Geopolitics of Environmental Disasters, Future
October 5, 2009 • Jason Hines
We’ve talked about environmental disasters earlier here – and it’s an interesting area to explore. Environmental issues/disasters have many impacts in both short and long term: economics, poverty, food/crop devastation, water supply, national security, and of course, the actual environment in itself.
Recorded Future collects environmental events from a variety of sources – government sources, newspapers, blogs, etc. – and organizes them by date of publication as well as the date of the event in itself – which might be historical (an environmental disaster in the 70s), something happening right now, a water shortage just over the time horizon, or a conclusion on global warming in 2025.
The two current core event types for environmental analysis are Environmental Issue and Natural Disaster (inheriting from our friends at OpenCalais!). We can explore these by publication time and event time – the former being the date an article gets published and the latter being the actual date of say a natural disaster – last year, today, or perhaps impending next week.
Natural disasters by publication date
Let’s begin by reviewing natural disasters by publication date – as below – with the blue curve being the daily count and the red area being the 7 day moving average of natural disasters. You clearly see ebbs and floods (not trying to be funny!) in disasters, and by clicking the flags you can see recent events such as Earthquake in Sumatra, rains in Karnataka, Chinese earthquake, Taiwanese typhoon, etc. You also can find interesting connections such as piracy being expected to increase around the Horn of Africa following end of Monsoon rains.
Note that the graph is interactive – you can zoom it and click the markers. It’s also updated in real time as you refresh this page. X axis is time, Y axis is count.
Clearly you’d like baseline this data and review cycles over long time periods. Alerts based on geographic areas going out of their normal cycle could also be highly helpful.
Environmental issues by publication date
Likewise we can explore environmental issues by publication date – as below. Now the peaks are much more driven (at least in this time period) by policy discussions. The big peak on September 22nd is a good example (big UN meeting). However – policy is connected to events – and natural disasters should be a good “connector”.
Natural disasters by event date
We can also explore the actual date of events, rather than just the dates they were published (the power of Recorded Future!). Sometimes of course these two dates align, but in many cases they don’t. An article might cover disasters from last year, today, or that might go down next week. Review this visualization where events are displayed by their event date, stretching in 2010.
The peak around October 1st, with earthquake in Sumatra and rains in Karnataka stands out again. The shapes of the natural disaster curve by publication date for the same time period is largely the same.
However – the interesting part obviously becomes to review the future. It would be a bit optimistic to find actual future natural disasters spelled out (although it happens), however we do find quite interesting content ranging from short term future to the very long term:
- Future changes of budgets and other short term policy actions to deal with disasters that just have happened (like in the Philippines)
- Indications of future water shortages – “Beijing faces a serious water shortage that cannot be alleviated before 2014 at the earliest”
- and all the way to “Here is the weather for 2080: floods, droughts and heatwaves”
Environmental issues by event date
Now, when we explore environmental issues the timeline and the overall information content is richer in the future – which makes sense since disasters tends to be more “point in time” whereas environmental issues are many times early indications of what’s to come.
Here we can find issues from just around the corner until predictions into 2050 – and global warming plays a big role.
There are certainly some overlaps between natural disasters and environmental issues, in particular when you look further into the future beyond say the next 12 months – and in particular around global warming issues where the distinction between a “natural disaster” and an “environmental issue” certainly can be pretty vague. But interesting follow-on analysis would include
- Certain classes of disasters – e.g. hurricanes have cycles to them and lend themselves to predictions. How can we take advantage of that?
- Exploring for classes of events, say water shortage, how does what is initially reported as an environmental issue turns into an actual disaster – can that be modelled? Are certain sources more interesting than others in meaningful early indications?
- Can we build chains of events from early warnings through intermediate events (indications of water shortage) to outright hunger/draught to downstream violence?
- For what types of environmental and natural disaster data does open sources such as those covered by Recorded Future bring an edge compared to more classic say meteorological approaches?
- How can Recorded Future data be combined with other approaches and data to provide new edges in environmental analytics?