Correcting inland water features

classic Classic list List threaded Threaded
16 messages Options
Reply | Threaded
Open this post in threaded view
|

Correcting inland water features

Andrew Wiseman-2
Hello,

My name is Andrew, I work for Apple’s Maps team. We are interested in doing some fixes and improvements to inland water features in Australia on OSM, such as adding and improving geometry of polygons for lakes and wide rivers, fixing broken relations, and correcting alignment issues for inland features when they meet the coast. 

We have a GitHub page here with more information about the project: https://github.com/osmlab/appledata/issues/85 

Please let me know if you have any suggestions or questions.

Thank you,

Andrew

Apple, Inc.

Andrew Wiseman |
  Maps | [hidden email]

APPLE CONFIDENTIAL
This message (including attachments if any) is for the private use of the addressee only and may contain confidential or privileged information. If you have received this message by mistake please notify the sender by return e-mail and delete this message and any attachments from your system. Any unauthorized use or dissemination of this message, and any attachments in whole or in part is strictly prohibited.





_______________________________________________
Talk-au mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-au
Reply | Threaded
Open this post in threaded view
|

Re: Correcting inland water features

Warin
On 16/05/18 03:05, Andrew Wiseman wrote:

> Hello,
>
> My name is Andrew, I work for Apple’s Maps team. We are interested in
> doing some fixes and improvements to inland water features in
> Australia on OSM, such as adding and improving geometry of polygons
> for lakes and wide rivers, fixing broken relations, and correcting
> alignment issues for inland features when they meet the coast.
>
> We have a GitHub page here with more information about the project:
> https://github.com/osmlab/appledata/issues/85
>
> Please let me know if you have any suggestions or questions.

Many inland waters in Australia are 'intermittent' meaning they only
flow when there is rain and that rain may only occur every 5 years or so
on average.

Some waterways are seasonal ... mostly in the north for the wet season
around Xmas time.

The LPI Base Map for the state of NSW is a fairly good source for
geometry of waterways and it is usable within OSM.

--------------
I have changed some of the inland waterways to intermittent but there
are others that I have left, these I lack knowledge of to tag them
intermittent.
The base map now differentiates the intermittent waters so they do show
up and are therefore easy to identify and correct any tagging.

_______________________________________________
Talk-au mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-au
Reply | Threaded
Open this post in threaded view
|

Re: Correcting inland water features

Ewen Hill
In reply to this post by Andrew Wiseman-2
Andrew,
   Welcome. As Warin mentioned, the vast majority of waterways are
intermittent outside the great dividing range and coastal areas and a map,
source and accuracy unknow is available as an example


https://amp.reddit.com/r/AussieMaps/comments/7sr4nz/where_the_water_flows_permanently_in_australia/?__twitter_impression=true

It would be interesting to see some small test areas and how you declare a
stream. Does it only run with extreme downpours or is it a collector.

There are also a number of areas I think in SA that have been innindated by
streams from a data source and these could possibly be trimmed back. Flood
water on flat deserts will make its own mind up next time. I would be really
keen to hear more and thank you for your time in improvung the map.

Ewen




--
Sent from: http://gis.19327.n8.nabble.com/Australia-f5416966.html

_______________________________________________
Talk-au mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-au
Reply | Threaded
Open this post in threaded view
|

Re: Correcting inland water features

Andrew Harvey-3
In reply to this post by Andrew Wiseman-2
Agreed with Warin and Ewen.

I'd like to point out the AGRI imagery [1], yes it's old (2006-2011), black and white, and missing parts of the imagery due to cloud and seams, but it may help as an extra source to compare. It's good enough to make out rivers and it should be accurate positionally since it was validated with ground surveys. CC BY 4.0 and usable in OSM from the blanket Geoscience Australia wavier[5].

GA also has a Surface Hydrology dataset [2] [3] [4] which could be used as well.

Some features are better tagged as a wetland https://wiki.openstreetmap.org/wiki/Tag:natural%3Dwetland. Local knowledge is always prefered, but so long as features mapped from local knowledge are not impacted, something mapped from other sources is better than nothing, which can be improved further by local knowledge.

I'm interested to know more about the alignment issues when they meet the coast, is there a specific example of where this needs work at the moment in OSM?



_______________________________________________
Talk-au mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-au
Reply | Threaded
Open this post in threaded view
|

Re: Correcting inland water features

Andrew Wiseman-2
Thanks for the feedback and suggestions everyone, that’s very helpful. We have also added AGRI to our list of sources.

Where the inland water meets the coastlines, we have seen a few different styles of modeling so I wanted to see if you had a local policy about it. 

In some cases we’ve seen things like this, where the coastline goes very far upstream: https://www.openstreetmap.org/way/14763476#map=11/-12.4279/132.4683 

In other cases, the coastline is farther downstream, like where the Pine River stops here: https://www.openstreetmap.org/way/159813122#map=14/-27.2842/153.0761

We were thinking the second example is better. If that’s the case for the Alligator River we would put the coastline here (which is also where the GA Surface Hydrology polygon has the river starting) https://www.openstreetmap.org/?mlat=-12.2296501875&mlon=132.402678198#map=15/-12.2293/132.4083 and change the existing coastline to riverbank.

Please let me know what you think.

Thanks,

Andrew

Andrew Wiseman |
  Maps | [hidden email]

APPLE CONFIDENTIAL
This message (including attachments if any) is for the private use of the addressee only and may contain confidential or privileged information. If you have received this message by mistake please notify the sender by return e-mail and delete this message and any attachments from your system. Any unauthorized use or dissemination of this message, and any attachments in whole or in part is strictly prohibited.




On May 16, 2018, at 9:26 AM, Andrew Harvey <[hidden email]> wrote:

Agreed with Warin and Ewen.

I'd like to point out the AGRI imagery [1], yes it's old (2006-2011), black and white, and missing parts of the imagery due to cloud and seams, but it may help as an extra source to compare. It's good enough to make out rivers and it should be accurate positionally since it was validated with ground surveys. CC BY 4.0 and usable in OSM from the blanket Geoscience Australia wavier[5].

GA also has a Surface Hydrology dataset [2] [3] [4] which could be used as well.

Some features are better tagged as a wetland https://wiki.openstreetmap.org/wiki/Tag:natural%3Dwetland. Local knowledge is always prefered, but so long as features mapped from local knowledge are not impacted, something mapped from other sources is better than nothing, which can be improved further by local knowledge.

I'm interested to know more about the alignment issues when they meet the coast, is there a specific example of where this needs work at the moment in OSM?




_______________________________________________
Talk-au mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-au
Reply | Threaded
Open this post in threaded view
|

Re: Correcting inland water features

Phil Wyatt

Hi folks,

 

Bay or river closing lines are defined here

 

http://www.ga.gov.au/scientific-topics/marine/jurisdiction/maritime-boundary-definitions

 

There is also the GA Coastline which may help

 

http://services.ga.gov.au/site_11/rest/services/Global_Map_Project_Australian_Base_Map/MapServer/6

 

 

Cheers - Phil

 

 

From: Andrew Wiseman [mailto:[hidden email]]
Sent: Friday, May 18, 2018 10:07 AM
To: OSM Australian Talk List
Subject: Re: [talk-au] Correcting inland water features

 

Thanks for the feedback and suggestions everyone, that’s very helpful. We have also added AGRI to our list of sources.

 

Where the inland water meets the coastlines, we have seen a few different styles of modeling so I wanted to see if you had a local policy about it. 

 

In some cases we’ve seen things like this, where the coastline goes very far upstream: https://www.openstreetmap.org/way/14763476#map=11/-12.4279/132.4683 

 

In other cases, the coastline is farther downstream, like where the Pine River stops here: https://www.openstreetmap.org/way/159813122#map=14/-27.2842/153.0761

 

We were thinking the second example is better. If that’s the case for the Alligator River we would put the coastline here (which is also where the GA Surface Hydrology polygon has the river starting) https://www.openstreetmap.org/?mlat=-12.2296501875&mlon=132.402678198#map=15/-12.2293/132.4083 and change the existing coastline to riverbank.

 

Please let me know what you think.

 

Thanks,

 

Andrew


Andrew Wiseman |
  Maps | [hidden email]

 

APPLE CONFIDENTIAL

This message (including attachments if any) is for the private use of the addressee only and may contain confidential or privileged information. If you have received this message by mistake please notify the sender by return e-mail and delete this message and any attachments from your system. Any unauthorized use or dissemination of this message, and any attachments in whole or in part is strictly prohibited.

 

 



On May 16, 2018, at 9:26 AM, Andrew Harvey <[hidden email]> wrote:

 

Agreed with Warin and Ewen.

 

I'd like to point out the AGRI imagery [1], yes it's old (2006-2011), black and white, and missing parts of the imagery due to cloud and seams, but it may help as an extra source to compare. It's good enough to make out rivers and it should be accurate positionally since it was validated with ground surveys. CC BY 4.0 and usable in OSM from the blanket Geoscience Australia wavier[5].

 

GA also has a Surface Hydrology dataset [2] [3] [4] which could be used as well.

 

Some features are better tagged as a wetland https://wiki.openstreetmap.org/wiki/Tag:natural%3Dwetland. Local knowledge is always prefered, but so long as features mapped from local knowledge are not impacted, something mapped from other sources is better than nothing, which can be improved further by local knowledge.

 

I'm interested to know more about the alignment issues when they meet the coast, is there a specific example of where this needs work at the moment in OSM?

 

 

 


_______________________________________________
Talk-au mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-au
Reply | Threaded
Open this post in threaded view
|

Re: Correcting inland water features

Andrew Davidson-3
In reply to this post by Warin
On 16/05/18 07:35, Warin wrote:
> Many inland waters in Australia are 'intermittent' meaning they only
> flow when there is rain and that rain may only occur every 5 years or so
> on average.

The vast majority of Australian stream are non-perennial. The WSJ has a
cool slider that shows this:

http://graphics.wsj.com/documents/wsj_sliders14/AustralianStreams

There are a least two problems for Australian mapping:

1. There is no OSM tag for ephemeral streams. On a stream length basis
the majority of Australian streams are ephemeral.

2. There is no OSM tag for the end of a watercourse. This causes
problems for validators because they can't tell if a stream finishes
because the mapper gave up or the stream itself just peters out.

_______________________________________________
Talk-au mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-au
Reply | Threaded
Open this post in threaded view
|

Re: Correcting inland water features

Andrew Davidson-3
In reply to this post by Ewen Hill
On 16/05/18 08:51, Ewen Hill wrote:
> There are also a number of areas I think in SA that have been inundated by
> streams from a data source and these could possibly be trimmed back. Flood
> water on flat deserts will make its own mind up next time. I would be really
> keen to hear more and thank you for your time in improving the map.

My personal view is that there is not much point in mapping unnamed
streams. By only bothering with the named streams you are in effect
applying something like the Wikipedia notability test.

I know the bit of SA you are referring to and it is a bit of a mess
because only the SA part of the streams have been imported and they all
get cut off at the NT border.

_______________________________________________
Talk-au mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-au
Reply | Threaded
Open this post in threaded view
|

Re: Correcting inland water features

Andrew Davidson-3
In reply to this post by Andrew Harvey-3
On 16/05/18 23:26, Andrew Harvey wrote:
>
> GA also has a Surface Hydrology dataset [2] [3] [4] which could be used
> as well.

+1

This is a good source as it covers all of the country and in effect has
the datasets from the states we already have permission from. It's also
useful for getting the name of the stream and they flag streams as being
perennial or not.

The downside is that the data comes from a range of mapping some of
which includes the 1:250,000 NatMap series, so the streams can be quite
inaccurate in places.

_______________________________________________
Talk-au mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-au
Reply | Threaded
Open this post in threaded view
|

Re: Correcting inland water features

Warin
In reply to this post by Andrew Davidson-3
On 19/05/18 11:02, Andrew Davidson wrote:

> On 16/05/18 07:35, Warin wrote:
>> Many inland waters in Australia are 'intermittent' meaning they only
>> flow when there is rain and that rain may only occur every 5 years or
>> so on average.
>
> The vast majority of Australian stream are non-perennial. The WSJ has
> a cool slider that shows this:
>
> http://graphics.wsj.com/documents/wsj_sliders14/AustralianStreams
>
> There are a least two problems for Australian mapping:
>
> 1. There is no OSM tag for ephemeral streams. On a stream length basis
> the majority of Australian streams are ephemeral.

Some are using stream=ephemeral ... low usage.
I'd rather go with ephemeral=yes as that then can be used on 'lakes' and
other things.

>
> 2. There is no OSM tag for the end of a watercourse. This causes
> problems for validators because they can't tell if a stream finishes
> because the mapper gave up or the stream itself just peters out.

I did strike one stream that flowed in to a river .. at both ends!! That
had me puzzled. I tracked it back using topo information to find it went
over a saddle, so I mapped it as flowing away from the saddle on both
sides ..

As for indicating an 'end' ... maybe exit=underground on the last node ???


_______________________________________________
Talk-au mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-au
Reply | Threaded
Open this post in threaded view
|

Re: Correcting inland water features

Ian Sergeant-2
On 19 May 2018 at 11:34, Warin <[hidden email]> wrote:

> Some are using stream=ephemeral ... low usage.
> I'd rather go with ephemeral=yes as that then can be used on 'lakes' and
> other things.

flow=ephemeral, maybe.  water-presence=ephemeral?

Ian.

_______________________________________________
Talk-au mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-au
Reply | Threaded
Open this post in threaded view
|

Re: Correcting inland water features

Warin
On 19/05/18 11:38, Ian Sergeant wrote:
> On 19 May 2018 at 11:34, Warin <[hidden email]> wrote:
>
>> Some are using stream=ephemeral ... low usage.
>> I'd rather go with ephemeral=yes as that then can be used on 'lakes' and
>> other things.
> flow=ephemeral, maybe.  water-presence=ephemeral?
>
> Ian.
>

'ephemeral=yes' matches the present use of 'intermittent=yes'. I like at
least some consistency in the tagging.


_______________________________________________
Talk-au mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-au
Reply | Threaded
Open this post in threaded view
|

Re: Correcting inland water features

Ian Sergeant-2
 On 19/05/18 11:38, Ian Sergeant wrote:
> flow=ephemeral, maybe.  water-presence=ephemeral?

On 19 May 2018 at 11:44, Warin <[hidden email]> wrote:

> 'ephemeral=yes' matches the present use of 'intermittent=yes'. I like at
> least some consistency in the tagging.

I think you picked the wrong mapping project :-)

The issue here of course, is that the next tag will something=yes.
Ugliness.  Best to fix it now.

Ian.

_______________________________________________
Talk-au mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-au
Reply | Threaded
Open this post in threaded view
|

Re: Correcting inland water features

Warin
I have started a draft for ephemeral.

https://wiki.openstreetmap.org/wiki/Proposed_features/ephemeral

On 19/05/18 12:08, Ian Sergeant wrote:

>   On 19/05/18 11:38, Ian Sergeant wrote:
>> flow=ephemeral, maybe.  water-presence=ephemeral?
> On 19 May 2018 at 11:44, Warin <[hidden email]> wrote:
>
>> 'ephemeral=yes' matches the present use of 'intermittent=yes'. I like at
>> least some consistency in the tagging.
> I think you picked the wrong mapping project :-)

Yep. landuse=grass ... stupid.

>
> The issue here of course, is that the next tag will something=yes.
> Ugliness.  Best to fix it now.

The properties keys are all this way. So I follow the crowd, as convincing the crowd is probably required.
Changing all of them would be a mammoth task. Changing one tag .. landuse=grass ... is just a little less of a challenge.


_______________________________________________
Talk-au mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-au
Reply | Threaded
Open this post in threaded view
|

Re: Correcting inland water features

cleary

I am struggling with how to use "ephemeral" rather than the "intermittent" tag, particularly when it comes to on-the-ground verification.

I have travelled in some rural and outback areas in western NSW and Queensland and, to a lesser extent, in South Australia.  I also regularly look at satellite imagery. If I see a waterway or lake without water, I am comfortable to add the intermittent tag. But I don't see every waterway and lake often enough to know how often the water flows. And satellite imagery is just a snapshot at one moment, perhaps several years ago. How can one expect to verify the frequency or duration of water flow? Another concern is that the total area of a lake may be covered with water only occasionally but half the area may be covered in water much more of the time. Do we try to work out which part of the lake is ephemeral and which is intermittent?

I understand why it may be useful to map water areas differently if they are not always filled with water. But trying to pictorially represent the dimension of time on a two-dimensional map and somehow show multiple degrees of absence of water, has me puzzled.

I don't think we would try to map roads differently according to whether they carry traffic all the time, some  of the time, or  just occasionally.

I will work with whatever the community agrees, but I need help to understand how to verify ephemeral as distinct from intermittent when mapping waterways and lakes.


 


On Mon, May 21, 2018, at 11:10 AM, Warin wrote:

> I have started a draft for ephemeral.
>
> https://wiki.openstreetmap.org/wiki/Proposed_features/ephemeral
>
> On 19/05/18 12:08, Ian Sergeant wrote:
>
> >   On 19/05/18 11:38, Ian Sergeant wrote:
> >> flow=ephemeral, maybe.  water-presence=ephemeral?
> > On 19 May 2018 at 11:44, Warin <[hidden email]> wrote:
> >
> >> 'ephemeral=yes' matches the present use of 'intermittent=yes'. I like at
> >> least some consistency in the tagging.
> > I think you picked the wrong mapping project :-)
>
> Yep. landuse=grass ... stupid.
>
> >
> > The issue here of course, is that the next tag will something=yes.
> > Ugliness.  Best to fix it now.
>
> The properties keys are all this way. So I follow the crowd, as
> convincing the crowd is probably required.
> Changing all of them would be a mammoth task. Changing one tag ..
> landuse=grass ... is just a little less of a challenge.
>
>
> _______________________________________________
> Talk-au mailing list
> [hidden email]
> https://lists.openstreetmap.org/listinfo/talk-au

_______________________________________________
Talk-au mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-au
Reply | Threaded
Open this post in threaded view
|

Re: Correcting inland water features

Warin
On 21/05/18 19:16, cleary wrote:

> I am struggling with how to use "ephemeral" rather than the "intermittent" tag, particularly when it comes to on-the-ground verification.

Good point. However ...
If the Todd River is ephemeral then by extrapolation other waterways in the surrounding area would be too.
Knowing the weather patterns where rain fall is nothing for years and then a few years worth within some days would suggest that
these areas have ephemeral water ways too. So, how is you knowledge of rainfall patterns? :)

>
> I have travelled in some rural and outback areas in western NSW and Queensland and, to a lesser extent, in South Australia.  I also regularly look at satellite imagery. If I see a waterway or lake without water, I am comfortable to add the intermittent tag. But I don't see every waterway and lake often enough to know how often the water flows. And satellite imagery is just a snapshot at one moment, perhaps several years ago. How can one expect to verify the frequency or duration of water flow? Another concern is that the total area of a lake may be covered with water only occasionally but half the area may be covered in water much more of the time. Do we try to work out which part of the lake is ephemeral and which is intermittent?

When Lake Eye has some water in it .. where it is will depend on which way the wind is blowing .. it is that flat due to the salt layer.
However Lake Eyre would be intermittent .. as there is water in it or more than 5 weeks .. when it is full,
so it does not comply with the ephemeral 'majority of the time' .. at least for me.

>
> I understand why it may be useful to map water areas differently if they are not always filled with water. But trying to pictorially represent the dimension of time on a two-dimensional map and somehow show multiple degrees of absence of water, has me puzzled.
>
> I don't think we would try to map roads differently according to whether they carry traffic all the time, some  of the time, or  just occasionally.
>
> I will work with whatever the community agrees, but I need help to understand how to verify ephemeral as distinct from intermittent when mapping waterways and lakes.

That help? I'll add it to the ephemeral page.

A side issue ... the meaning of 'intermittent' is poorly defined on the OSM wiki .. and some have taken it to be so similar to seasonal they are applying both tags.
  I think seasonal is not rendered and intermittent is .. and that could be motivation for some to apply the intermittent tags to the seasonal waterways...

>
>  
>
>
> On Mon, May 21, 2018, at 11:10 AM, Warin wrote:
>> I have started a draft for ephemeral.
>>
>> https://wiki.openstreetmap.org/wiki/Proposed_features/ephemeral
>>
>> On 19/05/18 12:08, Ian Sergeant wrote:
>>
>>>    On 19/05/18 11:38, Ian Sergeant wrote:
>>>> flow=ephemeral, maybe.  water-presence=ephemeral?
>>> On 19 May 2018 at 11:44, Warin <[hidden email]> wrote:
>>>
>>>> 'ephemeral=yes' matches the present use of 'intermittent=yes'. I like at
>>>> least some consistency in the tagging.
>>> I think you picked the wrong mapping project :-)
>> Yep. landuse=grass ... stupid.
>>
>>> The issue here of course, is that the next tag will something=yes.
>>> Ugliness.  Best to fix it now.
>> The properties keys are all this way. So I follow the crowd, as
>> convincing the crowd is probably required.
>> Changing all of them would be a mammoth task. Changing one tag ..
>> landuse=grass ... is just a little less of a challenge.
>>
>>
>> _______________________________________________
>> Talk-au mailing list
>> [hidden email]
>> https://lists.openstreetmap.org/listinfo/talk-au
> _______________________________________________
> Talk-au mailing list
> [hidden email]
> https://lists.openstreetmap.org/listinfo/talk-au
>

_______________________________________________
Talk-au mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-au