Inaccurate Admin Levels / Maps in South Korea

#1

Several of the iNat “places” in South Korea appear to be inaccurate with errors in the way their administrative levels are depicted on the site. I assume this is due to incorrect borders supplied by GADM but hope that a solution can be found to increase the accuracy of maps on iNat.

One quick example is the Admin Level 2 location “Mokpo”.

On iNat: https://www.inaturalist.org/places/mokpo
The selected area does not include the city of Mokpo (which it should) and is instead centered on the island Aphaedo, which should be in the Admin Level 2 location “Sinan County”.
Google Maps Link
Wikipedia Link

Unfortunately, the city where I live (Uijeongbu) is also affected by this issue which means a decent-sized chunk of the observations I upload end up being misplaced and serve as a constant reminder of how the maps don’t reflect the actual administrative borders here.

Admin Level 2 Uijeongbu is a location within Admin Level 1 Gyeonggi that, for the most part, should look similar to this image from Google Maps: Google Maps Link

On iNat the majority of the northern half is assigned to the Admin Level 2 region Yangju, the northeastern corner is assigned to Admin Level 2 region Pocheon, and the northernmost portion of Admin Level 1 Seoul’s Admin Level 2 Dobong district is mistakenly assigned to Uijeongbu.

  1. Observation from inside what should be the Admin Level 2 area (Korea - Gyeonggi - Uijeongbu):
    https://www.inaturalist.org/observations/2468439
    The address listed on iNat from Google is “358-36 Yijeongbudong, Yijeongbusi, Gyeonggi-do, KR”. iNat includes it under Korea - Gyeonggi - Yangju.

  2. Similarly:
    https://www.inaturalist.org/observations/17149005
    The address listed on iNat from Google is “Nagyang-dong, Uijeongbu-si, Gyeonggi-do, South Korea”. iNat includes it under Korea - Gyeonggi - Pocheon.

  3. Observation included within Korea - Gyeonggi - Uijeongbu that should not be placed there:
    https://www.inaturalist.org/observations/14532111
    The address listed on iNat from Google is “Dobong-gu, Seoul, South Korea”. iNat includes it under Korea - Gyeonggi - Uijeongbu but it should be treated as Korea - Seoul - Dobong. In this case it’s not ‘just’ the Admin Level 2 but also the Admin Level 1 that is different than expected, presumably due to where GADM decided to place their boundaries.

I’ve posted on the Google Group twice about this issue. The first time it was suggested that I contact GADM myself but despite doing so I have never heard back from them. The second time was in this thread – iNat Google Group Link – but without receiving a reply after I clarified what was occurring. I hope that bringing it up here in a dedicated thread here will help generate some attention to the matter.

In all honesty, it’s a little embarrassing to encourage people in Korea to try out iNat knowing that the maps aren’t accurate. I still do recommend the site to everyone I can, but it would be great if our observations were mapped in the correct place.

4 Likes

#2

It looks like what iNat currently has is for Admin-Level 1 and Admin-Level 2 are:

Jeollanam-do
…Wando
…Gwangyang
…Boseong
…Gurye
…Gangjin
…Naju
…Jindo
…Yeongam
…Muan
…Sinan
…Hwasun
…Hampyeong
…Suncheon
…Jangheung
…Mokpo
…Jangseong
…Yeosu
…Gokseong
…Yeonggwang
…Goheung
…Haenam
…Damyang
Incheon
…Namdong
…Nam
…Dong
…Ganghwa
…Bupyeong
…Gyeyang
…Jung
…Seo
…Yeonsu
…Ongjin
Busan
…Busanjin
…Gangseo
…Seo
…Buk
…Dong
…Dongnae
…Geumjeong
…Gijang
…Haeundae
…Nam
…Saha
…Sasang
…Suyeong
…Yeongdo
…Yeonje
Chungcheongbuk-do
…Danyang
…Goesan
…Cheongju
…Boeun
…Eumseong
…Chungju
…Cheongwon
…Okcheon
…Jecheon
…Jeungpyeong
…Jincheon
…Yeongdong
Seoul
…Gangseo
…Geum-cheon
…Gangnam
…Yeongdeungpo
…Dong-daemun
…Mapo
…Nowon
…Yongsan
…Gangbuk
…Dongjak
…Eun-pyeong
…Gandong
…Dobong
…Gwanak
…Gwang-jin
…Jongro
…Jung
…Guro
…Jungnang
…Seocho
…Seodaemun
…Seongbuk
…Seongdong
…Songpa
…Yangcheon
Ulsan
…Nam
…Buk
…Dong
…Jung
…Ulju
Daejeon
…Daedeok
…Seo
…Dong
…Jung
…Yuseong
Daegu
…Suseong
…Seo
…Jung
…Dong
…Nam
…Dalseo
…Dalseong
…Buk
Gangwon-do
…Hongcheon
…Chuncheon
…Taebaek
…Wonju
…Yanggu
…Donghae
…Goseong
…Gangneung
…Hoengseong
…Samcheok
…Hwacheon
…Inje
…Jeongseon
…Pyeongchang
…Sokcho
…Cheorwon
…Yangyang
…Yeongwol
Jeju
…Jeju
…Seogwipo
Jeollabuk-do
…Imsil
…Jeongeup
…Gunsan
…Namwon
…Jangsu
…Jinan
…Muju
…Jeonju
…Sunchang
…Wanju
…Iksan
…Buan
…Gimje
…Gochang
Gwangju
…Dong
…Gwangsan
…Buk
Chungcheongnam-do
…Gyeryong
…Seosan
…Nonsan
…Dangjin
…Cheongyang
…Buyeo
…Geumsan
…Hongseong
…Seocheon
…Cheonan
…Boryeong
…Taean
…Gongju
…Asan
…Yesan
Gyeongsangnam-do
…Miryang
…Tongyeong
…Uiryeong
…Namhae
…Geochang
…Changnyeong
…Haman
…Gimhae
…Jinhae
…Hapcheon
…Masan
…Sancheong
…Yangsan
…Sacheon
…Hamyang
…Goseong
…Hadong
…Jinju
…Changwon
…Geoje
Gyeonggi-do
…Pyeongtaek
…Seongnam
…Siheung
…Yangju
…Yangpyeong
…Gwacheon
…Gunpo
…Hwaseong
…Icheon
…Guri
…Namyangju
…Uiwang
…Uijeongbu
…Gwangju
…Gwangmyeong
…Hanam
…Goyang
…Gimpo
…Osan
…Paju
…Pocheon
…Yeoju
…Yeoncheon
…Yongin
…Suwon
…Ansoeng
…Anyang
…Bucheon
…Dongducheon
…Gapyeong
…Ansan
Gyeongsangbuk-do
…Gunwi
…Yecheon
…Andong
…Bonghwa
…Cheongdo
…Cheongsong
…Goryeong
…Gumi
…Gyeongju
…Gimcheon
…Chilgok
…Pohang
…Sangju
…Yeongcheon
…Gyeongsan
…Seongju
…Uljin
…Yeongdeok
…Yeongyang
…Yeongju
…Uiseong
…Mungyeong
…Ulleung
Sejong
…Yeongi

Are the Provincial-level divisions and Municipal-level divisions described here Municipal-level divisions what you want for Admin-Level 1 and Admin-Level 2? If so, where does one get that data?

2 Likes

#3

Thanks for taking the time to explain the boundary problem, @whaichi. It looks like there are unfortunately many boundary and naming mistakes/generalizations in the GADM files for South Korea.

Can you confirm if this is a correct summary of the problems you noted?

  • In GADM/iNat, Uijeongbu (in Gyeonggi province) the boundary is grossly generalized so the center is correct but the borders are far from accurate.
    Compare:
    Google maps
    GADM

  • In GADM/iNat, the boundary for Mokpo (Jeollanam-do province) is completely incorrect and is instead showing an island part of neighboring Sinan. Is Mokpo incorrectly included in Muan?
    Google maps Mokpo
    Mokpo GADM
    Sinan GADM
    Muan GADM

The best way to identify other errors is probably to explore the GADM subdivisions for South Korea. It looks like many of the non-coastal boundaries are not very precise.

I did get a reply from GADM in January about a similar kind of problem with Bogota, Colombia. If you can help point us to a more correct authoritative source for provincial and administrative divisions, we can come up with a plan for making the corrections (and hopefully GADM can too).

3 Likes

#4

Thanks for taking a look into this, Scott. It’s much appreciated!

Jeollanam-do

The 22 Admin Level 2 places looks correct.

Incheon

The Admin Level 2 “Nam” (District) was renamed to “Michuhol” (District) in July 2018 but otherwise looks good.

Busan

The Admin Level 2 “Jung” (District) appears to be missing.

Chungcheongbuk-do

The 12 Admin Level 2 places look correct.

Seoul

The 25 Admin Level 2 places are correct but “Gandong” should be “Gangdong”. To the best of my knowledge, “Jongro” is typically transliterated as “Jongno” these days but that’s a smaller issue. I’m not sure why hyphens are used for some places (Geum-cheon, Dong-daemun, Eun-pyeong, Gwang-jin) and not others since Jung is the only place in the list that is only one syllable. For the sake of consistency, I would suggest removing hyphens from all of those places.

Ulsan

The 5 Admin Level 2 places look correct.

Daejeon

The 5 Admin Level 2 places look correct.

Daegu

The 8 Admin Level 2 places look correct.

Gangwon-do

The 18 Admin Level 2 places look correct.

Jeju

The 2 Admin Level 2 place looks correct.

Jeollabuk-do

The 14 Admin Level 2 places look correct.

Gwangju

“Dong”, “Gwangsan”, and “Buk” are correct but the list is also missing the Admin Level 2 places “Nam” and “Seo”.

Chungcheongnam-do

The 15 Admin Level 2 places look correct.

Gyeongsangnam-do

“Jinhae” and “Masan” were incorporated into Changwon in 2010 so should be combined with “Changwon” instead of being Admin Level 2 places in their own right.

Gyeonggi-do

“Ansoeng” should be “Anseong” but otherwise the 31 Admin Level 2 places look correct.

Gyeongsangbuk-do

The 23 Admin Level 2 places look correct.

Sejong

From what I can tell it looks like Sejong should have 19 Admin Level 2 places within it. The city’s official webpage - https://www.sejong.go.kr/eng/sub01_02_02.do - lists that many (near the bottom of the page). Sejong (Admin Level 1) city was only created in 2012 so I imagine the discrepancy is from the administrative divisions being redrawn since GADM’s last update.

Admin Level 1 looks good, I’ve added comments about the Admin Level 2 places. As for getting the data, I honestly don’t know what would be a good centralized and official location to receive the most up-to-date data. Things like the merger between Changwon, Masan, and Jinhae into “Changwon” and the creation of “Sejong” city was reported in the news here. The list on Wikipedia - https://en.wikipedia.org/wiki/Administrative_divisions_of_South_Korea#Provincial-level_divisions - looks accurate though I don’t know if iNaturalist would want to rely on it or how easily that information could be applied to any potential updates to Admin Level places.

2 Likes

#5

Yes, the summary about the boundary issues related to Uijeongbu and Mokpo is correct. I wasn’t sure how to access the GADM bounday maps myself but the one for Uijeongbu certainly explains how observations were being put in incorrect Admin Level 2 places. It looks like [Korea - Gyeonggi-do - Hanam] is another one with generalized borders that presents accuracy problems – especially as it borders an Admin Level 1 place (Seoul).

-Google maps
-GADM

I’m not sure where to find a great authoritative source regarding the boundaries - especially one with kd3 files available for import - but if I do find one I’ll be sure to share it here. I imagine the issue is made more complicated by Korean institutions often favoring local organizations such as Naver, Daum, and Kakao over global companies.

2 Likes

#6

Thanks for looking around for the borders, @whaichi. I think we/GADM can work with/convert different file formats such as shp, kml, and kmz (and maybe others), so the most important part is the authoritative source(s) of the boundaries as some kind of polygon data.

0 Likes