The need for better geohealth information infrastructure

In the early hours of the morning a North Brunswick warehouse caught fire and burned. The fire was less than three miles from our new GeoHealth Lab at the Center for Remote Sensing and Spatial Analysis, so it caught our attention.

The smoke from the fire could be seen from Manhattan and from space and appeared as a large plume on weather radar with trains and traffic delayed throughout the region. The warehouse stored goods reportedly ranging from plastics to automotive parts to hydrogen cyanide, so this could have been a terrible situation. As of this afternoon, the EPA has declared that there is 'no environmental hazard' from the fire, but I don't know if that means there was not a health hazard. (It isn't clear and nothing in this blog post is meant to suggest either way)

With the situation appearing to be under control, we can now look at this as an opportunity to imagine the next generation of geospatial health information and how it might change our response to a situation like this.

As an experiment, I watched the situation unfold and looked for corresponding spatial information, of which there was little. The morning news did show weather radar showing the plume of smoke, which is a digital map. But that was about all I found.

Imagine going to the NJ Department of Health and getting live data that helps you assess the situation. Maybe they post a map of where the fire is or show you the plume of smoke so that you can avoid exposure. Maybe you can help them by marking your location and whether you are experiencing any breathing problems or smell burning plastic. By the time I write this, the cloud has dissipated and reconstructing the event is suddenly much harder without timely evidence from the field.

To simply make a map of the plume (above), I had to look at NOAA radar and other images on news sites to estimate where it was.  A reliable plume map, made at regular time intervals over the course of the fire would help professionals assessing the health impacts of the fire. (please note that the data shown is for illustration purposes and should not be used for assessments. Much of this plume was fairly high and at other times it stretched out into the Atlantic - so a simple polygon isn't enough)

Imagine a homeowner near the fire. After living close to the warehouse for 30 years, they still don't know how close they are. A simple map like the one above, showing buffers of 300 yards and 500 yards from the building, could change a simple decision about whether to stay or go. I chose the distances fairly arbitrarily, but a health professional could help explain what happens at specific distances and what the level of risk in each area might be.

Mapping the plume would allow a comparison with population centers. For instance, the above map shows population density in the areas around the plume. We can quickly pick out Jamesburg and Englishtown as places to check. Freehold is nearby, as well. 

Counting up the people impacted is tricky. We need to figure out whether we are using the right plume map. And we need to understand what we mean by "impacted" as well as which population dataset we use to estimate that number. But if we settle for a very simple figure, the grey polygon in the map can be estimated to cover an area populated by more than 80,000 residents.

As a public safety issue, we might ask where are there emergency treatment facilities nearby. But as a geohealth issue, this might not be the right question.

Instead, we might make ask about pulmonologists in the area. Digging around on the Internet can get you some information about these doctors. But it would take very good data to figure out whether these doctors have the capacity for a sudden increase in cases from such a crisis.

With a little time we could have done much more.  We could have compared radar images to map different levels of exposure. We could have added land use/land cover to show where forests might counter the health impacts. We might have been able to find other datasets, like building footprints and cadastral details, which could reveal new details. We could have looked at demographic characteristics to ask whether specific groups (elderly, poor, minorities) were impacted more than others. Maybe we could combine census and land use and parcels to model where people work and where they work at home. Maybe we would discover that information about the event needs to be distributed in more than one language (Hungarian? Hindi? Spanish?). We could have certainly added some context to the maps along with scale bars and other mandatory basics.

For GeoHealth, the situation demonstrates how a geospatial information infrastructure could really help with unanticipated events like this. As a new field, we are just beginning to understand what is essential and what is peripheral. And these maps (which were made quickly, without scale bars or good data or a background in pulmonology) are meant to simply point out that we need to think about the health ramifications of an event like this from a spatial perspective. Health is a spatial phenomenon that we can see stretched across our landscapes if we learn how to look for it.

For three decades the Center for Remote Sensing and Spatial Analysis we have created a wide variety of spatial data and advised agencies on ways to use geospatial data for scientific and professional applications. One of our primary interests at the GeoHealth Lab is to extend that track-record of success to encourage the development of spatial information that will make our communities healthier. 

---

UPDATE: One of our readers pointed out that adding a map of asthma clusters would help us illustrate the potential impact on at-risk populations. That would be something else that a health/wellness spatial data infrastructure would include. As a proxy, I thought that age might let us isolate either infants and toddlers or elderly. For the purposes of rapid assessment I offer you instead of map of median age by block group which still lets us almost instantly identify one of the largest retirement communities in central New Jersey at Jamesburg. To find it, just look for the dark cluster near the middle of the map.