Improve TAF BECMG time support

marqh commented 7 years ago

In TT-AvXML-4 Mark Oberfield noted that there are two times of interest on a TAF BECMG group: a transition period and a period when the phenomenon exists. Both may need to be represented on TAF, whereas only one is carried today.

"OM_Observation provides a time at which the procedure completed (the result time) and the time or time-period when the result is relevant (the phenomenon time).

The validPeriod defines the period during which this TAF is valid.

TAF HGK 221754Z 2218/2318 18007MPS 4000 BR SCT100

BECMG 2300/2302 NSW CAVOK

What is the phenomenonTime for the BECMG group? Is the TimePeriod two hours long? Or 18 hours from 00Z to 18Z on the 23rd?

If it’s the former (two hours), for a WFS request what forecast will be returned if user request conditions at 06Z on the 23rd for Hong Kong International Airport? If it’s the latter, then how will the two hour duration for the BECMG group be derived? (problem for converting TAF XML->TAC) To my understanding of phenomenonTime, I believe the latter option is correct."

ref; https://software.ecmwf.int/issues/browse/AVXML-32

Created: 29/Sep/15 9:45 PM

Reporter: Aaron Braeckel

Tertzoid commented 7 years ago

I think latter intrepetation is "more" correct. In your example baseForecast phenomenonTime TimePeriod 2218-2302 and changeForecast (BECMG) TimePeriod 2300-2318. So Periods overlaps 2300-2302.

In example: http://schemas.wmo.int/iwxxm/2.1/examples/taf-A5-1.xml FM group looks like right TimePeriod, but BECMG "only" as in TAC. FM also repeats all weather groups, which makes life easier. Hm, maybe not best example, some other mess with times too. Sadly many IWXXM converter implementations uses former intrepetation.

braeckel commented 6 years ago

In TT-AvXML 7 it was agreed:

The definition of phenomenonTime is a bit stretched and less than ideal for becoming conditions
Additional documentation/guidance should be added to the TAF MeteorologicalAerodromeForecastType phenomenonTime to indicate the proper interpretation in these cases

Further, conclusions were that in the long-term:

A "derived" set of meteorological conditions in sectional time series was desirable rather than the change indicator approach used today. This is more consumer friendly in that no work is required to interpret the TAF conditions at any point in time. This is also easier to generate directly from weather models
This TAF reporting change would have implications on data producers and would therefore require METP discussion, and there is not any urgency to work on this

wmo-im / iwxxm

Improve TAF BECMG time support #13