ICTworks

Here is one of the basic questions for a mass Long Lasting Insecticidal Net distribution campaign, “How do you know when you are done?” Population data isn’t all that great and is often inflated at the local level. In the past the way this works was – estimates were made, mobilization took place for x number of days and when those days were complete – the campaign was deemed complete.

Essentially, you are blind.

During the pilot though we had an extra tool to help with this. Relatively high-resolution satellite imagery courtesy of Google Earth, which is free to everyone.

Register now for MERL Tech DC to find new GIS tools for your work

Using this we were able to see actual settlements. Try it yourself. Go to google maps, set it to satellite imagery, find Nigeria and zoom in. You’ll easily be able to see settlements and even individual houses.

But this still doesn’t solve our problem. We need two more things. First we need to know where we have been and superimpose that on the map. That is easy – see figure 1 below.

Human in the loop

The second task, though, is much harder. We can see with our eyes if there are settlements that haven’t been visited. There aren’t any dots there. If we were dealing with just a small geographic area, we could simply scan the map and tell people, “hey, go look at this settlement, seems like they haven’t been visited.” But you can’t do that at scale, considering we might be dealing with hundreds of thousands of settlements. We needed a way to automate.

Not so easy.

Automate

To understand how you might automate take a look at (image 1) In this image you see places that were visited (the little colorful circles) and if you look closely you’ll see little black polygons surrounding some settlements that were not visited. It’s a little hard to tell because of the resolution but those boxes are drawn around what looks like settlements.

So now we have something to work with. 1 – we know where we’ve been. 2 – we know areas that haven’t been visited indicated by the polygons. Using simple GIS analysis you can confirm whether or not one of the little circles appears within the polygons. If it is, it has now been mobilized.

We could theoretically only highlight the areas that have not been mobilized. Take a look at (image 2). In the lower portion you’ll see one of those polygons with little circles in it. These circles were not present in (image 1)

Not quite there yet

But we still have a problem. We had to draw those polygons.

From here we had two options – crowd source or get really lucky and find someone who has a file that has all of the settlements already identified with polygons around it.

First for crowd sourcing. We spoke to HOT|OSM and after a brief discussion they suggested the best way to do this was using an application called MapSwipe. Essentially, this is a crowd sourcing map application that you can download. You swipe from screen to screen indicating if you see something (in this case settlements) or not. The backend software then recombines all of that into your settlement files.

I was honestly really excited to try this method, and will on other projects but we got really lucky.

We got lucky

As it turns out through a grant from the Bill and Melinda Gates Foundation we already have very detailed settlement data for all of Nigeria. We were able to access this information and will be able to use this as the basis for automation. (See image 3)

Right now we are in the process of implementing this in the Long Lasting Insecticidal Net distribution program. Once this is implemented teams at the LGA level will be able to log into the dashboard and see the map of their area highlighting settlements that need to be visited. Very cool.

And, while we still may not get 100% coverage – at least we aren’t blind anymore.

By Nate Barthel covering his experience in a large LLIN distribution campaign in Nigeria for CRS.