Lake Victoria
| Lake Victoria | |
|---|---|
This page is localised for  Uganda - UG
| |
| Status: | Draft (under way) |
| Proposed by: | Bert Araali |
| Tagging: | water=lake |
| Applies to: | 
|
| Definition: | Improved ampping and tagging scheme for Lake Victoria |
| Drafted on: | 10 Jan 2021 |
| Voting members: | talk-ug+Core Members |
V・T・E Lake Victoria, East Africa, Africa  |
| latitude: -1, longitude: 33 |
| Browse map of Lake Victoria 1°00′00.00″ S, 33°00′00.00″ E |
| Edit map |
|
External links:
|
| Use this template for your locality |
Lake Victoria is a lake in East Africa, Africa at latitude 1°00′00.00″ South, longitude 33°00′00.00″ East.
 |
This page is a proposal and under development. The described tagging and mapping conventions/guidelines are based on existing approved and proposed tagging schemes, and used in East-Africa. New features might be added as new requirements evolve or the project evolves. This page is intended to help mappers and users (data consumers) in general in East Africa about a new concept for the mapping of Lake Victoria and references to other data sources. This is intended as a guide based on what contributors to the page think is best. There may be other equally acceptable alternatives - if so present them to the community after discussion on the Talk page. At least it is advised to consult the page administrator(s) Bert Araali (on osm, edits, contrib, heatmap, chngset com.). |
OSM legacy mapping
The lake was originally mapped as a single multi-polygon with reference: Relation 2606941
This worked fine until over the years more details were added. By the end of 2020 the multi-polygon has grown to 1650 members consisting of over 410 000 nodes. This resulted in timeouts when obtaining the history or members list online. On top of that the lake is located in an area with lots of HOT projects, with participants with low mapping experience. This over the years resulted several times in relation errors due to minor changes. Some of them where none closing ways, incorrect role definitions, overlapping new ways, sometimes even accidental deletion of the lake. The errors resulted in the lake disappearing from the map. Editing these large relations is a very complex and time consuming task, as the whole relation needs to be downloaded and edited and analysed with specialised editors or JOSM.
2020 was also a disastrous year as the lake water risen to record heights, resulting in lots of floods. When mapping the shores of the lake, users found large deviations (sometimes up to 20 to 50 meters inland, islands disappeared completely under water) from the original mapping, mostly based on alder satellite images from Bing. So users kept on shifting the shore lines inland, assuming they were mapping inaccuracies. The lake shores should be mapped as medium high water level lines, according to the OSM guidelines. Due to this constant shifting and the lack of free available other sources or data many these lines were lost.
So, in January 2021 I (Bert Araali) took the intuitive to work out a new mapping and tagging scheme. The following key constraints were applied to the new solution:
- Use only existing OSM tagging and mapping solutions. Prefer to use approved concepts. For special purposes, like the grouping, new proposed concepts might be used but the proposal should be at least 2 yeras old and used in multiple other instances for similar purposes (minimal use number should be around 50). Although OSM is tagged for data, not for rendering, prefer to use solutions which provide the most obvious rendering solutions in the standard OSM renderer (Carto). This will make the use more friendly for novice users.
- Chop the lake in pieces to support more convenient and flexible concurrent mapping. Chopping the lake into pieces will allow concurrent mapping without creating conflict during upload. Also the amount of data to be downloaded or uploaded for the changes will be reduced significantly. In case of errors, only a part of the lake will disappear from the rendered map. It will be easier and cheaper in data to resolve the errors and restore proper rendering. The use of the validation tool in JOSM will be faster and more convenient.
- Add high and low water boundaries. To support and provide data for new and innovative applications like environmental, hazard and navigation mapping. Make the concept of Median High Water Line shores more clear to novice users and reduce the need to adjust the shorelines to reflect momentarily satellite images.
Abstract
As described above there was a need to redesign the mapping and tagging of Lake Victoria.
Lake Victoria is one of the biggest inland water bodies in the world. Since it is a lake, not a sea or ocean, it's shores cannot be tagged with natural=coastline. Traditionally, 1 single multi-polygon is used to map a lake. However this solution becomes very hard to edit and maintain for these large bodies, over time as the detail level increases the online solutions and Normatim even fail to show the complete lake. Since it covers many countries, different communities were implementing different solutions to cope with these restrictions. There was a need for one coordinated and simple solution.
How to map
Lake components
The actual lake components - lake shore lines
The lake shoreline is a rich and complex environment at the edge of the lake including cliffs, beaches, harbors, headlands (also called points), bays, wetlands and islands.
Every component is a multi-polygon, which can be edited and easily maintained as a single OSM component. Therefore, the size of every component should not exceed 300 items. This include both outer and inner roles. The number of nodes should be kept limited. Nodes aligned in a straight line, with a deviation less then 3 m must be avoided at all times.
To determine the limits we apply to the individual components we did some calculations on the legacy lake multipolygon:
- Number of ways with role:outer making up the outer boundary of the lake: 376
- Total length of the outer boundary: 4506.694 km
- Number of nodes in this outer way: 236 152
Calculated results: one node every 0.019 km or every 19 m
- Number of ways with role:inner making up all the islands and islets: 1277
- Total length of the inner boundaries: 3638.085 km
- Number of nodes in the inner ways: 179 633
Calculated results: one node every 0.0202 or every 20 m
We should be able to do better. So the new boundaries are created by copying and offsetting the existing one and then reducing the number of nodes. An accuracy of non-alignment not more then 3 m should do the job. We increase the distance in nearly straight lines like large bays, we decrease the distance between nodes to accommodate strongly curved borders. We did a test on the first lake component outer boundary:
In the new concept, at the lake shores, you will find 3 lines:
- The actual lake shore: the actual rendered lake shore line, this is located at the
mean high water springs (also known as Mean High Water Level / MHWL) line at the edge of the lake. This is the same concept as with the coastline of seas (see natural=coastline and Coastline). Although not strictly required, the land should be located on the left, the water on the right, meaning all outer borders go clockwise, all inner boundaries go counter-clockwise.
This makes sense since the lake, due to it's size has some tidal effects. For the lake this is the historical mean average water level since the start of the observations in 1900. Currently this stands at
- The low water line: similar to the maritime Baseline, this line is also the outer border for administrative areas land_area boundaries. The historical register low level was registered on 1923 and stand at 1133.19 m (m.a.m.s.l)
- The high water line: there is no sea tagging alternative. A new key value for . The historical high level until 2020 was at May 12th 1964 and stands at 1136.10 m (m.a.m.s.l).
Data sources for the shore line (MHWL), the low water line () and the high water line
Data to compose these lines is readily available and free to reuse (in the public domain). The lake levels (gage levels) have been closely monitored and registered since 1896, with complete and reliable data available as from 1950 (see []. The record low water level appeared in 1923 at 1133 m (), the record high level, until 2020, in 1961 at (). This data however has some challenges:
- GPS reference and surveying: due to the lack of a geoid modal for GPS/levelling projects (until 2020 in Uganda the best geoid modal commonly used is EGS2008), the gage levels or water levels are hard to accurately convert to the WGS84 or any local reference system, we rely on the WGS84 system. See Geoid determination in Uganda: Current status. By 2020, a new UGRN (Uganda Geodetic Reference Network) was established (see Establishing an accurate geodetic reference network for Uganda and [https://www.ignfi.fr/en/portfolio-item/reseau-geodesique-ouganda/ UPDATING AND IMPLEMENTING OF GEODETIC REFERENCE FRAME FOR LAND ADMINISTRATION, UGANDA). The location of the new UGRN markers (426) and reference stations (12 CORS) is not yet mapped in OSM by the time of this writing (10/01/2021).
As recently, due to the effects of climate change and the increased rainfall in the area, the lake levels have risen, breaking new records year after year (2019 and 2020). This changes the high water lines inland. The mean lines, as a rule of thumb, should be located somewhere in the middle between the low and high water lines. Although the slope, and thus the distance from the mean, of the lake bottom and the land area may differ. Slopes on land can be easily determined from topographic maps, freely available. Lake bottom slopes are a different story as we couldn't find any recent nautical or navigation maps with depths near the shoreline. When referring to satellite pictures, the user should take into account that the lake has some tidals and the shorelines can differ significantly from year to year. So in general the mean shore line should not change so fast. Investigate carefully before you do so.
Terms and abbreviation
m.a.m.s.l. = meters above sea level, []
Lake group
The cluster should not contain more then 300 children.
Lake Victoria component list
| Nr | Name | Type | Reference | Description | Comments |
|---|---|---|---|---|---|
| 1 | Duwafu Bay of Lake Victoria | bay | Single large bay. East end point is Rubabu Point, a navigation landmark, north end point is Kyabasimba point in Uganda. | It is unclear if the Rubabu Point is located in Tanzania or Uganda. The original country border assigns it to Uganda, although most of the land area point is located in Tanzania.
In the legacy version, the lake shoreline had a length of 49.51 km and consisted of 1477 nodes, meaning one every 33 meters. In the new version the number of nodes for the lake shore was reduced to |