Wikidata talk:WikiProject Tropical cyclones

Launch of WikiProject Wikidata for research

Hi, this is to let you know that we've launched WikiProject Wikidata for research in order to stimulate a closer interaction between Wikidata and research, both on a technical and a community level. As a first activity, we are drafting a research proposal on the matter (cf. blog post). It would be great if you would see room for interaction! Thanks, --Daniel Mietchen (talk) 01:37, 9 December 2014 (UTC)

Wikimania 2016

Only this week left for comments: Wikidata:Wikimania 2016 (Thank you for translating this message). --Tobias1984 (talk) 12:00, 25 November 2015 (UTC)

Change of properties for cyclone items

I'm working to fill infobox tropical cyclone related (in cawiki) with WD information.

1. There is no property for "category", because when it was proposed, people rejected it assuming that can be calculated via maximum sustained winds (P2895).

   I checked how present infoboxes calculate the category and, in addition to the speed wind, could be necessary the basin (Atlantic, NW Pacific, NE Indian, etc...) and, in some of these basins, the duration and wind gust. It is, we need more than one new property to calculate just one rejected property. Even more, there is no only one categoritzation category, but one (or more) for each basin.

   In any case, even knowing the formula, the category attribution is finally certified by en:Regional Specialized Meteorological Center of each zone.

2. The subclasses were poorly structured.
3. The instance of (P31) of events are mainly focused in one of the generic names: tropical cyclone (Q8092), North Atlantic tropical cyclone (Q2547976), typhoon (Q140588).

To be able to have the category without a new property, I have done:

• create an entry for each classification criteria (in en:Tropical cyclone scales):
  • Saffir–Simpson hurricane wind scale (Q205801). For Atlantic North, Eastern and Central Pacific Tropical Cyclones (Q63100492) >>> North Atlantic tropical cyclone (Q2547976)
  • JMA typhoon wind scale (Q63100860) and JTWC typhoon wind scale (Q63100804). For Northwest Pacific tropical cyclone (Q63100732) >>> typhoon (Q140588)
  • IMD cyclone wind scale (Q63106281). For North Indian Ocean tropical cyclone (Q27491739)
  • MF cyclone wind scale (Q63107049). For South-West Indian Ocean tropical cyclone (Q27491742)
  • BOM/FMS cyclone wind scale (Q63166758). For Australian region tropical cyclone (Q1284783) and South Pacific tropical cyclone (Q1284799)
• create an entry for each value of the en:Tropical cyclone scales
  • Include them inside the subclass of (P279) tree.

With this elements, my proposal for each event is:

• Exchange the P31 value by the specific category + qualifier with the corresponding scale. : Ex.:

⟨ Hurricane Easy (Q272072)    ⟩ instance of (P31) ⟨ Category 4 hurricane (Q63100601)    ⟩
determination method or standard (P459) ⟨ Saffir–Simpson hurricane wind scale (Q205801)    ⟩

• For those cases with two classifications because the cyclone cross from one basin to another, the P31 could be multivalue and should incorporate P518 with the basin:

⟨ Hurricane Easy (Q272072)    ⟩ instance of (P31) ⟨ Category 4 hurricane (Q63100601)    ⟩
determination method or standard (P459) ⟨ Saffir–Simpson hurricane wind scale (Q205801)    ⟩
applies to part (P518) ⟨ North-East Pacific Ocean (Q63860551)    ⟩

• For subtropical or extratropical cyclons, in addition to its classifications, add a P31 to indicate this complementary characteristic:

⟨ Hurricane Easy (Q272072)    ⟩ instance of (P31) ⟨ subtropical cyclone (Q2331851)    ⟩

• Add located in/on physical feature (P706) with the value of the specific ocean/sea basin where the cyclone was : Ex.:

⟨ Hurricane Easy (Q272072)    ⟩ located in/on physical feature (P706) ⟨ North Atlantic Ocean (Q350134)    ⟩

• Add part of (P361) pointing to the season item/article. Ex.:

⟨ Hurricane Hilda (Q6854262)    ⟩ part of (P361) ⟨ 1955 Atlantic hurricane season (Q2445148)    ⟩

• Add part of the series (P179) with the previous and next event of same category scale. For those cases with two classifications because the cyclone cross from one basin to another, the P179 could be multivalue. Ex:

⟨ Hurricane Hilda (Q6854262)    ⟩ part of the series (P179) ⟨ North Atlantic tropical cyclone (Q2547976)    ⟩
follows (P155) ⟨ Hurricane Ione (Q5947817)    ⟩
followed by (P156) ⟨ Hurricane Janet (Q3303181)    ⟩

⟨ Hurricane Hilda (Q6854262)    ⟩ subclass of (P279) ⟨ recurrent event edition (Q27968055)    ⟩

, to avoid a P179 constrain

I'm preparing a quickstatement to make this changes gradually.

I'll appreciate any suggestions. Thanks, Amadalvarez (talk) 21:50, 15 April 2019 (UTC)

This scheme needs to be discarded almost entirely (sorry, I really honestly believe that about your classification scheme) due to numerous problems, some ontological and some related to computer science. The main problem is that it is not in Boyce–Codd normal form (Q2460153), leading to much redundancy (and constraint violations; BCNF is useful for preventing constraint violations in the first place). Other problems are:

• part of the series (P179) should be used to indicate the season, not storms of a given category or type; part of (P361) could be used instead. The current linked list (Q7003418) data structure (Q175263) is not resilient at all and is very cumbersome for when a storm needs to be retroactively added to or removed from a category (as is expected from the Atlantic hurricane reanalysis project (Q4816630) or any scale definition change). It is also inefficient for random access. Also, linked lists require an order to be defined on the set; what order is chosen? (formation date? date the storm first reached that category? etc.).
• located in/on physical feature (P706) should not be used here; instead, we should create a new property called "tropical cyclone basin". These basins, unlike things like "south Atlantic ocean", have very strictly defined boundaries, unlike the terrain features used. In addition, especially in the western Pacific, the specific terrain feature is often different from the main ocean (for example, the South China Sea (Q37660) as a subset of the western Pacific basin).
• instance of (P31) emphatically should not be used to indicate storm category. This is a blatant violation of BCNF, and is essentially circumventing the consensus that the category should be computed from the peak intensity value. It should simply take on values of "tropical cyclone" or "subtropical cyclone".
• maximum sustained winds (P2895) needs to be qualified by agency, especially outside the central/eastern Pacific and Atlantic basins where there are usually many agencies. The Regional Specialized Meteorological Center (Q1326212) value should of course take primary rank. It also needs to be qualified by duration of sustained winds; this could be a new property called "sustained wind time", and the values are time, either 1-, 3-, or 10-minute. In particular, in all basins not monitored by the Central Pacific or National Hurricane Centers, we will have both 1-minute and either 3- or 10-minute.

Enhancements we need are:

• Only limited data on each tropical cyclone is kept currently. We can keep far more information. For each item to be more useful, we should import the entire best track into the storm's Wikidata item. This would allow nice things such as automatic computation of duration at a given category, automatic calculation of accumulated cyclone energy (Q339687), and what not. To do this, we need to create a property called "best track entry" that will store a given entry of a best track file; each entry contains the time, location, and intensity of the storm, as well as many other data fields, and could be imported automatically (for example HURDAT). I propose the following format:
  • Value: time.
  • Qualifiers (all of which that are supplied in the given source; not all will always be present): maximum sustained winds (P2895), location (P276), lowest atmospheric pressure (P2532), radius of wind (new property, see below), speed (P2052) (for storm movement, not intensity), direction (P560) for storm movement direction, gusts (new property), best track status (new boolean property, true if finalized, false if provisional), radius of outermost closed isobar (new property), is a landfall (new boolean property), etc.
• We need a new radius of wind property to indicate 34-knot (gale), 50-knot (storm), and 64-knot (hurricane) force wind radii to act as a qualifier to the above best track property. Its value will be a scalar, and it will be qualified by direction (P560) for which quadrant (northeast, northwest, southwest, southeast), either maximum sustained winds (P2895) or a new property to specify which wind type (gale, storm, or hurricane).
• As an exception to BCNF, we should keep maximum sustained winds (P2895) and location (P276) as computing it from the best track claims can be cumbersome for such a frequent operation; both should be qualified by agency.
  • We should also include notable measurements, such as the 899 hPa measured for Typhoon Rammasun (Q17342662) and the 120-knot gust measured in Cyclone Winston (Q22817489). Measurements should be subordinate the the official best track value.
• There should be a property "Fujiwhara interaction with" to hold information about Fujiwhara effect (Q1376614)'s between storms.
  • An "interaction result" qualifier can be added to indicate whether the storm absorbed or was absorbed by the other storm, or merely altered course.

Besides meteorological data, we should also store casualty counts (number of deaths (P1120)) and damage totals on each item. Since many storms have multiple names (today, mostly because PAGASA (Q747963) assigns additional names to storms already named by the Japan Meteorological Agency (Q860935)), we should have a tropical cyclone name property, with value not string but rather, item; since many storms have the same name, we also need an item for each name, on which we can then also indicate information as to where (which basin) the name is used, whether it has been retired, who contributed it, and the storms it was used on (each of which should be a new property). The tropical cyclone name property should be qualified by agency given the frequency of multiple-name storms.

This is just for tropical cyclone items. Season items have to also be considered. The value of image (P18) for them should be the season summary map (for example what I just added to 2015 Pacific hurricane season (Q18604546)). Most season data such as number of named storms, number of major hurricanes, and casualties can be inferred by summing the individual storms' data, so most seasons should just include follows (P155) (for previous season), followed by (P156) (for next season), and a has part(s) (Property:P527) for each storm.

I welcome comments from my fellow English Wikipedia tropical cyclone project members @Cyclonebiskit, Juliancolton, TheAustinMan, TropicalAnalystwx13, Titoxd, JasonRees: etc. to comment on this proposal.--Jasper Deng (talk) 01:28, 5 August 2019 (UTC)

I agree with the principal qualms Jasper notes above, and agree with the proposals forwarded for maximum sustained winds (P2895) and associating a "sustained wind time" property with them. Also agree with some of the proposals forwarded by Jasper. Would support importing International Best Track Archive for Climate Stewardship (Q43269466) to all instances of tropical cyclone (Q8092) with associated properties per time (see IBTrACS), but disagree with adding a "best track status" boolean property given that is hard to determine (because of Atlantic hurricane reanalysis project (Q4816630), Atlantic storms out to 1989 are technically "provisional" which is misleading). Disagree with a "Fujiwhara interaction with" property as that is often not well-established in any sort of verifiable manner. — the•austin•man (works) 02:20, 5 August 2019 (UTC)

We should drop the "applies to" for similar reasons; that doesn't seem to be a very natural or clear way to indicate basin.--Jasper Deng (talk) 04:37, 5 August 2019 (UTC)

• If no one objects, I will begin implementing this new structure by first requesting the properties.--Jasper Deng (talk) 07:15, 11 August 2019 (UTC)

Should unnamed tropical cyclones without popularized names but with associated Wikipedia articles be assigned those articles names or using some other designation? There are storms like 1916 Texas hurricane (Q199003) which do not have any typically used names by popular press or publications, but are so-named simply to describe them on-wiki. the•austin•man (works) 00:19, 13 August 2019 (UTC)

@TheAustinMan: We should use the Wikipedia-used names if no better alternative is available. Machines should be able to still lookup tropical cyclones by official identifies like AL052019.

In fact I think we should even create an item for each storm name. The name item can give its etymology, its pronunciation, whether it was retired, which name it replaced, and what not.--Jasper Deng (talk) 07:14, 15 September 2019 (UTC)

@Jasper Deng, TheAustinMan: Once you finish with all the definitions, new properties and changes in the current information I already uploaded, please make me a ping in order I can apply the changes in my accesses. It's a pity you appear after I made thousands of uploads of statements, but if you see clear and are willing to go ahead, I'll be waiting that you end all the work. Thanks, Amadalvarez (talk) 16:02, 2 November 2019 (UTC)

Wikidata item for each class of storm

Hello. Based on en:Tropical cyclone#Intensity classifications, I went ahead and created the missing Wikidata items for the remaining storm intensities: See here for overview. Individual storms should ideally be tagged with such (i.e. instance of). Obviously, the table can be expanded to include other scales (i.e. Taiwan), and should ideally be saved on this Wikiproject. Thoughts welcome. Cheers, Rehman 12:54, 6 June 2020 (UTC)