Solar Panels Quarterly Project: 39 days to go

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

Solar Panels Quarterly Project: 39 days to go

sk53.osm
Thought I'd write a really quick summary of progress on the quarterly project:
  • As of last night we had a total of 67,475 installations (solar farms enclosing several groups of panels are treated as one installation. At the time of writing another 500 have been added this morning.
  • 8 local authorities have more than 50% of the estimated total number (based on FIT data), in descending order : Nottingham, Ashfield, Liverpool, Knowsley, Tameside, Wrecsam, Bassetlaw, Peterborough, and Manfield.
  • During August anywhere from 1500 to nearly 2500 panels have been mapped each day. Even if only average 1000 a day for the remainder of the project it means we should comfortably exceed 100k mapped installations (between 10 & 12% of the total).
  • There's no shortage of places where it is easy to add a lot of panels in a short time (I added nearly 500 in Worksop yesterday). Gregory's site has a list of candidates at the bottom of the main page.
  • As expected urban areas are easier to do than rural areas.
  • General areas with a lot of mapping are: West Midlands, North-East (Tyneside, Wearside etc), North-West, East Mids and Kent. I think you can guess who the likely suspects are.
  • I don't have any immediate stats on solar farms, but Dan pointed out that we have over 50%. perhaps Dan or Jex can provide an update.
Apart from a general target of 100k installations some other things are worth focussing on for the remainder of the project:
  • Getting a few LAs over 75%. Ashfield is surprising because installations are widely distributed. Liverpool is more typical: two-thirds are mapped but perhaps 80% of the LSOAs have not been touched, and over a 1000 are in 4 LSOAs around Speke. So finding the additional panels may be less rewarding.
  • Searching a small number of rural LAs intensively: small ones are probably best: Anglesey, Isle of Wight, Rutland etc. My suspicion is that panels are harder to find, but also that imagery is often quite a bit older.
  • Get more done in Scotland (and Northern Ireland). Gregory's site doesn't allow the micro-targeting by LSOA which has been so effective for England & Wales.
The first two are because there is the possibility of using located rooftop solar panels as training sets for more automated identification using machine learning. Dan may want to say more on this.

Lastly, a couple of remarks stemming from jumping around England & Wales about OSM mapping in general:
  • Buildings have been diligently mapped in all sorts of unexpected places. Unfortunately address data is scarcer.
  • Former council estates often lack many of the footpaths which give a better sense of their layout. Particularly true for the post-war Radburn style estates.
  • Some places were obviously mapped when OS Streetview first became available and the road networks could do with tidying up (I'll draw up a separate list at some stage).
  • MS StreetSide is really valuable particularly around Manchester. Although the imagery is 7 years old it allows a much more detailed appreciation of building types, and checking of other detail..
Happy hunting,

Jerry

PS. I was only the 3rd OSMer to look to add the new solar plant at Aldershot station!

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

Re: Solar Panels Quarterly Project: 39 days to go

Dan S
Hi Jerry and all,

Indeed we've over 500 solar farms - I can check the exact number this weekend. (From the OfGEM data, I estimate that approx 1100 exist.)

Here are some comments from the point of view of a data consumer:

As I think Gregory commented - we don'tt really "need" much detail on solar farms (e.g. the panels contained within them). In general, we have metadata for solar farms, so we just need their outlines, plus either their output capacity and/or their ID in the REPD dataset (repd:id=*).

Much more important is to spot the small-scale solar - that's a vital piece of the puzzle that none of the official data sources are very good at.

As Jerry suggests, it's good to aim for high coverage on a few chosen regions (e.g. local authorities). This will help with the machine learning but the more direct importance is that it will help piloting solar energy forecasting, using those regions.
In particular, it might be helpful for at least someone to focus on Cornwall. That's an area with a lot of solar installations, and the National Grid know it as an area that often gives a lot of "reverse flow", i.e. a large amount of small-scale generation feeding back into the grid. I've had a bit of a look at Cornwall and various imagery looks really nice and clear there.

Have a sunny weekend!

Cheers
Dan



Op vr 23 aug. 2019 om 13:05 schreef SK53 <[hidden email]>:
Thought I'd write a really quick summary of progress on the quarterly project:
  • As of last night we had a total of 67,475 installations (solar farms enclosing several groups of panels are treated as one installation. At the time of writing another 500 have been added this morning.
  • 8 local authorities have more than 50% of the estimated total number (based on FIT data), in descending order : Nottingham, Ashfield, Liverpool, Knowsley, Tameside, Wrecsam, Bassetlaw, Peterborough, and Manfield.
  • During August anywhere from 1500 to nearly 2500 panels have been mapped each day. Even if only average 1000 a day for the remainder of the project it means we should comfortably exceed 100k mapped installations (between 10 & 12% of the total).
  • There's no shortage of places where it is easy to add a lot of panels in a short time (I added nearly 500 in Worksop yesterday). Gregory's site has a list of candidates at the bottom of the main page.
  • As expected urban areas are easier to do than rural areas.
  • General areas with a lot of mapping are: West Midlands, North-East (Tyneside, Wearside etc), North-West, East Mids and Kent. I think you can guess who the likely suspects are.
  • I don't have any immediate stats on solar farms, but Dan pointed out that we have over 50%. perhaps Dan or Jex can provide an update.
Apart from a general target of 100k installations some other things are worth focussing on for the remainder of the project:
  • Getting a few LAs over 75%. Ashfield is surprising because installations are widely distributed. Liverpool is more typical: two-thirds are mapped but perhaps 80% of the LSOAs have not been touched, and over a 1000 are in 4 LSOAs around Speke. So finding the additional panels may be less rewarding.
  • Searching a small number of rural LAs intensively: small ones are probably best: Anglesey, Isle of Wight, Rutland etc. My suspicion is that panels are harder to find, but also that imagery is often quite a bit older.
  • Get more done in Scotland (and Northern Ireland). Gregory's site doesn't allow the micro-targeting by LSOA which has been so effective for England & Wales.
The first two are because there is the possibility of using located rooftop solar panels as training sets for more automated identification using machine learning. Dan may want to say more on this.

Lastly, a couple of remarks stemming from jumping around England & Wales about OSM mapping in general:
  • Buildings have been diligently mapped in all sorts of unexpected places. Unfortunately address data is scarcer.
  • Former council estates often lack many of the footpaths which give a better sense of their layout. Particularly true for the post-war Radburn style estates.
  • Some places were obviously mapped when OS Streetview first became available and the road networks could do with tidying up (I'll draw up a separate list at some stage).
  • MS StreetSide is really valuable particularly around Manchester. Although the imagery is 7 years old it allows a much more detailed appreciation of building types, and checking of other detail..
Happy hunting,

Jerry

PS. I was only the 3rd OSMer to look to add the new solar plant at Aldershot station!
_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb

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

Re: Solar Panels Quarterly Project: 39 days to go

ael
On Fri, Aug 23, 2019 at 04:33:17PM +0100, Dan S wrote:
> In particular, it might be helpful for at least someone to focus on
> *Cornwall*. That's an area with a lot of solar installations, and the
> National Grid know it as an area that often gives a lot of "reverse flow",
> i.e. a large amount of small-scale generation feeding back into the grid.
> I've had a bit of a look at Cornwall and various imagery looks really nice
> and clear there.

I am often in East Cornwall, and have mapped the odd Solar Farm when out
surveying - and also wind farms.

But I haven't had time to do much armchair imagery mapping. There are
maybe around 3 fairly active mappers in East Cornwall.

ael


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

Re: Solar Panels Quarterly Project: 39 days to go

Brian Prangle-2
In reply to this post by Dan S
Stupendous though the progress is - we're unlikely to achieve more than about 20% of the total by the end of the project so perhaps it would be better to concentrate for the remainder of the time on the larger solar farms where we already have approx 50% of the total and have reasonable prospect of getting them all. It would be good to  have a quartlery project where we achieve completion. Perhaps someone with the data skills could compare the REPD ofgem data with OSM data and prepare a map (or even just a csv with lat/lon for use in JOSM) showing the missing ones?

Regards

Brian

On Fri, 23 Aug 2019 at 16:34, Dan S <[hidden email]> wrote:
Hi Jerry and all,

Indeed we've over 500 solar farms - I can check the exact number this weekend. (From the OfGEM data, I estimate that approx 1100 exist.)

Here are some comments from the point of view of a data consumer:

As I think Gregory commented - we don'tt really "need" much detail on solar farms (e.g. the panels contained within them). In general, we have metadata for solar farms, so we just need their outlines, plus either their output capacity and/or their ID in the REPD dataset (repd:id=*).

Much more important is to spot the small-scale solar - that's a vital piece of the puzzle that none of the official data sources are very good at.

As Jerry suggests, it's good to aim for high coverage on a few chosen regions (e.g. local authorities). This will help with the machine learning but the more direct importance is that it will help piloting solar energy forecasting, using those regions.
In particular, it might be helpful for at least someone to focus on Cornwall. That's an area with a lot of solar installations, and the National Grid know it as an area that often gives a lot of "reverse flow", i.e. a large amount of small-scale generation feeding back into the grid. I've had a bit of a look at Cornwall and various imagery looks really nice and clear there.

Have a sunny weekend!

Cheers
Dan



Op vr 23 aug. 2019 om 13:05 schreef SK53 <[hidden email]>:
Thought I'd write a really quick summary of progress on the quarterly project:
  • As of last night we had a total of 67,475 installations (solar farms enclosing several groups of panels are treated as one installation. At the time of writing another 500 have been added this morning.
  • 8 local authorities have more than 50% of the estimated total number (based on FIT data), in descending order : Nottingham, Ashfield, Liverpool, Knowsley, Tameside, Wrecsam, Bassetlaw, Peterborough, and Manfield.
  • During August anywhere from 1500 to nearly 2500 panels have been mapped each day. Even if only average 1000 a day for the remainder of the project it means we should comfortably exceed 100k mapped installations (between 10 & 12% of the total).
  • There's no shortage of places where it is easy to add a lot of panels in a short time (I added nearly 500 in Worksop yesterday). Gregory's site has a list of candidates at the bottom of the main page.
  • As expected urban areas are easier to do than rural areas.
  • General areas with a lot of mapping are: West Midlands, North-East (Tyneside, Wearside etc), North-West, East Mids and Kent. I think you can guess who the likely suspects are.
  • I don't have any immediate stats on solar farms, but Dan pointed out that we have over 50%. perhaps Dan or Jex can provide an update.
Apart from a general target of 100k installations some other things are worth focussing on for the remainder of the project:
  • Getting a few LAs over 75%. Ashfield is surprising because installations are widely distributed. Liverpool is more typical: two-thirds are mapped but perhaps 80% of the LSOAs have not been touched, and over a 1000 are in 4 LSOAs around Speke. So finding the additional panels may be less rewarding.
  • Searching a small number of rural LAs intensively: small ones are probably best: Anglesey, Isle of Wight, Rutland etc. My suspicion is that panels are harder to find, but also that imagery is often quite a bit older.
  • Get more done in Scotland (and Northern Ireland). Gregory's site doesn't allow the micro-targeting by LSOA which has been so effective for England & Wales.
The first two are because there is the possibility of using located rooftop solar panels as training sets for more automated identification using machine learning. Dan may want to say more on this.

Lastly, a couple of remarks stemming from jumping around England & Wales about OSM mapping in general:
  • Buildings have been diligently mapped in all sorts of unexpected places. Unfortunately address data is scarcer.
  • Former council estates often lack many of the footpaths which give a better sense of their layout. Particularly true for the post-war Radburn style estates.
  • Some places were obviously mapped when OS Streetview first became available and the road networks could do with tidying up (I'll draw up a separate list at some stage).
  • MS StreetSide is really valuable particularly around Manchester. Although the imagery is 7 years old it allows a much more detailed appreciation of building types, and checking of other detail..
Happy hunting,

Jerry

PS. I was only the 3rd OSMer to look to add the new solar plant at Aldershot station!
_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb
_______________________________________________
Talk-GB mailing list
[hidden email]
https://lists.openstreetmap.org/listinfo/talk-gb

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

Re: Solar Panels Quarterly Project: 39 days to go

Dan S
Hi Brian,

I see where you're coming from. But I'd argue it's not that vital to
push people forward on the larger solar farms. I'm confident they'll
all get done, by the end of the year at least, since there's a finite
set of them and they're highly visible. (We've gone from 138 to 538
solar farms tagged so far.) To my eyes, the quarterly project is
already on track to be a big success - we started with 4k panels and
now we have 69k.

But of course if people would like to map the outlines of the big
farms then please do - indeed I've been doing some of that. I do
actually have a CSV of the type you suggest, but it's a mess with
personal notes. I can share it but not immediately (I was hoping to
merge it back into the wiki list at [1]). In my experience, about 5%
of the REPD entries, I don't see in any aerial imagery (so cannot map)
and I don't know whether that means they're very recent or that they
didn't get built. I've checked all the entries of 10 MW or more, so if
others would like to help then a good starting point is items in the
list [1] of less than 10 MW (e.g. between 5 and 10 MW - there are lots
of those).

Best
Dan



[1] https://wiki.openstreetmap.org/wiki/Renewable_energy_in_the_United_Kingdom#List_of_under_construction_and_operational_UK_Ground_Mounted_Solar_Farms

Op za 24 aug. 2019 om 10:24 schreef Brian Prangle <[hidden email]>:

>
> Stupendous though the progress is - we're unlikely to achieve more than about 20% of the total by the end of the project so perhaps it would be better to concentrate for the remainder of the time on the larger solar farms where we already have approx 50% of the total and have reasonable prospect of getting them all. It would be good to  have a quartlery project where we achieve completion. Perhaps someone with the data skills could compare the REPD ofgem data with OSM data and prepare a map (or even just a csv with lat/lon for use in JOSM) showing the missing ones?
>
> Regards
>
> Brian
>
> On Fri, 23 Aug 2019 at 16:34, Dan S <[hidden email]> wrote:
>>
>> Hi Jerry and all,
>>
>> Indeed we've over 500 solar farms - I can check the exact number this weekend. (From the OfGEM data, I estimate that approx 1100 exist.)
>>
>> Here are some comments from the point of view of a data consumer:
>>
>> As I think Gregory commented - we don'tt really "need" much detail on solar farms (e.g. the panels contained within them). In general, we have metadata for solar farms, so we just need their outlines, plus either their output capacity and/or their ID in the REPD dataset (repd:id=*).
>>
>> Much more important is to spot the small-scale solar - that's a vital piece of the puzzle that none of the official data sources are very good at.
>>
>> As Jerry suggests, it's good to aim for high coverage on a few chosen regions (e.g. local authorities). This will help with the machine learning but the more direct importance is that it will help piloting solar energy forecasting, using those regions.
>> In particular, it might be helpful for at least someone to focus on Cornwall. That's an area with a lot of solar installations, and the National Grid know it as an area that often gives a lot of "reverse flow", i.e. a large amount of small-scale generation feeding back into the grid. I've had a bit of a look at Cornwall and various imagery looks really nice and clear there.
>>
>> Have a sunny weekend!
>>
>> Cheers
>> Dan
>>
>>
>>
>> Op vr 23 aug. 2019 om 13:05 schreef SK53 <[hidden email]>:
>>>
>>> Thought I'd write a really quick summary of progress on the quarterly project:
>>>
>>> As of last night we had a total of 67,475 installations (solar farms enclosing several groups of panels are treated as one installation. At the time of writing another 500 have been added this morning.
>>> 8 local authorities have more than 50% of the estimated total number (based on FIT data), in descending order : Nottingham, Ashfield, Liverpool, Knowsley, Tameside, Wrecsam, Bassetlaw, Peterborough, and Manfield.
>>> During August anywhere from 1500 to nearly 2500 panels have been mapped each day. Even if only average 1000 a day for the remainder of the project it means we should comfortably exceed 100k mapped installations (between 10 & 12% of the total).
>>> There's no shortage of places where it is easy to add a lot of panels in a short time (I added nearly 500 in Worksop yesterday). Gregory's site has a list of candidates at the bottom of the main page.
>>> As expected urban areas are easier to do than rural areas.
>>> General areas with a lot of mapping are: West Midlands, North-East (Tyneside, Wearside etc), North-West, East Mids and Kent. I think you can guess who the likely suspects are.
>>> I don't have any immediate stats on solar farms, but Dan pointed out that we have over 50%. perhaps Dan or Jex can provide an update.
>>>
>>> Apart from a general target of 100k installations some other things are worth focussing on for the remainder of the project:
>>>
>>> Getting a few LAs over 75%. Ashfield is surprising because installations are widely distributed. Liverpool is more typical: two-thirds are mapped but perhaps 80% of the LSOAs have not been touched, and over a 1000 are in 4 LSOAs around Speke. So finding the additional panels may be less rewarding.
>>> Searching a small number of rural LAs intensively: small ones are probably best: Anglesey, Isle of Wight, Rutland etc. My suspicion is that panels are harder to find, but also that imagery is often quite a bit older.
>>> Get more done in Scotland (and Northern Ireland). Gregory's site doesn't allow the micro-targeting by LSOA which has been so effective for England & Wales.
>>>
>>> The first two are because there is the possibility of using located rooftop solar panels as training sets for more automated identification using machine learning. Dan may want to say more on this.
>>>
>>> Lastly, a couple of remarks stemming from jumping around England & Wales about OSM mapping in general:
>>>
>>> Buildings have been diligently mapped in all sorts of unexpected places. Unfortunately address data is scarcer.
>>> Former council estates often lack many of the footpaths which give a better sense of their layout. Particularly true for the post-war Radburn style estates.
>>> Some places were obviously mapped when OS Streetview first became available and the road networks could do with tidying up (I'll draw up a separate list at some stage).
>>> MS StreetSide is really valuable particularly around Manchester. Although the imagery is 7 years old it allows a much more detailed appreciation of building types, and checking of other detail..
>>>
>>> Happy hunting,
>>>
>>> Jerry
>>>
>>> PS. I was only the 3rd OSMer to look to add the new solar plant at Aldershot station!
>>> _______________________________________________
>>> Talk-GB mailing list
>>> [hidden email]
>>> https://lists.openstreetmap.org/listinfo/talk-gb
>>
>> _______________________________________________
>> Talk-GB mailing list
>> [hidden email]
>> https://lists.openstreetmap.org/listinfo/talk-gb

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