NOAA’s National Weather Service Tools To Access the NDFD Arthur Taylor RS Information Systems / Meteorological Development Laboratory National Digital Forecast Database Technical Workshop August 13, 2003

NDFD: How do I use it? Since NDFD uses the WMO’s GRIB2 format, which is relatively new, NWS is providing a decoder library. The current versions of the library are located at: http://www.nws.noaa.gov/mdl/iwt/grib2/decoder.htm http://www.nws.noaa.gov/mdl/iwt/grib2/encoder.htm In addition, NWS is providing a driver “degrib” (aka “NDFD_GRIB2Decoder”) for the library. Degrib’s primary purposes are: Allow users to convert GRIB2 to other formats for use with various Geographic Information Systems (GIS), without need for a programmer. Provide an example of how to call the decoder libraries Enable some form of access to the data without requiring other packages Generate Images Probe a give lat / lon location for all the relevant data

Degrib: Installation Degrib can be downloaded from: http://www.nws.noaa.gov/mdl/NDFD_GRIB2Decoder/register.htm On the download page, MS-Windows users should get “ndfd-demo.exe”, which contains: An installation wizard Compiled code Source code for the libraries, and the “degrib” and “tkdegrib” programs (superImageGen and htmlmaker source code are not provided) Documentation Unix users should get “degrib-unix.tar.gz”, which contains: Source code for the libraries and the “degrib” and “tkdegrib” programs Documentation. To compile the code you need: FORTRAN (g77 should work) : for the decoder library C (gcc should work) : for the “drivers” and projection library Optional : Tcl/Tk (free from http://scriptics.com) : for the Graphical User Interface

Degrib: Data download Step 1: Download some data. The GUI version of degrib, “tkdegrib”, provides a convenient interface to get NDFD either via ftp or http. Highlight the desired sector Click on either “Download by ftp” or “Download by http” Alternatively, you can use various UNIX tools to get the data directly from: http://weather.noaa.gov/pub/SL.us008001/ST.expr/DF.gr2/DC.ndfd/ ftp://tgftp.nws.noaa.gov/SL.us008001/ST.expr/DF.gr2/DC.ndfd/

Degrib: Convert Currently degrib can convert to the following formats: ESRI Point .shp files ESRI Polygons .shp files GrADS lat / lon grids GrADS Projected grids .flt files for use with ESRI Spatial Analyst .flt files for use with other programs that can read NxM 4 byte floats. In addition, degrib creates a .txt file which contains the meta data that was in the GRIB2 message.

Degrib: Convert to .shp file To convert to .shp format: Double click on the GRIB2 file (top pane) Select message to convert (middle pane) Choose an “output file name” Click on “Generate .shp files” “Polygon .shp” creates better images in ESRI and is easier to manipulate, but it is larger than the “Point .shp” “Include Missing Values” allows you to study the entire grid, or limit the result to where data exists

CONUS MaxT in ArcView

Polygon vs Point, DC area

Degrib: Convert to .flt file To convert to .flt file (for use with GrADS or ESRI Spatial Analyst): Double click on the GRIB2 file (top pane) Select message to convert (middle pane) Choose an “output file name” Click on “Generate .flt files” “GrADS .ctl file”: creates a control file which GrADS can use to access the .flt file “Interpolate”: bi-linear interpolates to a lat / lon coverage grid. “M.S.B. first”: creates the .flt files in “Big Endian” format.

CONUS MaxT in GrADS

NDFD, GRIB2 & Weather To encode Weather (Wx), NDFD disseminates “ugly strings”. Advantage: A flexible format that can encode a description of weather such as “Chance of thunderstorms and chance of heavy rain showers” Difficulty: An “ugly string” is not a number, so it is challenging to store in GRIB2. Solution: Use GRIB2 section 2 to provide an ASCII look up table, and store the numeric values in the regular GRIB2 Note: each weather grid, for each forecast projection, has a different ASCII look up table, so without section 2 it has no meaning. Some Questions: What does “Chc:T:<NoInten>:<NoVis>:^Chc:RW:+:<NoVis>:” mean, and how, for example, is an Emergency Manager supposed to know that? What can be done for the .flt file?

Weather & .shp files Degrib looks up the value, and parses it when it saves Wx to a .shp file Wx-Index: Ugly string index used in the GRIB2 message. Visibility: The lowest visibility in the 5 “ugly string words” (in statute miles). NDFD wxCode: An encoding of the ugly string using the first wx-type, intensity, and coverage, and the second wx-type. Weather 1: English translation of the “ugly word” Wx-Inten 1: A number code for the wx-type and wx-intensity Cover 1: A number code for the coverage Hazard 1: A number code for hazards. There can be 5 hazards per word, there are 9 types of hazards. So each hazard is given a digit, the digits are sorted. Result: “00045” = 45 for “Damaging Wind (4) and “Small Hail (5)”

Weather Codes: “Coverage” Table “Wx-Inten” Table “Hazards” Table No Coverage 1 Isolated 2 Scattered 3 Numerous 4 Widespread 5 Occasional 6 Slight Chance of 7 Chance of 8 Likely 9 Definite 10 Patchy 11 Areas of “Wx-Inten” Table No Weather 1-5 Smoke (no Intensity, very light, light, moderate, heavy) 6-10 Blowing Dust (no Intensity, very light, light, moderate, heavy) 11-15 Blowing Snow (no Intensity, very light, light, moderate, heavy) 16-20 Haze (no Intensity, very light, light, moderate, heavy) 21-25 Fog (no Intensity, very light, light, moderate, heavy) 26-30 Drizzle (no Intensity, very light, light, moderate, heavy) 31-35 Rain (no Intensity, very light, light, moderate, heavy) 36-40 Rain Showers (no Intensity, very light, light, moderate, heavy) 41-45 Hail (no Intensity, very light, light, moderate, heavy) 46-50 Frost (no Intensity, very light, light, moderate, heavy) 51-55 Freezing Drizzle (no Intensity, very light, light, moderate, heavy) 56-60 Freezing Rain (no Intensity, very light, light, moderate, heavy) 61-65 Ice Pellets (no Intensity, very light, light, moderate, heavy) 66-70 Snow (no Intensity, very light, light, moderate, heavy) 71-75 Snow Showers (no Intensity, very light, light, moderate, heavy) 76-80 Thunderstorms (no Intensity, very light, light, moderate, heavy) “Hazards” Table No Hazards 1 Frequent Lightning 2 Gusty Winds 3 Heavy Rain 4 Damaging Wind 5 Small Hail 6 Large Hail 7 Outlying Areas 8 On Bridges and Overpasses

Example: Weather in ArcView In “/degrib/ArcView/”, there is a weather.avl which was applied to the “Wx-Inten 1” to create this image. Problem: Chance of thunderstorms (T), and severe T have the same color. Intent: get people started. If you have a better .avl, and want to share, let me know.

Weather & .flt “-SimpleWx” <NoWx> 20 R/IP (Rain / Ice Pellets) Low 40 R/ZR (Rain / Freezing Rain) High 1 L (Drizzle) Low Probability 21 S/IP (Snow / Ice Pellets) Low 41 ZR- (light Freezing Rain) High 2 R- (light Rain) Low 22 R/S (Rain / Snow) High Probability 42 ZR (Freezing Rain) High 3 R (Rain) Low 23 RW/SW (Rain / Snow Showers) High 43 ZR+ (heavy Freezing Rain) High 4 R+ (heavy Rain) Low 24 R/IP (Rain / Ice Pellets) High 44 5 R/T (Rain & Thunder) Low 25 S/IP (Snow / Ice Pellets) High 45 ZR/IP (Freezing Rain / Ice Pellets) High 6 RW (Rain Showers) Low 26 IP- (light Ice) Low 46 SW (Snow Showers) Low Probability 7 RW/T (Rain Showers & Thunder) Low 27 IP (Ice) Low 47 S- (light Snow) Low 8 T (Thunderstorms) Low 28 IP+ (heavy Ice) Low 48 S (Snow) Low 9 L (Drizzle) High Probability 29 ZL (Freezing Drizzle) Low 49 S+ (heavy Snow) Low 10 R- (light Rain) High 30 R/ZL (Rain / Freezing Drizzle) Low 50 SW (Snow Showers) High Probability 11 R (Rain) High 31 ZR- (light Freezing Rain) Low 51 S- (light Snow) High 12 R+ (heavy Rain) High 32 ZR (Freezing Rain) Low 52 S (Snow) High 13 R/T (Rain & Thunder) High 33 ZR+ (heavy Freezing Rain) Low 53 S+ (heavy Snow) High 14 RW (Rain Showers) High 34 R/ZR (Rain / Freezing Rain) Low 54 Fog 15 RW/T (Rain Showers & Thunder) High 35 ZR/IP (Freezing Rain / Ice Pellets) Low 55 Haze 16 T (Thunderstorms) High 36 IP- (light Ice) High 56 Smoke 17 Severe Thunder storms 37 IP (Ice) High 57 Blowing Snow 18 R/S (Rain / Snow) Low Probability 38 IP+ (heavy Ice) High 58 Blowing Dust 19 RW/SW (Rain / Snow Showers) Low 39 ZL (Freezing Drizzle) High

Example: .flt “-SimpleWx”

NDFD CONUS Map Projection For CONUS (continental U.S.), NDFD uses a Lambert Conformal Conic Map Projection: Tangent latitutude 25, orientation longitude -95, mesh latitude 25, mesh size 5079.406 m. In ESRI ArcMap that would be: 1st Parallel = 2nd Parallel = 25 degrees north (tangent latitude) Lon of center of projection = -95 degrees (orientation longitude) Lat of center of projection = 25 degrees (tangent latitude) False easting = 0 m, False northing = 0 m, X Shift = 0, Y Shift = 0 In addition, NDFD uses a spherical earth with radius 6371.2 km Note: 6367.47 km is currently in the GRIB2 messages, but NDFD used the same methods that NCEP used, which depended on the 6371.2 km sphere. To assist ESRI ArcView 3.x, degrib creates a .ave (ArcView Avenue script) to set the projection and radius correctly.

Un-projected vs Projected

When Earth Radii go Bad. GOOD: Point used 6371.2 km Polygon used 6371.2 km BAD: Point used 6371.2 km Polygon used 6367.47 km

Degrib: Using superImageGen To generate images using superImageGen Select the “Download” tab Highlight the sector in question Download the data Click on “Generate Images” This results in the following: Tkdegrib generates a set of “.mosaic” files “superImageGen” generates the corresponding .png images htmlmaker generates a set of web pages to browse the .png files IE is called to view the web pages.

CONUS MaxT using superImageGen

Degrib: Probe point This is for people who know their lat/lon and just want text output. To probe a point from a command prompt (not available yet in “tkdegrib”) : C:\ndfd\degrib1.2\bin\degrib maxt.bin –P –pnt 38.99,-77.03 C:\ndfd\degrib1.2\bin\degrib maxt.bin –P –pntFile point.txt The first example probes just 1 point, while the second reads a file which is of the form: Station ID, lat, lon per line. The second form allows for numerous stations. By default, probe chooses the “nearest” grid cell, but using the “-Interp” option, causes it to perform bi-linear interpolation. Everything previously discussed can be done using degrib, except data download, so the degrib step can be automated. See (./degrib1.2/docs/degrib.txt for help)

Probe point (MaxT, MinT, PoP) element, unit, refTime, validTime, (38.993600,-77.022400) PoP12, [%], 200307212300, 200307220000, 39.000 PoP12, [%], 200307212300, 200307221200, 39.000 PoP12, [%], 200307212300, 200307230000, 50.000 PoP12, [%], 200307212300, 200307231200, 50.000 PoP12, [%], 200307212300, 200307240000, 50.000 PoP12, [%], 200307212300, 200307241200, 50.000 PoP12, [%], 200307212300, 200307250000, 20.000 PoP12, [%], 200307212300, 200307251200, 20.000 PoP12, [%], 200307212300, 200307260000, 14.000 PoP12, [%], 200307212300, 200307261200, 5.000 PoP12, [%], 200307212300, 200307270000, 5.000 PoP12, [%], 200307212300, 200307271200, 11.000 PoP12, [%], 200307212300, 200307280000, 11.000 PoP12, [%], 200307212300, 200307281200, 21.000 PoP12, [%], 200307212300, 200307290000, 21.000 element, unit, refTime, validTime, (38.993600,-77.022400) MaxT, [F], 200307212300, 200307230000, 86.882 MaxT, [F], 200307212300, 200307240000, 82.375 MaxT, [F], 200307212300, 200307250000, 85.233 MaxT, [F], 200307212300, 200307260000, 87.753 MaxT, [F], 200307212300, 200307270000, 89.395 MaxT, [F], 200307212300, 200307280000, 87.775 MaxT, [F], 200307212300, 200307290000, 86.800 MinT, [F], 200307212300, 200307221200, 72.625 MinT, [F], 200307212300, 200307231200, 73.165 MinT, [F], 200307212300, 200307241200, 69.025 MinT, [F], 200307212300, 200307251200, 65.965 MinT, [F], 200307212300, 200307261200, 67.945 MinT, [F], 200307212300, 200307271200, 69.745 MinT, [F], 200307212300, 200307281200, 69.025

Degrib: Future? Maintain the program by keeping up with NDFD as more variables / sectors are introduced. In order to inform users of updates, degrib has an “announcement mailing list”, which currently has 119 people on it. Improvements: Based on user feedback, current thoughts are: Create merged polygons. Instead of a polygon per cell, join similar values together to form large polygons. Create some form of graphical probe point capability Create some form of un-compressed database so that a CGI callable probe program can be used to access the data quickly Add some way for UNIX users (without GrADS) to plot the images Add some way for users to zoom in on a map Interface to more file formats