<html xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40"><head><meta http-equiv=Content-Type content="text/html; charset=utf-8"><meta name=Generator content="Microsoft Word 15 (filtered medium)"><style><!--
/* Font Definitions */
@font-face
{font-family:"Cambria Math";
panose-1:2 4 5 3 5 4 6 3 2 4;}
@font-face
{font-family:Calibri;
panose-1:2 15 5 2 2 2 4 3 2 4;}
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0cm;
font-size:11.0pt;
font-family:"Calibri",sans-serif;}
a:link, span.MsoHyperlink
{mso-style-priority:99;
color:blue;
text-decoration:underline;}
p.MsoNoSpacing, li.MsoNoSpacing, div.MsoNoSpacing
{mso-style-priority:1;
margin:0cm;
font-size:11.0pt;
font-family:"Calibri",sans-serif;
mso-fareast-language:EN-US;}
.MsoChpDefault
{mso-style-type:export-only;}
@page WordSection1
{size:612.0pt 792.0pt;
margin:70.85pt 70.85pt 70.85pt 70.85pt;}
div.WordSection1
{page:WordSection1;}
--></style></head><body lang=EN-GB link=blue vlink="#954F72" style='word-wrap:break-word'><div class=WordSection1><div><p class=MsoNoSpacing> There are many broken coastlines in the recent source data. Practically, all polygons crossing the antimeridian are broken, sometimes even not starting/ending on that meridian. The user side closing activity is not a simple exercise and it requires rerunning of a complex and complicated land/water area structuring program. Consequently, strange mismatch errors may appear on the World borders and large number of objects may simply disappear in the corresponding maps. Some examples (please, MK, do not repair them immediately):</p><p class=MsoNoSpacing>-Gap on the World border <span lang=NO-BOK><a href="https://osm.org/go/VBBVAUgp5-?layers=D&way=823391084"><span lang=EN-GB>https://osm.org/go/VBBVAUgp5-?layers=D&way=823391084</span></a></span> .</p><p class=MsoNoSpacing>-Broken polygon closing error at the World border <span lang=NO-BOK><a href="https://osm.org/go/FEAEFBGK7--?layers=YD"><span lang=EN-GB>https://osm.org/go/FEAEFBGK7--?layers=YD</span></a></span> . </p><p class=MsoNoSpacing>-What and how to connect dilemma on the World border <span lang=NO-BOK><a href="https://osm.org/go/~q6r665b--?layers=D"><span lang=EN-GB>https://osm.org/go/~q6r665b--?layers=D</span></a></span> .</p><p class=MsoNoSpacing>-Missing island <span lang=NO-BOK><a href="https://osm.org/go/b5j82T2f--?layers=HD&way=11491640"><span lang=EN-GB>https://osm.org/go/b5j82T2f--?layers=HD&way=11491640</span></a></span> , or a missing lake <span lang=NO-BOK><a href="https://osm.org/go/zBrvxUJK--?layers=D&way=718141147"><span lang=EN-GB>https://osm.org/go/zBrvxUJK--?layers=D&way=718141147</span></a></span> , or both <span lang=NO-BOK><a href="https://osm.org/go/ZfyUCvrZi-?layers=D&way=829792631"><span lang=EN-GB>https://osm.org/go/ZfyUCvrZi-?layers=D&way=829792631</span></a></span> . Other map makers have World border problems too: <span lang=NO-BOK><a href="https://goo.gl/maps/HiSXSfh4zSvNjKjz6"><span lang=EN-GB>https://goo.gl/maps/HiSXSfh4zSvNjKjz6</span></a></span><span class=MsoHyperlink><span lang=NO-BOK> </span></span>(misalignment, the famous white/light stripe, missing water…) or this one <span lang=NO-BOK><a href="https://binged.it/3qqi5rE"><span lang=EN-GB>https://binged.it/3qqi5rE</span></a></span> . To avoid the former issues, IMO we should make a fresh review of the coastline data under the control of OSMF and correct the errors. Also, we should detect all coastline holes, holes-in-holes… polygons and integrate them into other (water) layers where they belong. </p><p class=MsoNoSpacing> The tiled (…split-3857) coastline data is just an instance of a regular vector tiling, based on a square frame vector/geometry clipping. The strange tiling strategy, though formally correct, is risky and easily ends up with errors and difficulties in applications. Some notes:</p><p class=MsoNoSpacing>-The east and the west World borders do not meet at the antimeridian. The gap is around 5mm or close to 45 pixels in 200dpi. In the Large World this gap seams ignorable but it may generate accumulated errors. It should be avoided.</p><p class=MsoNoSpacing>-The tiles are too large and probably further tiling is required in applications. Note that users never need explicitly the trivial (TT) or empty tiles (i.e. the water tiles). Knowing the border/none-trivial land tiles (NT) the others are predefined. Also, the tile overlaps are unnecessary.</p><p class=MsoNoSpacing>-The multi-tiling technology is so much more effective compared to the usual vector pre-tiling technology that the latter may be considered as old fashion today. The multi-tiling opens for on-the-fly tiling service affordable even for individuals. For instance, the green tiles are the 10km NT tiles of the land polygons generated in seconds (on may laptop), with the size just slightly larger than the coastline, here</p><p class=MsoNoSpacing><span lang=NO-BOK><a href="https://drive.google.com/file/d/1UcRYG37Rmjb_4w6FavwGmQeclP-smmc2/view?usp=sharing"><span lang=EN-GB>https://drive.google.com/file/d/1UcRYG37Rmjb_4w6FavwGmQeclP-smmc2/view?usp=sharing</span></a> . Green </span><span class=MsoHyperlink><span style='color:windowtext;text-decoration:none'>tiles are the only necessary tiles/data to have on the server side. A</span></span>ny tiles of e.g. 1km size (like the yellow once) is clipped-out from a green tile in about 3 micro seconds, or it is outside of greens and is TT tile or empty. So, for the Planet, you can serve tiles on-the-fly for many data layers.</p><p class=MsoNoSpacing> Finally, some words about the lev9 simplified land polygons. Maps at this and lesser scales are mostly used for location overview/estimates of large objects. There are no details and as a rule hundreds of meters, or even kms, are compressed into just a several display pixels. Therefore, if you scale the 1:1 land areas to 1:1million scale and draw the borders with a darker line, you will see just a messy thick dark stripe and the data size is still exactly the same as in the 1:1 scale, like here</p><p class=MsoNoSpacing><span lang=NO-BOK><a href="https://drive.google.com/file/d/19OMuQaensHIZRsHPmiP32i0zB9NuhhSK/view?usp=sharing"><span lang=EN-GB>https://drive.google.com/file/d/19OMuQaensHIZRsHPmiP32i0zB9NuhhSK/view?usp=sharing</span></a> . </span>To avoid these negative side effects in (digital) cartography we usually run, after the down scaling, a process known as data generalisation (sometimes called simplification). Within this, the dark stripes are replaced with lines approximately running through the middle of the stripes (vector smoothing) and small and tiny spots are simply ignored (removed). There are many models doing the mentioned generalisation and production of scale levels. Most of them are heuristic and subjective criteria based. The best once radically reduce the data size and provide nice images hardly distinguishable from the image just scaled from 1:1. A good model should not produce spiky borders, brakes on solid objects or the contrary. Especially longish (not convex) objects make problems when thin sections alternate with thick (like on fjords, peninsulas, rivers…). To understand the differences let us look at a slightly enlarged extract from the former image overlapped with two simplified borders, here</p><p class=MsoNoSpacing><span lang=NO-BOK><a href="https://drive.google.com/file/d/1wOsth-PJbi2_ZM3EiE8dLb9SxVdqmCz1/view?usp=sharing">https://drive.google.com/file/d/1wOsth-PJbi2_ZM3EiE8dLb9SxVdqmCz1/view?usp=sharing</a> . </span>The black border is an extract from the “simplified-land-polygons-complete-3857” while the red border was generated by another simplification model that has a special longish object configuration handling in its vector smoothing. The latter, while following the green border, detects sharp U-turns (pointer by the blue arrow) and removes the tiny sections without destroying the topology connectivity. In this way the red border is even more reduced compared to the black (40%). When both border lines are rendered in the lev9 scale, or lower, it is hard to see any difference. The two lines almost perfectly overlap (red on top), like in the same extract here <a href="https://drive.google.com/file/d/1p2R0ZEkNeVKvOc1HJulTS31KkaVv6x0I/view?usp=sharing">https://drive.google.com/file/d/1p2R0ZEkNeVKvOc1HJulTS31KkaVv6x0I/view?usp=sharing</a> . Some more hints. If the simplified data is intended to be rendered by anti-aliasing, do not keep convex spots of several display pixels in diameter. Also, defragmentation should be done before simplification to avoid erroneous brakes like here on forests <span lang=NO-BOK><a href="https://osm.org/go/WS_z64-?layers=H"><span lang=EN-GB>https://osm.org/go/WS_z64-?layers=H</span></a></span> or on rivers <span lang=NO-BOK><a href="https://www.google.com/maps/@67.5253181,115.6199111,4z"><span lang=EN-GB>https://www.google.com/maps/@67.5253181,115.6199111,4z</span></a></span> or the same here</p><p class=MsoNoSpacing><span lang=NO-BOK><a href="https://drive.google.com/file/d/1_PQK4WBAVvNc-E3ewgn3DjUHwALmTQSZ/view?usp=sharing"><span lang=EN-GB>https://drive.google.com/file/d/1_PQK4WBAVvNc-E3ewgn3DjUHwALmTQSZ/view?usp=sharing</span></a></span> . </p><p class=MsoNoSpacing>The generalisation of object classes should be synchronized. It is both, aesthetic and formal error when large rivers disappear while small islands, lakes or titles are still rendered likes here <span lang=NO-BOK><a href="https://www.google.com/maps/@2.2836181,-52.031399,11.91z"><span lang=EN-GB style='color:windowtext;text-decoration:none'>https://www.google.com/maps/@2.2836181,-52.031399,11.91z</span></a></span> .</p><p class=MsoNoSpacing>Developers and researchers may find more details and hints in the paper here <span lang=NO-BOK><a href="https://drive.google.com/file/d/1Q6MSuWPcAtggn6Hx5paBa62IlsLOPadc/view?usp=sharing"><span lang=EN-GB>https://drive.google.com/file/d/1Q6MSuWPcAtggn6Hx5paBa62IlsLOPadc/view?usp=sharing</span></a></span><span lang=NO-BOK> </span></p><p class=MsoNoSpacing><o:p> </o:p></p></div><p class=MsoNormal><span class=MsoHyperlink><span style='color:windowtext;text-decoration:none'> Regards, Sandor</span></span></p><p class=MsoNormal><o:p> </o:p></p><p class=MsoNormal>Sent from <a href="https://go.microsoft.com/fwlink/?LinkId=550986">Mail</a> for Windows 10</p><p class=MsoNormal><o:p> </o:p></p></div></body></html>