1 of 34

Climate: Investigating impacts

Ryan McNeill

Deputy Editor, Date Journalism

Reuters

2 of 34

3 of 34

As the world was dealing with SARS-CoV-2, we were thinking about the next one. And how to stop it.

4 of 34

As the world was dealing with SARS-CoV-2, we were thinking about the next one. And how to stop it.

My colleague Deb Nelson was intrigued by a growing body of scientific research linking zoonotic spillovers to land use change.

5 of 34

As the world was dealing with SARS-CoV-2, we were thinking about the next one. And how to stop it.

My colleague Deb Nelson was intrigued by a growing body of scientific research linking zoonotic spillovers to land use change.

She started sending me academic papers. And I thought: I wonder if we could use past spillovers to identify other areas with similar conditions?

6 of 34

Other news organizations had covered the growing links between land use change.

7 of 34

Other news organizations had covered the growing links between land use change.

But it often lacked geographic specificity. Or the why. Or who was responsible.

8 of 34

Other news organizations had covered the growing links between land use change.

But it often lacked geographic specificity. Or the why. Or who was responsible.

Three glaring unanswered questions:

Where is risk highest?

How has risk changed?

Who is responsible?

9 of 34

10 of 34

We built a model

We identified 95 spillovers of bat-linked viruses between 2002 and 2020.
We divided nearly entire land surface of the earth into 25 sq km areas.
For each 25 sq km area, we assembled a dataset of covariates, most of them derived from satellites, for each year from 2002 to 2020.
Covariates included bat species richness, land surface temperatures, precipitation, vegetation indices, tree loss, land cover and estimates of pig and cattle populations. There were billions of observations.
The model first examined environmental variables around those 95 spillovers, then we used those results to find where else on earth was most environmentally similar.

11 of 34

Now what?

We had risk scores for each 25 sq km patch of earth for each year from 2002-2020.
How to turn this from an academic to a journalistic endeavor?

12 of 34

Who is affected?

The number of people living in areas at highest risk for spillover has grown by 57% over the two decades.
Nearly 1.8 billion people -- 1 of 5 on the planet -- now live in these areas.

13 of 34

New population datasets offer new opportunities

Programs like Worldpop and Gridded Population of the World make it easier to estimate who is affected by an floods, wildfires, hurricanes, tornadoes, heat or other events.
Particularly useful in areas where good census data doesn’t exist or is very infrequent.
It’s not perfect. But it is a big step up from what we had previously: Nothing.

14 of 34

The risk of pandemics is linked to the battle between the global economic system and the environment.

So too is climate change.

The stakes are existential.

15 of 34

West Africa

We focused on the role of mining in West African countries.
A number of countries across Africa have online mining cadastrals, which gives you the geographic boundaries of a concession as well as the ownership and the metals.
Many countries use a product from Trimble. You can scrape this data. I provide some simple R code at the end to do this. It works on each of their portals.
Why scrape it? Because you can use GIS methods to link it to other data, such as tree loss, which you can’t do when it’s locked up in the web portal.

16 of 34

17 of 34

Scrape that data!

rev_pro_getter <- function(my_url) {

mining_licenses <- geojson_sf(my_url) %>%

rowid_to_column()

}

rev_pro_getter("https://repo-prod.revenuedev.org/api/map/geojson/GH/ws/-1?status=Active%20Licenses&type=&owner.id=&minerals.id=")

18 of 34

Satellite data shows that more than 100 sq km of tree cover has been lost across the concession – about 22% of the forest that existed in 2000. The company says farmers cut down most of those trees. That’s a typical pattern around mines in the region, because the sites attract more people than can find jobs. The newcomers often then turn to farming.

19 of 34

Kerala

20 of 34

21 of 34

Yunnan Province and northern Laos

This story relied on the intersection of academic literature, infrastructure and habitat disruption.
Google Scholar is your friend. Never forget that academic literature can be an amazing source of structured and unstructured data.
Look for appendices, supplemental data, etc.

22 of 34

Structured data

23 of 34

24 of 34

Sorta-kinda unstructured data

25 of 34

26 of 34

Interlude

27 of 34

Get familiar with OpenStreetMap.

28 of 34

Brazil

Brazil is a black box of viral risk.
The Amazon is so vast that we still don’t know what lurks in one of the most bat species rich places on earth.
As we build roads deeper into the Amazon, humans risk coming into contact with potentially deadly viruses unknown to scientists.

29 of 34

30 of 34

Final thoughts…

If you cover climate change, investing in data skills is an extremely good idea.
If you cover climate change and already have some data skills, GIS (mapping!) is an excellent second step.

We have incredible amounts of spatial data on population, infrastructure, pollution, sea level rise, temperatures, fires, floods and more.
Commercial satellite providers growing by leaps and bounds. If you have mapping skills, you’re not just limited to looking at jpgs of imagery. You can layer multiple layers on top of one another to gain new insights.
If you’re not familiar with Google Earth Engine, there’s no time like the present. It’s one of the most important journalistic tools to come along in a long while.