Parc National de la Comoé is so beautiful that Openstreetmap currently records it twice:
At first glance, the relation is more sophisticated - more detail, more complicated modelling… Surely it is better. But no… Parc National de la Comoé’s boundaries are legally defined by “Décret n° 2018-497 du 23 mai 2018 portant redéfinition des limites du Parc National de la Comoé” :
“Le présent décret redéfinit les limites du Parc national de la Comoé, en application de la loi la n° 2002-102 du 11 janvier 2002, telle que modifiée par la loi n° 2013-864 du 23 décembre 2013. Toutes les dispositions antérieures contraires au présent décret sont abrogées, notamment le décret n°68-81 du 9 février 1968 portant création du Parc national de la Comoé”
In summary, it defines the boundaries - and supersedes any previous definition.
It is particularly helpful, in supplying the nodes of the boundary polygon - here are the two pages of the decree in full:
I extracted them using Claude Opus 4.8 High, and I got the following GeoJSON - sorry for the 253 lines block, but I want every piece of this debate to be contained in this conversation:
{
"type": "FeatureCollection",
"features": [
{
"type": "Feature",
"properties": {
"name": "Parc national de la Comoe",
"decret": "n 2018-497 du 23 mai 2018",
"superficie_ha_declaree": 1148756,
"crs_source": "WGS84 geographic"
},
"geometry": {
"type": "Polygon",
"coordinates": [
[
[
-3.92102621,
8.61482255
],
[
-3.934894,
8.63767
],
[
-3.96969,
8.621069
],
[
-3.994925,
8.643363
],
[
-3.9772101,
8.6406878
],
[
-3.98265312,
8.6750878
],
[
-3.99811214,
8.66998328
],
[
-3.9977156,
8.75322383
],
[
-3.97613799,
8.74709138
],
[
-3.93790151,
8.84395223
],
[
-3.96618532,
8.8755769
],
[
-4.02152767,
8.86295828
],
[
-4.04158976,
8.89394961
],
[
-4.03219088,
8.93971782
],
[
-4.12129423,
9.08164744
],
[
-4.24851138,
9.11794238
],
[
-4.24921794,
9.03342889
],
[
-4.34741974,
9.04243735
],
[
-4.35179024,
9.15839444
],
[
-4.38458871,
9.14138636
],
[
-4.42110041,
9.16427563
],
[
-4.39916323,
9.30898538
],
[
-4.36093287,
9.33889841
],
[
-4.33480452,
9.32783895
],
[
-4.33702687,
9.40033194
],
[
-4.37630183,
9.47457327
],
[
-4.31332379,
9.5801773
],
[
-4.27282443,
9.60609377
],
[
-4.20027611,
9.61572953
],
[
-4.14546384,
9.57016592
],
[
-4.11183579,
9.56240061
],
[
-4.06387521,
9.59298608
],
[
-4.01747979,
9.5807186
],
[
-3.96368703,
9.58469459
],
[
-3.75229319,
9.61100977
],
[
-3.70036226,
9.60812873
],
[
-3.6946398,
9.58746772
],
[
-3.6306244,
9.56111775
],
[
-3.62082057,
9.58502066
],
[
-3.53203793,
9.59257503
],
[
-3.43247708,
9.473231
],
[
-3.24612246,
9.34340232
],
[
-3.22455625,
9.2154269
],
[
-3.11592353,
9.16877234
],
[
-3.195497,
8.68347
],
[
-3.21769075,
8.61448081
],
[
-3.28028303,
8.58326364
],
[
-3.34000496,
8.60420622
],
[
-3.39671665,
8.59130096
],
[
-3.43085985,
8.54857622
],
[
-3.49503255,
8.54178585
],
[
-3.54054234,
8.50485635
],
[
-3.63247827,
8.58649351
],
[
-3.66793155,
8.61778185
],
[
-3.75169804,
8.63141355
],
[
-3.77909953,
8.65513358
],
[
-3.85439181,
8.64581687
],
[
-3.92102621,
8.61482255
]
]
]
}
}
]
}
None of the two current records of Parc National de la Comoé match the extracted data:
- Relation: Comoé National Park (9450443) differs in many ways (haha, pun !) - clearly not faithful to the legal definition.
- Way: Parc National de la Comoé (796523031) is reasonably close to the legal definition in overall shape, but contains 174 nodes - whereas the legal definition only takes 57 nodes to represent… The difference is a stretch of river that the current way follows - which is wrong.
So, the fate of Relation: Comoé National Park (9450443) is sealed: I’ll copy all its excellent tags to the correct way.
As for Way: Parc National de la Comoé (796523031), two options:
- Delete it & replace it with the imported legal definition GeoJSON, with all the excellent tags from the deleted relation… Full fidelity to the source.
- Correct it - only removing the section that wrongly the river, and add to it the excellent tags from the deleted relation… Almost as good, and extremely close - but preserves history, which is good Openstreetmap practice.
So, I’ll chose one of those options ? I’m thinking out loud here - I needed to write to clarify my own thoughts, and also document my process in case someone wonders why large objects are suddenly deleted in northern Ivory Coast.
No so fast - I am having second thoughts… Look at this screenshot - ESRI World orbital imagery, with Relation: Comoé National Park (9450443) on the left and the legally defined boundary on the right… Relation: Comoé National Park (9450443) follows a clear boundary between two types of vegetation - typical of denser vegetation inside a protected area, and cultivated land outside of it:
So, while Relation: Comoé National Park (9450443) doesn’t follow the legal definition, it does map at least some facts on the ground… The plot thickens.
So, yes - I need your input before I commit one way or the other.
