Issues with the Adm1 boundaries

[LaurensJN]

Hi, I noticed a few issues with the Adm1 boundaries in the exposure model. I'm only posting the issues that I verified should be different in both the GADM administrative boundaries and OpenStreetMap admin_boundaries. Is it possible to look into these districts and change the respective exposure?

To verify I have used the up-to-date OSM boundaries and the GADM boundaries v4.1.

Africa

Country	Area	Problem
France	Reunion / Mayotte	It's missing from the exposure model
Ghana	Western North	It's a separate unit since 2018, before it was part of Western
Ghana	Bono East & Ahafo	Both are separate units since 2018, before they were part of Brong-Ahafo
Ghana	Savanna & North East	Both are separate units since 2018, before they were part of Northern
Ghana	Oti	It's a separate unit since 2018, before it was part of Volti
Tanzania	Songwe region	It's a separate unit since 2016, before it was part of Mbeya region

Caribean and Central America

Country	Area	Problem
France/Netherlands/separate	Saint Martin / Saint Barthélemy / SSS islands / ABC islands	With exception from Aruba, they are all missing from the exposure model
Nicaragua	Lago Nicaragua	It's a lake, not an administrative boundary. Therefore it should not contain exposure.

Central Asia

Country	Area	Problem
Kyrgyzstan	Osh City	Osh City is not part of the Osh District but a separate entity
Kyrgyzstan	Bishkek City	Bishkek City is a separate entity

East Asia

Country	Area	Problem
North Korea	Kaesong	Split from North Hwanghae Province in 2019

Europe

Country	Area	Problem
Belarus	Minsk	City of Minsk is a separate entity and not part of the Region of Minsk
Estonia	Peipsii	It's a lake, not an administrative boundary
France	All regions	The 22 districts from <2015 have been used, instead of the 13 of nowadays (see https://en.wikipedia.org/wiki/Regions_of_France)
Luxembourg	All regions	The 3 districts from <2015 have been used, instead of the 13 of nowadays (see https://en.wikipedia.org/wiki/Administrative_divisions_of_Luxembourg)
Moldova	Administrative-Territorial Units from the Left Bank of the Dniester	It's missing from the exposure model
North Macedonia	Oslomej, Vraneshtica, Zajas	Regions have been merged with Kichevo Municipality into one administrative unit in 2013

Middle East

Country	Area	Problem
Azerbaijan	Shaki	Shaki City is not part of Shaki District but a separate entity
Azerbaijan	Yevlakh	Yevlakh City is not part of Yevlakh District but a separate entity
Azerbaijan	Lankaran	Lankaran City is not part of Lankaran District but a separate entity
Lebanon	Baalbek-Hermel Governorate, Akkar Governorate	Separated from the North Governorate in 2003
Oman	Musandam Governorate	Entity is missing
Oman	Al Buraymi	Separated from Ad Dhahirah in 2006
Oman	Al Batinah south/north	Al batinah split in 2013
Oman	Ash Sharqiyah south/north	Ash Sharqiyah split in 2011

Oceania

Country	Area	Problem
Australia	Other territories	Should be subdivided into: Jervis Bay Territory; Ashmore and Cartier Islands; Coral Sea Islands; Heard Island and McDonald Islands; Norfolk Island
Cook Islands	Nassau	Is actually part of Pukapuka

South America

Country	Area	Problem
Falkland Islands	All	Missing from model

South Asia

Country	Area	Problem
Bangladesh	Mymensingh Division	Separated from Dhaka Division in 2015
Maldives	All	Missing from the exposure model

Southeast Asia

Country	Area	Problem
Indonesia	Kalimantan Utala (North Kalimantan)	Split from Kalimantan Timur in 2012

Other:

Greenland is also missing entirely

[CatalinaYepes]

Hi Laurens,

Thank you for submitting your detailed report. We greatly appreciate your contribution!
Rest assured that we are actively working to address the issues you raised in upcoming updates. However, it is worth noting that we may not be able to address all issues in a short timeframe. We will work on better ways to communicate any limitations or discrepancies related to region names to the users.

We have added a brief description of the administrative regions used in our models to the FAQ section. Please keep in mind that the exposure models in this repo have been developed based on available country/territory information, which may differ from the current regions. Tracking the current list of administrative divisions and names even at level 1 is unfortunately not a trivial task. For instance, Indonesia now has four new provinces (Central Papua, Highland Papua, South Papua, and Southwest Papua) that are not even covered by GADM4.1 or OSM. Such changes happen on a frequent basis globally, but eventually we hope to establish a release cadence where we bring all models up to date to a certain year.

Once again, thank you for your valuable input.

[LaurensJN]

Thanks a lot for your comment! It's a great dataset, I hope I can make my contribution with this.

There are actually many more problems, such as the Indonesia one you mentioned. I only addressed issues that are already up-to-date in OSM and GADM. I assumed you use GADM4.1 as reference boundary dataset and as all the other problems also exist there, I sent the list of remaining problems to them instead of to you. If you'd like to, I can also post that here, for your own documentation?

By the way, the four new regions of Indonesia are actual in OpenStreetMap, just by there non-translated names. If you look at the name:en, you see that these are in fact the correction regions: Papua Barat Daya, Papua Tengah, Papua Pegunungan, Papua Selatan.

As a suggestion from my side, maybe it is better to use the ISO3166-2 code as reference, instead of the values that there are now (although sometimes the code is already in use). These are also easy to search for, even to look if they are still up-to-date or not. In the GADM ADM1 files, they all have the code HASC_1.

In any case, OpenStreetMap also has all the regions here. If you are interested, I made a mapping with all OSM IDs and your boundary names/codes. I find that OSM has the most accurate boundaries.