ClimAte DISASTERS

Matthew Wielicki repeatedly posts these charts, claiming that climate disasters are not getting worse with climate change.

What should we make of these?

We can easily download the source global disaster data from EM DAT (https://www.emdat.be/), and get an Excel spreadsheet with 8,147 rows, each row giving details of one ‘disaster’.  If we mindlessly plot the number of rows by year, we can easily recreate Wielickis chart.

Does it mean anything to merely count rows? Let’s look at two such rows:  the Canadian fires in 2001 is counted as ‘one’ disaster in 2001. Similarly, the record fires across multiple Canadian provinces in 2023 are also just one ‘disaster’.   

From other sources, we can see that the 2001 CA fires burned 607,000 hectares, the 2023 fires burned 11.5 million hectares (19 times as much). Yet Wielickis visual counts these equally. So according to Wielickis “analysis”, the EM –DAT data shows that there was no change to Canadian forest fires between 2001 & 2023 !

Similarly a storm in 2000 in the US that effected just 34 people counts equally to the 2022 floods of Pakistan that impacted 33 million and (from other sources) destroyed 1 million homes. Yet Wielicki treats these as being equal – they both count as 1.  

Possibly Wielikis chart of deaths, as in his second chart, is more meaningful and captures the actual human impact. 

Deaths, rather than lives severely impacted, and livelihoods destroyed, is obviously a flawed way of measuring climate change. As medicine advances over the years, and emergency response improves, we should see deaths reduce, even as the human suffering and costs from climate change might increase. And deaths are relatively rare – the dataset also includes a count of those affected (lost homes, displaced etc.) that outnumbers deaths by a factor of 77,000 to 1 !    That would have made a much more sensible measure to use.

But let’s look at the deaths data as Wielicki did, and does it make sense to talk of trends by mindlessly looking at deaths per year?

In the period since 2000, there have been 630,000 deaths from climate related disasters according to the EM DAT data. A staggering 138,000 (22%) occurred from just one event – the Myanmar cyclone in 2008.   This already raises alarm bells for a data analyst – talking about trends over a short period of 20 years where you have only seen ONE event of such magnitude is just silly.

Let’s look at a specific type of event that we would expect to increase with climate change – extreme heat events and droughts. The EM DATA website in their ‘must read documentation’ state that:    “For droughts, EM-DAT fails to capture the associated mortality”, and provide charts showing how infrequently deaths get captured/reported for droughts. 

Indeed, from the raw data we see that since 2021, droughts have affected 1.6 billion people, more than any other type.

Yet in almost all those cases, covering droughts that effected 1.55 billion people, associated deaths were not recorded, so those huge climatic events do not even feature on Wielickis chart !

Given droughts are more likely to arise from a warmer planet, v.s. say cyclones, it is incredibly foolish to use deaths as the metric if it is rarely even recorded for the type of event that effects the most people, and is most likely to be impacted by climate change!

So let’s look at heatwaves, which account for the most deaths in the data, are much more commonly reported (about 80% of the time), and of course we know heatwaves are more linked to increasing temperatures.

According to the EM DATA data, heatwave deaths totaled 221,000 since 2020, which is over a third of ALL recorded climate related deaths. The vast majority of these (94%) occurred in Europe, likely a reflection of the quality of data reporting from Africa and India vs Europe, more than a reflection of actual reality (this is born out by this statement by EM DAT “In general, EM-DAT has a relatively worse coverage for Sub-Saharan Africa regarding the occurrence and the accounting of impact variables”).

We see it is concentrated in just 3 years – 2003, 2010, and 2022 – corresponding to the 3 European heatwaves.

To talk of ‘trends’ in the data, as Wielicki does, when you effectively have just three data points of these three European heatwaves, is ridiculous. And this begs the question, what of the record high temperatures in  2023? The dataset shows no entry at all from Europe 2023, when we  know that the planet had record high temperatures, and other sources report heatwave deaths in Europe.  How could there be no entry for Europe 2023? Because it can take months for these entries to be made – the database has a record of when entries are added, and as we can see from the 2022 heatwave events, it took more than a year (till July 17 2023) before entries were made.  

So Wielickis errors/misrepresentations on his chart of deaths from climatic events are numerous:

• He aggregates data from all types of event, mixing those that are predicted to be impacted from warming (heatwaves, droughts) with those where the prediction is much weaker, or non-existent (eg cyclones).

• Whilst drought affects more people than any other type of climatic events, and drought is more likely to increase with climate change (vs, say, cyclones), he totally ignores the fact that the number of deaths is rarely even recorded for droughts!  So droughts that affected 1.55 billion people do not even feature on this chart.

• He ignores the fact that you cannot expect to see clear trends in data over a relatively short period, if the data is dominated by a tiny handful of events (for heatwave deaths, just 3 large events).   

• He ignores the fact that the data record for recent years is incomplete, and additional entries are very likely.