Collier
County ARES/RACES
PROPOSED NWS PROTOCOL FOR APRS
Dale
Huguley, KG5QD
kg5qd@earthlink.net
Collier
County ARES/RACES
Dale
Huguley, KG5QD
 |
Background: This proposal is an outgrowth of a project at the
Collier
County Florida Emergency Operations Center called WXSVR (weather
server or weather severe). Information arriving on the GTE weather wire
from the National Weather Service
via satellite is broken into "products" and sent to a co-located packet
BBS by way of a serial cable using AX.25 protocol and is accessable via
VHF packet radio by any amateur in range. Tropical weather updates are
sent as bulletins to the Florida BBS network automatically using a protocol
developed by Gary Arnold, WB2WPA.
In 1996, an attempt was made to add APRS output of the positions of tropical cyclones (the generic name for hurricanes, typhoons, tropical storms and depressions). While somewhat successful, this approach left a lot to be desired as far as stability. After meeting with Keith Sproul at the second annual Amateur Radio hurricane meeting at the National Hurricane Center during the 1997 Miami 'Hamboree' I began work on an extended tropical cyclone protocol with help from Keith and Mark Sproul. Mark is working on an interface for WinAPRS from the GTE wire and EMWIN (Emergency Managers Weather Information Network). EMWIN is available on satellite and on VHF in some locales. Thinking through the problems associated with APRS "objects" coming from multiple sources has led me to the following proposal. |
The APRS system is designed to plot and track "objects" which contain information which determined by individual operators, or by local automatic systems such as weather stations or GPS units. Information parsed from NWS bulletins are fundamentally different in that they are "official" and are universal. This means that the source station should be irrelevant to the receiving station and multiple paths should only add to reliability, not clutter. Therefore, objects generated have to be named consistantly and be time stamped the same for the system to sort out what is the most up to date. There will be need a character in the APRS packet which tells the system that this is automatically derived universal information and should be treated differently that normal APRS packets. I will use the proposed Tropical Cyclone Protocol to illustrate this system.
Example 1 is a "marine" forecast from the National Hurricane Center for Hurricane Danny. This type bulletin is normally sent at 0300Z, 0900Z, 1500Z, and 2100Z which is a six hour interval. It contains the present and predicted positions with sustained winds, gusts, etc., and the wind fields associated with those positions. The APRS system allows this to be transmitted in very efficient manner using the following protocol by the author of APRS Bob Bruninga, WB4APR who originally created APRSDos.
DANNY *DDHHMM/LAT/LONG@CSE/SPD/TS/www^GGG/ppp>RRR&rrr | | | | | * - - Radius of TS winds | | | | * - - - - Radius of HC winds | | | * - - - - - - Central Pressure | | * - - - - - - - - Gusts | * - - - - - - - - - - Sustained winds * - - - - - - - - - - - - HC, TS, or TDwhich for the "Present Position" is:
DANNY *1809z2930.00N/08948.00W@.../.../HC/065^080/989>020&100
(the dots in the course and speed are because "DRIFTING" is not parsable to a vector and speed)
Wind fields are extremely important in that they define the dangerous parts of the tropical system. Last year Bob, WB4APR devised a very clever offset for the wind fields based on forward motion and vector which works very well at forward speeds of 7 knots (or so) and above. Contained in the forecast is also the "gale force" wind field, i.e. winds above 50 knots (whole gale). In fact, world-wide the while gale wind field is more universal. It is available for the 48 and 72 hour forecast positions for Atlantic and Pacific basin (out to 180 degrees) cyclones where they are the only wind fields provided.
Also "WMO" type forecast positions for Western Pacific, Indian ocean and Southern Pacific (I think) have whole gale wind fields plus others. Taking these and other factors into account I propose the following:
Extended APRS Tropical Cyclone Protocol
DANNY *180900z2930.00N/08948.00W@340/010/HC/065^080/989>see wind fields below | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * Central Pressure in | | | | | | | | | * millibars 3 digits | | | | | | | | | * "1" truncated > 1000 | | | | | | | | * Max Gusts in Knots | | | | | | | * Max Sustained Winds in Knots | | | | | | * Type (see Type Protocol) | | | | | * Forward Speed in Knots | | | | * Course in Degrees | | | * Standard APRS type position | | * TIME- always time that bulletin is valid | * "*" Means object in APRS (good place to define NWS derived object) *-----NAME (see naming protocol) padded to 9 characters
Wind Fields
(added after ">" to packet above)
Present Position:
020&100%030v~nn~~xx~~zz 15A *---Package number (2 digits, plus letter if needed) | | | | | | | | | * Diameter of Eye (space indicates not available) | | | * 12 character compressed wind fields offsets defined as: | | | * Trop Storm Gale Force Hurricane | | | * NE SE SW NW NE SE SW NW NE SE SW NW | | | * Character =(126 - ((max radius - quadrant radius)/5)) | | * Maximum Radius of Gale Force Winds (50 knots) in Nautical Miles | * Maximum Radius of Tropical Storm Force Winds (35 knots) in Nautical Miles * Maximum Radius of Hurricane Force Winds (64 knots) in Nautical Miles Eye diameter is defined as (126 - (Eye Diameter/5))The Hurricane Center reports or predicts wind fields in 5 nautical mile increments so this scheme defines the wind fields as accurately as the NHC reports them.
The icon for the "Present Position" should be the "Tropical Cyclone" icon.
Predicted Positions:
For the predicted positions add the following field to the end of the wind fields: VLDDDHH or VLD1803 which would read "Prediction for the 18th at 0300z"
Predicted positions will never have course, speed, or barometic pressure fields so at some future time it may be possible to shorten these packets by 12 characters by defining a seperate object as a "predicted position object"
The icon for "Predicted Positions" should be a dot possibly colored to show class of storm. (suggested by Bob, WB4APR)
Reference Position:
Each "marine forecast" has the position when the computer run was done (the synoptic time) and gives one a "bread crumb" to trace the object and also is the time to which the predicted postion's time is relative. Since this object carries only time and position it can be used to carry additional information that will not fit into any of the other packets.
DANNY_REF*180900z2912.00N/08954.00W@VLD1806NXT1815nxt1812 | | | | | * Next Intermediate Update | * Next Complete Update * Reference TimeNote: these times are always Zulu and are on the hour.
The icon for the Reference Position should be a gray dot.
Naming Protocol
In order for objects to be updated from multiple sources and at multiple times, a naming convention has to be consistent from update to update. This precludes having the time of a predicted position imbedded in the name. In the discussions published by the Hurricane Center, they refer to the 12,24,36,48 and 72 hour positions. These are relative to the reference time and position which is known as the "synoptic". We will adopt this convention as follows:
Present Position Name:
If Named Cyclone- 9 characters ALL CAPS padded with spaces or truncated If Numbered Cyclone -TD for Atlantic - ED for Eastern Pacific - CD for Central Pacific - WD for Western Pacific etc. plus two digits for number padded with a zero for single digit.Predicted Position Names:
Name from above padded to 5 characters with "_" or truncated to 5 characters plus "+12h" for the 12 hour predicted position "+24h" for the 24 hour predicted position "+36h" for the 36 hour predicted position "+48h" for the 48 hour predicted position "+72h" for the 72 hour predicted positionReference Position Name:
First 5 characters as above plus "_REF"
Type Protocol:
In order to make APRS more universal I propose: "HC" for hurricane (Atlantic and Pacific to 180 degrees) winds > 65 Knots "TS" for Tropical Storm i.e. winds > 35 knots but < 65 knots "TD" for Tropical Depression - closed circulation "TY" for Typhoon "ST" for Super Typhoon "CY" for Cyclone which is what the Typoon/Hurricane type is in the Indian Ocean and the South Pacific "EX" for an extratropical cyclone in the transition phase to non-tropicalExample 2 is a "Public" advisory which can come 6, 3, or 2 hour intervals, but is generally issued on a 6 hour basis. The NHC goes to 3 hour intermediate advisories (consisting of the public advisory only) when the center is within range of a coastal radar, and 2 hours just before landfall. Between the intermediates, very brief tropical cyclone estimates (MIATCEATx) or updates (MIATCUATx) are issued. This information is intended for human reading and is therefore harder to parse consistantly for APRS. But I have had good enough results, especially with the center of circulation, to issue an updated "Present position" on the off times the more reliable "marine" parsing is not available. It may be possible to represent this "intermediate" position with a seperate icon to denote the wind fields are rudamentary.
The "Public," a simplified text for general consumption, gives a short informative "Bulletin" between the end of the header and the place the present position is sent. This area generally contains headlines (denoted by ... on either side), areas for watches and warnings, and other important information. Limiting the line length to 65 characters gives 12 lines for the longest I could find in 1996, most of the time very much less. You can bet the longer the text between the header and the position statement, the more important the information is.
An example from Lili package 15 "public" from last year:
BLN@LI?01:"...LILI EXITING THE NORTH COAST OF CUBA AFTER WREAKING BLN@LI?02:HAVOC...HURRICANE WARNINGS ARE IN EFFECT FOR THE BLN@LI?03:CENTRAL AND NORTHWEST BAHAMAS AND FOR MAINLAND CUBA BLN@LI?04: FROM MATANZAS PROVINCE EASTWARD TO CAMAGUEY. RADAR BLN@LI?05:INDICATES THAT EYE OF LILI EXITED THE NORTH COAST OF BLN@LI?06:CUBA NEAR CAIBARIEN." | | | * Two digit Line number | | * "Package" Number (see below) | * First two letters of named Cyclone- two digits if numbered * BLN@ Designator for Tropical BulletinPackage Number Character is defined as (decimal 48 + package Number) so that package 15 would have a "?" (ascii 63) as its character. An intermediate advisory will have an "A" for the first intermediate, "B" for the second, and "C" for the third, attached to the package number in the bulletin so to make these sort properly the two digit line number will start with 21 for "15A", 41 for "15B", and 61 for "15C", for example.
(I am not including the ending {1 stuff as I hear it is no longer needed.)
This assures that each line number is unique during any given season for named storm (and probably unique for a given time for a numbered cyclone) allowing multiple paths and times for reception. I have a problem reusing the same BLN number because there is no way to know if you are viewing an update, or mixed lines from various bulletins, giving confusing results.
An attempt will be made to parse the Watch and Warning areas from the "Public" Advisory and transmit this information via a format suggested by Keith Sproul for a graphical overlay for maps. My discussion with the folks at the Hurricane Center about the "breakpoints" reveals that the points are fairly fixed for the United States and the areas that the Hurricane Center serves as the meteorological service, but in other areas the local governments determine the points on an "ad-hoc" basis making parsing problematical. For this reason the watches and warnings should be sent both as text and graphically.
(Click on the image to isolate and enlarge it. Use the "BACK" button to return to this page.)
In WinAPRS/MacAPRS, a tropical cyclone watch area should be defined by an outline of the coast- Red for Hurricane Watch, Yellow for Tropical Storm Watch. Filling the area between the outline and the coast denotes "Warning". This takes care of the common case of an area being under a watch and warning at the same time. I have no idea if this can be done in APRSDos.
SKYWARN
OPERATIONS
Example_3 is a Tornado Watch Bulletin which has imbedded corners for the watch box, and useful information for planning skywarn operations. The NWS uses a format of 356,1047 for position which translates to 35.6N 104.7W - with points separated by spaces. This is a compact method of thansmitting multiple points, so we will use this method with some modifications. I would like to keep this packet to be "human readable" because it would be useful to be able to send it as an alpha-numeric page to Skywarn spotters as a "heads-up" message. With a map, a spotter away from his ham equipment can still get an idea about when and where he might be needed.
TORWCH226*012027z3830994,3540994,3531020,3831021EXP020300zMXT500HAL212GST075MMV26040 (see below) | | | | | | | * Max Surface Gusts in MPH | | | | | | * Max Hail in inches | | | | | | * with fractions | | | | | * Max Cloud Tops in hundreds ft | | | | * Expiration Time in Zulu | | | * NWS type positions separated with commas: may be 3 or more | | * Issue time: I would rather this be "Valid Time" (hard to parse) | * Object character: good place to denote NWS derived object * Type of watch and numberThe MMV is the "Mean Storm Motion Vector" which is the direction and speed the storms should be moving FROM, not TO, in this case moving from West Southwest at 40 knots! The Discussion portion of the watch bulletin could be sent as a bulletin using somewhat the same format as the Hurricane Bulletins:
Types of watches:
SVRWCH - Severe Thunderstorm (yellow outline. Comments requested on
visibility of yellow on APRS maps!)
TORWCH - Tornado (red outline)
(Click on the image to isolate and enlarge it. Use the "BACK" button
to return to this page.)
The discussion portion of the watch bulletin contains useful information which can be sent as a short bulletin using a format similar to the tropical cyclone bulletin:
BLN=T226A:DISCUSSION...STRONG THUNDERSTORMS DEVELOPING ALONG COLD FRONT NEAR BLN=T226B:CO/KS BORDER EXPCD TO BUILD SOUTHEASTWARD ALONG DRY LINE INTO AREA BLN=T226C:AHEAD OF DRY LINE BULGE. PROFILERS INDICATE SHEAR PROFILE QUITE BLN=T226D:STRONG FOR SUPERCELL DEVELOPMENT.The "=" could be almost any character. Using a character besides a letter or number would allow the output programmer to move the weather bulletins to another area and leave the regular APRS bulletins uncluttered. Since both the Tropical Cyclone and Skywarn type bulletins generate graphic and text information, it would be useful to tie the object to the text associated with it, i.e. be able to see both on the screen at once.
System Issues:
When the WinAPRS interface is available for the GTE weather-wire and EMWIN, there will be multiple stations sending the same information, great for backup, but a potiential problem with channel load. My experience with the "Thor" exercise during May was that a great deal of information could be sent on HF if it was done in a "laid-back" fashion. You have about 2 hours to move about 20 packets throughout the tropics and send it enough times to make sure it made it thru the qrm and fading. Once per hour from a given station is probably enough if multiple stations are transmitting. Remember the packets will have the same things in them: the only difference would be the path portion of the packet.
On VHF, I plan to send the information to TROPIC,TROPIC with "TROPIC" being an alias for the KG5QD-3 digi in Southwest Florida. If other areas want the information before their area has an automatic server, a local digi can changed from "WIDE" to "TROPIC" temporarily or changed to the new Paccomm firmware which allows 4 aliases. On HF, if the path from a server is set to TROPIC,TROPIC, then a well situated station could set its HF digi to "TROPIC" from "ECHO" to bounce the packets back to an threatened area, or set its gate to "TROPIC" from "GATE" to provide the info on vhf locally.
Use of a path which will keep local information close to home, but allow fringe areas to still receive vital information, is possible with the new firmware.