[Talk-GB] Incorrect use of OS VectorMap District when mapping?

[Michael Collinson]

At 13:29 09/03/2011, Chris Hill wrote:
>On 09/03/11 11:57, Michael Collinson wrote:
>>At 12:32 10/02/2011, Andy Robinson (blackadder-lists) wrote:
>>>Henry Gomersall [mailto:heng at cantab.net] wrote:
>>> >Sent: 10 February 2011 11:07 AM
>>> >To: Peter Miller
>>> >Cc: Talk GB
>>> >Subject: Re: [Talk-GB] Incorrect use of OS VectorMap District 
>>> when mapping?
>>> >
>>> >On Thu, 2011-02-10 at 10:30 +0000, Peter Miller wrote:
>>> >> On reflection possibly we should use river-bank as that has more
>>> >> information in it, but recommend that anyone importing does a 'bridge
>>> >> cleanup' at the same time.
>>> >
>>> >This is an area I'm actually really interested in (for rural rivers) and
>>>keen to
>>> >contribute. So far I've been put off by exactly this problem. Is a
>>>reasonable
>>> >approach to use the OS data for river edges and then fill in the gaps
>>>(bridges
>>> >etc) with OSM data?
>>>
>>>+1
>>>
>>>If the OS vector data is only assumed to be the banks and the additional
>>>data for flow direction, bridges and other features are added from
>>>survey/BING etc then we should end up with a very functional dataset.
>>
>>A late response to this thread, but a word of caution. Comparing 
>>Bing imagery recently for several Yorkshire rivers with folk's 
>>riverbanks derived from OS data indicates that very frequently  the 
>>OS are not tracing the riverbank as the dividing line between water 
>>(clear river channel) and land (grass, scrub) but the top of the 
>>riverbank or where the rough "verge" meets pasture land.
>A further word of caution: Bing and all other imagery only shows a 
>snapshot of the way things are, often many years ago, and in an 
>indeterminate state of water level. Some rivers have tidal 
>influences, some rivers have very different levels in flood or 
>drought. Sometimes where the rough "verge" meets pasture land is the 
>highest point the water reaches regularly, but still only occasionally.

Certainly both Chris' and Phillip's cautions are certainly true but 
I've paid particular attention to the River Wharfe mid-reaches, which 
I know very well and flows in a well-defined channel with high banks 
and  has not shifted markedly in the last 40 years. In places, it is 
almost twice as wide as it should be. Chris may be right in 
suggesting that the highest water mark is being mapped, but why map 
the 10 - 25-year flood event level rather than the natural bank line? 
I am tempted to think that automated software has been used which 
like PGS coastlines occasionally gets confused by nearby lineaments. 
I also recall comparing with digitised 25:000 maps (vintage 1900 - 
1960 surveying) and noticing that it correlates much more closely 
with Bing than StreetView. Needs more analysis but be aware!

Mike