obs. These are now being called by the traditional LAPS
surface analysis to report temperature and dewpoint bias, as well
as "stuck sensor" checks for wind direction and speed.
The stuck sensor wind checks now are used to withhold
erroneous observations from the traditional LAPS surface analysis.
Radar Processing
I have worked towards more efficiency and other functional improvements
for radar remapping and mosaicing.
This is described in detail as follows.
We now look for volumetric radar netCDF files in addition to the tilt ones.
Clouds / Precipitation Analyses
THe radar reflectivity access routine should work now using sparse
(non-filled) arrays for STMAS. The "Kessler" option for precipitating rain and
snow now has correct units. Preparations were made to use first guess humidity
data within the satellite cloud top processing routine.
Wind / Temperature Analyses
Working with Jacques Middlecoff of the HPCC group, MPI changes to parallize key
portions of the wind analysis are now in place. This is now running in
real-time on our JET linux cluster.
Related script refinements were made to have more flexibility and efficiency
when we run the parallelized analysis on JET, including during the transition
from the WJET to HJET clusters.
Radiometer temperature profiles now get used in the lowest 1km for
the traditional LAPS temperature analysis.
Stability Indices
LAPS Model Initialization/Post Processing
The LFMPOST test script now works better and
can be set up automatically via the LAPS configuration.
LAPS Fire weather now gets calculated in LFMPOST by default.
General Software Improvements / Portability
Land Fraction is now unsmoothed (and is prevented from going slightly
negative), so it will show more detail on high resolution (1km)
analyses.
Software error checking, documentation and logging were improved.
Variable initialization was improved for machines that do not automatically
initialize variables to zero.
The purger can now be run in the background to speed up the run
cycle. Core files are now also purged.
Support was improved for running
LAPS on different JET clusters (such as WJET and HJET).
Portability improvements to LVD satellite ingest were made for
the Intel compiler.
Software builds are now being set up (and improved) for Intel,
gfortran, (and PGI on the Mac).
LAPS Implementation
Other interaction within and outside of GSD
I continue to give ~1 weather briefing per month.
WWW LAPS Interface
Support was added for "montage" comparisons with the "on-the-fly" page.
Additional fields were added and/or refined, including cloud and microphysics
fields.
Scaling was improved for certain variables.
Forecast LAPS fire weather is now supported in our plotting software.
NCAR graphics "ctrans" usage now works better on various machines.
Pre-generated forecast graphics now more reliably read the domain aspect
ratio from the grid specification file.
The NCAR graphics and "on-the-fly" page now work more efficiently
for forecast plots, while raster fields like 'divergence' now
work for animations. Colorbars will now display for zoomed plots.
Wind gusts were added to obs plots.
PROJECT HIGHLIGHTS
National Science Foundation (NSF)
Department of Water Resources (DWR)
Windsor Tornado Case Study
AFTAC/ATMET
Science on a Sphere (SOS)
Space-Time Mesoscale Analysis System (STMAS)
Added Sigma-Pressure grid support for library subroutines and modules.
Fire Weather
The LAPS/Kelsch fireweather index is now included in our model post-processor so
we can generate this for model forecasts.
NetCDF output files are now consistently set up for two of our main fireweather
indices.
GeoTargeted Alert System (GTAS)