Tags
Australia, buildings, bushfires, climate change, cooling, correlation coefficent, Craig Kelly, deaths, deniers, early warning systems, emergency services, extreme events, fatalities, fossil fuels, global warming, heating, heatwave, insurance losses, natural disasters, p-value, population growth, safety campaigns, temperatures, weather, wild weather
I’ve been posting comments to Australian politician Craig Kelly’s Facebook page. Kelly and his followers are climate change deniers, so there’s a lot of misinformation, selective information and non-information there. I posted the following after Kelly put up a graph showing the number of deaths from natural disasters in Australia from 1900 to 2010 normalised for population growth only (although he didn’t say it was only normalised for population growth and nothing else such as huge changes in building quality, early warning systems, emergency services, etc). The graph shows a declining number of deaths. To Kelly, this is proof that fossil fuels have done a great job in providing heating, cooling, etc in homes and elsewhere and therefore the number of deaths from natural disasters has fallen. His followers agreed, without of course checking the rest of the report the graph came from. I made an edit to my post on two occasions and on the second one, Facebook wouldn’t post it; it disappeared. I think they’ve got some algorithm in there to try and prevent spam or something, like I’d posted it three times. Anyway, I thought I’d post it here. The study and report, linked to below, is by the Australian Department of Industry, Innovation and Science …
I would question a number of things about this government study (which includes figure 3 shown above [in Kelly’s post]). Here’s the link: http://www.bnhcrc.com.au/news/2015/century-natural-disasters-what-are-costs.
It looks firstly at annual weather related insurance losses from 1966 to 2013. In nominal dollar terms, the losses go up a lot, which you would expect. These dollar figures are then normalised for such things as increases in the number of houses and their value, building code changes and better construction (which is a broadly similar process that temperature data goes under to get it from raw to adjusted, for factors such as weather stations moving, more concrete and steel around existing stations, and different temperature measuring instruments, although climate deniers prefer raw data as it shows much less of an increase in temperatures over time). The report says there is ‘no statistically significant upward trend’, which is broadly true over the whole period. The correlation coefficient (which can be between -1 and +1), r, for the years 1966 to 2013 is 0.09, or a possible slight increase in normalised losses. Leaving off 2013 (which was a year of many natural disasters and for which claims don’t appear to have been complete at the time of the study), r = 0.12. If we start at 1980, the time when temperatures started to increase rapidly, r = 0.23, and leaving off 2013, r = 0.29. The p-value for this last one is 0.053, so it’s just over the threshold (0.05 or 1 chance in 20 of no actual increase) between ‘reject’ and ‘not reject’; so we can reject, and only just reject, the hypothesis that there is a statistically significant upward trend in insurance losses. However, the graph (figure 1) doesn’t seem to include the 2011 Qld floods of $2.5b normalised (table 1). Adding this in, r = 0.366 for 1980 to 2012, giving a p-value of 0.018 or about a 1 in 55 chance of there being no actual increase in normalised losses over the period 1980 to 2012.
The study then looks at fatalities. The raw data for annual fatalities from 1900 to 2010 is shown in figure 2 and there looks to be no long term increase or decrease. This is then normalised in figure 3 for population growth and now shows a significant decrease in fatalities. But it doesn’t take into account the huge leaps in early warning systems, the vastly larger and better emergency and rescue services, much better knowledge, and all the community safety campaigns, all of which were virtually non-existent in 1900 (some dot points near the bottom of the page suggest that further studies might include some of this). We of course have better buildings, and they do have better heating and cooling from electricity which so far has mainly come from fossil fuels. We probably also spend more time indoors both at work and when not at work these days, therefore staying out of the heat more. I would also question the raw numbers in figure 2. Table 2 has 8256 deaths from natural disasters from 1900 to 2011. This is an average of 74 deaths per year. Yet figure 2 shows an average of perhaps around 25, with only six years where the number exceeds 74, and in four of those years it’s not much above 74. Further, a list of disasters at Wikipedia shows that fatalities from Australia’s worst disasters far exceed what is shown in figure 2, e.g. a heatwave in Victoria claimed 438 lives in 1938-39, one in SE Australia 374 lives in 2009, a heatwave in the southern states 246 lives in 1907-08, bushfires in Victoria 173 in 2009, and many more.
Thus there seems to be an increase in normalised losses, certainly since 1980. But we really can’t make much out of the normalised fatalities numbers as they only take into account population increase.
The increase in wild weather worldwide due to climate change between 1980 and 2014 is shown quite vividly in a National Geographic article: ‘Wild weather’, which starts off: ‘Torrential hurricanes, devastating droughts, crippling ice storms, and raging heat waves—all are extreme weather phenomena that can claim lives and cause untold damage. Climate change influences severe weather by causing longer droughts and higher temperatures in some regions and more intense deluges in others, say climate experts.’ The number of extreme events increased steadily from 291 in 1980 to 904 in 2014, nearly all of them weather related. See https://www.nationalgeographic.com/climate-change/how-to-live-with-it/weather.html.