I’m working on warehouses location in US.
According to the CBP Census 2019, there are 1081 warehouses (NAICS 493—) in New-York CSA. OSM finds 912 building=warehouse or industrial=warehouse, which is OK.
It’s pretty the same for Chicago CSA.
But in the case of Los Angeles / Riverside CSA, CBP Census says there are 1 267 establishments, whereas OSM finds more than 18 000 entities !
I’m wondering why there is such a difference in this city. Is it because of a wrong automatic detection system ? A specific tagging rule for LA ? A too much optimist contributor ? A lack in the CBP Census ?
Good question and hint.
I would suggest go out to compare those units, jus to get to the right senses.
And count the units afterwards again, when they are marked or categorised under the same rules or descriptions.
I don’t know what the situation is elsewhere, but a lot of commercial buildings in California are “tilt ups”. See: https://en.wikipedia.org/wiki/Tilt_up At least a bunch of older ones are, earthquake codes have resulted in different construction techniques in recent years. I don’t know how one can tell from aerial imagery if a tilt up building was designed and used for warehouse operations, light manufacturing or office space. Often it is not even obvious from the ground. For example, I once worked for a small company that had the offices, manufacturing and warehousing all in one tilt up building.
So I can see where either OSM or CDP Census could have a strong bias one way or the other.
Not sure about the Riverside area, but there was a large building import for LA about 5 years ago (can’t remember if for the whole county or just the city at the moment). The identification of those buildings was, if I recall correctly, based on a county dataset. That dataset may have been biased too.
Hi, it all comes down to the maintenance of the high-quality standard of Open Street Map (OSM), based on the modus operandi (MO) ‘go out and survey’. If you’re not able to verify the use or original purpose (fi a church building) don’t ad any extra tags and stick to building=yes.
OSM is not a global mapper with lots of false facts or entries, due to uncontrolled and without maintenance data imports.
I would think the warehouse category from the census refers only to very large fulfillment centres. For instance there are around 2500 warehouses greater than 9000 m2 (I presume 100k sq ft) in the England & Wales serving a population of ~60 million which compares reasonably with 1000ish in NYC with a pop of 20-23 millions (MSA or CSA).
In many cities there are lots of old C19 and early C20 buildings which were built as warehouses and have long been re-used for other things, I think there’s even an area of downtown Denver known as the “Warehouse District”, see https://digital.denverlibrary.org/digital/collection/p15330coll22/id/64183/. This is also true across the western world: Copenhagen has many old dockside warehouses converted to other uses: office, hotels etc. The warehouses at Albert Dock in Liverpool house a range of things from museums, apartments and various eateries & bars.
If you want to evaluate the LA buildings I would take a random sample of perhaps 1% and look at them on aerial imagery & Bing Streetside imagery to get an impression of the type of building. This may alert you to discrepancies, or merely confirm that these data are accurate.
Thank you all for your answers, it helped me a lot.
Indeed, the LA Building Import @kartonage mentioned (https://wiki.openstreetmap.org/wiki/Los_Angeles,_California/Buildings_Import) says that 17 936 "“Warehousing, Distribution, Storage” buildings were imported as “building=warehouse”, including 8 396 under 10 000 square feet.
After an aerial imagery verification, it seems that a lot of shops having a storage space are tagged as warehouses.
I can’t find a way to select and remove this kind of buildings (very few tags building:use=warehouse).
How to map and code the complexity of reality, such a hard question !