Geographic Information Systems Stack Exchange is a question and answer site for cartographers, geographers and GIS professionals. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

INASAFE seems like an interesting QGIS plugin but as I wade through the docs, I really need some sample data to play with in order for me to understand. There is a tutorial which uses Indonesia data, but I can't find that dataset either.

If anyone is using this and has any example datasets, it would be very appreciated.

OK, after working thru the INASAFE tutorial it was unclear how to use Census data, so I waded thru the python scripts. It became clear that INASAFE couldn't deal with my data. I'm not familiar enough with either QGIS or python to add my capability to INASAFE so I decided to try my luck doing a similar analysis using QGIS (newbie here, but a bold one).

INASAFE appears to work well with OSM building data and Indonesia has done a terrific job mapping what appears to be thousands of their buildings in OSM, but my county has very few and census data has much more detailed data.

So I took my 2010 census block data, which had fields of POP10 for # of people in block and HOUSING10 for # of housing units in block and added a calculated AREA field. Since I had everything in Google Mercator, this area is in some nonsense units, but I don't care about that. As long as cutting the polygon in half gives an area half as big, then it will work for my purposes.

Next I did an intersect with some circa 2002 FEMA 100-yr floodplanes (remember when FEMA floodplanes were free!). On the intersect layer I added a calculated AREA field called FloodAREA.

I calculate the flooded Population as POP10*FloodAREA/AREA (ratio of area of flooded polygon to original polygon). As expected, some times FloodAREA=AREA (completely flooded block), and sometimes was less (partially flooded block), but never more (sanity check).

I ran basic statistics on the intersect layer and got the surprising result that 45,000 people (10% of pop) in my county live in 100 year flood plane.

I was wondering about thresholding - if polygon has like 0.1 people flooded, probably shouldn't count it - people and houses are not uniformly distributed in block. Especially in flood plane, I suspect lowest spots are less likely to have houses built on them. So, if a block has 250 people, for example, and it shows 25 people flooded - this really is saying that lowest 10% of block is flooded. Housing density rarely seems to approach 50%, so it's very possible (likely?) that when 10% of a block is flooded, there are not 10% of the houses in the flooded area.

So 2 questions:

  1. Are there any ideas or standards out there for discarding partial people from the sum or requiring minimum % of block flooded before counting?
  2. How do I get a sum of a field in QGIS with a thresholding formula applied (like sum(if Flooded_Pop>1 and FloodAREA/AREA>0.05)?
share|improve this question

closed as too broad by PolyGeo Oct 1 '15 at 0:29

There are either too many possible answers, or good answers would be too long for this format. Please add details to narrow the answer set or to isolate an issue that can be answered in a few paragraphs.If this question can be reworded to fit the rules in the help center, please edit the question.

OK, I'm working on it but I haven't gotten any smarter. I have Census data for my county, which is shapefiles for each Census block with fields of HOUSING10 for # of housing units in the block and POP10 for number of inhabitants. Since nothing in the INASAFE keyword descriptions seems to allow specifying a specific field, I have to assume that a raster is needed with pixel values. Since population is allowed a datatype of count, I don't understand how that works. Anyone know more about this? - by this I mean is there a way to use the shapefiles with count for exposure? – Dennis Conklin Dec 25 '12 at 21:14
Please open a new thread for new questions. Unlike a forum, this Q&A site is built around the idea of only one question per thread. – underdark Dec 27 '12 at 18:28
If it's Lucas county in Ohio, NOAA shows 9% are within the 100 yr flood plain: (general link, not county-specific) – mkennedy Dec 28 '12 at 17:58
That is not available for Summit County Ohio, where I am. Using the 2002 FEMA floodplane I got about 10% of the people in the floodplane, but I got the 2009 FEMA floodplane and they are significantly different. Now I have about 2.5% of the people in the floodplane. And this is way too high. Thanks to modern zoning ordances, when I examine the high population blocks in the floodplane, they have little or no houses in them. So I'm a little flustered about how to do anything else than manually count houses in each block and enter them manually - with 2500 blocks, that's not very appealing. – Dennis Conklin Dec 29 '12 at 17:55
up vote 6 down vote accepted

The INASAFE data used in the tutorial can be found at the inasafe_data Github repository. You might want to play with it first so you can get acquainted with the data types. Once you're confident enough, you can the use the equivalent US data. Good luck :)

share|improve this answer
Thanks for that pointer - I'm not a developer so I'm not good at Github – Dennis Conklin Dec 25 '12 at 11:34
Just download the zip version. Just make sure to use a download manager though. It's a bit large ~139 MB – R.K. Dec 25 '12 at 11:41

Apologies for writing a reply a year later, I am currently managing the InaSAFE work in Indonesia and hope I can shed some light onto your situation. When we built InaSAFE we made sure it worked with 2 types of exposure data: buildings (vector) and population count(raster). We are about to release a new version that will now allow for roads (lines). In the next version we hope to have polygons. The reason why we have left the development of polygons until last is that we need to figure out some way to logically split the polygon by the hazard and produce a tangible/actionable answer for a disaster manager to use ie. percentage of crops distorted, km2 of forest burnt etc.

However population as a polygon is additionally tricky as we have to assume that the density within the polygon is the same where ever you are. Hence if the county has 30% forest covering, InaSAFE has to assume that the coverage is the same - people living in forest. Its our challenge to work out how to ensure the count of people is at least similar to reality (no one living on the vent of a volcano or on the bank of a river).

For the training datasets they are now located at Cheers

share|improve this answer
I have one example Inasafe exercises in Indonesian language.. Might provide insight into the application. You can down load the link… – syafrauf Feb 8 '13 at 20:16

Not the answer you're looking for? Browse other questions tagged or ask your own question.