Aerodrome MET Observation and Forecast Discussion 2024 Feb 28 - team-sx/iwxxm-sx GitHub Wiki

2024-Feb-28, 14:00-14:00 UTC, Webex

1. Action items of last discussion
   • Ilkka will see if TSML has primitive types that can be used to describe time-value pairs in iwxxm:PhenomenonProperty
   • Choy will find out if there is any information model on probabilistic information
   • All are invited to share existing and potential use cases on consumption of MET information, including those in other aviation domains like air traffic control, search and rescue operations, etc., as well as challenges on creating mock ups, on GitHub Discussion
2. IWXXM-SXRC2 has been made available that includes changes in IWXXM 2025-2RC1
3. Geospatial community's views with regard to the physical implementation of information models

• Regarding TSML:

  • Ilkka prepared examples on using TSML TVP as part of the aerodrome forecast content model
  • TSML TVP supports the provision of "measurement uncertainty". It may not be best suited for describing forecast uncertainties
  • The examples depicts the structure of a time series at indicated time instants. The representation would be much simpler if the time interval is fixed
  • While it is possible to express MET parameters with multiple values (e.g. wind has speed and direction) in separate time series with TSML TVP, it may not be desirable to do so in view of possible duplication of representation as well as increased chance of errors

• Regarding probabilistic information:

  • ECMWF and NCEP have descriptions on their probabilistic products
  • Apart from (i) probabilities of exceeding a threshold and (ii) percentile accumulations, one can also provide (iii) CDF/PDF to allow the consumer to do their own calculation (see example below)

  

  • For (i) and (ii), we may want to have a standard set of thresholds and percentiles for use. Producer provision of these values should also be allowed. For (iii) one may provide the first 3 moments of the distribution, or data points describing the CDF/PDF

  • Additional CDF/PDF on climatology and/or model climatology may be able to provide extra information conducive to conclusions like confidence of prediction or occurrence of extreme?

  • Available standards:

    • Checking WMO No.306 Vol. I.2 there are (GRIB) templates and (BUFR) sequences on provision of probability and percentile data but not the associated CDF/PDF
    • The OGC Uncertainty Markup Language (UncertML) is a conceptual model and it has representations on probability distributions (Caution: Don't visit the website www.uncertml.org mentioned in the document as it is no longer showing information on Uncertainty Markup Language):

    

    • The OGC NetCDF Uncertainty Conventions (NetCDF-U) is based on UncertML and has provisions to encode distributions:

    

  • While it is uncertain whether a consensus could be reached in the provision of CDF/PDF, it would be good to include representations of CDF/PDF as optional items in the information model for future use

  • There were discussions on the information provided by human against those derived directly from weather prediction models like probabilistic forecasts:

    • Forecasts are "best guesses" from human forecaster taking into account of all the information the forecaster has, including outputs from weather prediction models
    • Human intervention are normally involved to ensure consistency of all information provided in a forecast
    • The inclusion of probabilistic information, however, creates a problem as it is usually derived directly from weather prediction models and it would be hard for a forecaster to manipulate if not impossible
    • If we put forecaster's "best guesses" side-by-side with information derived directly from weather prediction models like CDF/PDF with no intervention, consumers could be confused. This could be a safety hazard
    • Having said that, it is well known that users like airlines, after reading authentic forecasts from MET producers, will also check against information from unofficial sources like Windy to make decisions upon the best information they can obtain. May be a clear indication of those parts in a forecast that are authentic and those parts that are providing additional/supplementary information can help? We may probably need a debate on this to clarify the function of a forecast particularly within WG-MRAD.

• Regarding use cases:

  • Jens has tried to create an airport forecast and would be further explored.
  • Joonas has raised a question on the provision of information regarding representation of cloud

• Regarding views on physical implementation of information models:

  • Ilkka commented on the use of XML vs. JSON in OGC standardization:
    • Despite the fact that XML technologies and schemas are still actively used in various application domains, and that there is technically nothing wrong in XML / XML Schema based workflows, JSON has been been the weapon of choice for web developers for many years. Generally XML processing is seen too complicated and resource-heavy by software developers, especially in the context of web APIs and browser-based web applications.
    • In OGC there has been very little activity in developing new XML based data models or APIs standards or their revisions in the past few years. The 2nd generation of OGC geospatial web APIs, such as OGC API Features and OGC API EDR, gradually replacing 1st generation OGC Web Services (WMS, WFS, WPS, CSW) are based on JSON technology and the Open API specification. While XML payloads are still supported in these APIs, they are not used very often.
    • XML still has its place in data exchange between existing information systems, especially when the data is not provided or used in web browser applications. XML Schema an Schematron technologies currently provide more standardized and stable data validation capabilities than JSON Schema, although the situation is likely to change for the better for JSON content.
    • Holding on to XML techonologies, without considering JSON based alternatives in parallel, is risky, because the tool support and technical skills for producing and using XML is bound to decline as XML based data exchance gradually transitions from mainstream to niche.
    • The OGC Testbed-17: Aviation API Engineering Report contains a good discussion on the benefit of GeoJSON and JSON-FG as alternatives to the XML based data models and APIs in the aviation domain, see Chapter 13.2.3.1. "Use of Encodings. The Value of JSON-FG" in particular.
  • Universities are no longer teaching XML and it is anticipated that works based on XML will be obsoleted/superseded in near future
  • Choy will bring this up in the upcoming CWGP/2 on the general trend of leaving XML behind with the use of other formats like JSON, and to seek views from AIXM and FIXM teams

• Ilkka will take a look at TSML TVP on the possibility to describe a time series of MET parameter with multiple values like wind

• Date: With the week 11-15 Mar 2024
• Additional discussion topics:
  1. Do we want to pull together WxObjects (i.e. iwxxm:MeteorologicalFeature), and may be addition information, to create an iwxxm:MeteorologicalFeatureCollection to tell a story (c.f. TAF) and what they are
  2. Whether we allow distribution of WxObjects without the use of iwxxm:MeteorologicalFeatureCollection in certain information services
  3. We may need to identify typical use cases to substantiate decisions regarding the above

• BL Choy
• Dirk Zinkhan
• Joonas Eklund
• Ilkka Rinne
• Kuscherka Jens