Predicting the severity of drought using multiple indices
WHO: Liu Sun, Scott W. Mitchell (Department of Geography and Environmental Studies, Geomatics and Landscape Ecology Research Laboratory, Carleton University, Ottawa, Ontario, Canada)
Andrew Davidson (Department of Geography and Environmental Studies, Carleton University, Ottawa National Land and Water Information Service, Agriculture and Agri-Food Canada, Ottawa, Ontario)
WHAT: Improving the accuracy of drought prediction in the Canadian prairies
WHEN: September 2012
WHERE: International Journal of Climatology, Vol 32, Issue 11, September 2012
Are you tired of your drought prediction methods using only two layers of soil structure to track moisture? Sick of having to work with constants when you’d much rather be using dynamically calculated values? Well, this paper is for you.
Drought is going to be a big issue with climate change as rainfall patterns change and move. Agricultural yields are not able to increase as quickly as the world’s population increases but people still need to eat.
Drought is going to affect all of us as extreme weather increases from climate change, whether it’s through increased food prices (I’m still upset about bacon), local water restrictions (stop hosing down concrete – stop it now), local ecosystems being stressed or climate refugees from newly arid areas. This is one of the great ironies about climate change – you can’t negotiate with or spin physics. The laws of physics aren’t going to change because of some slick advertising campaign trying to prop up a floundering status-quo, and climate change isn’t going to avoid you if you ignore it.
In terms of drought modelling and prediction, each method currently used has slightly different ways of predicting drought, which means they can’t easily be compared. The method the researchers used for this paper was to modify the original Palmer Drought Severity Index to include more variable data. They accounted for six soil layers and a new evaporation calculation. Instead of using constant numbers for the characteristics of the climate, they allowed each of those to be calculated too. This means most people end up with a giant math headache from extra calculations, but by allowing for greater variability, they also allowed for greater sensitivity and accuracy in their model. The new model was also tested for accuracy against the Palmer Drought Severity Index, Standardised Precipitation Data and Palmer Moisture Anomaly Index methods.
For any of these models to work, they need approximately 30 years of monthly weather data (temperature, rain etc). This paper looked at 1976 – 2003 as it was the period of most consistent data in the area they were studying (the Canadian Prairies).
Then they got into the serious math using all kinds of things like a ‘thin plate smooth spline surface fitting method’ to remove the noise from the data and a linear regression to remove yield differences from better agricultural practices, allowing them to just look at the data that was climate affected.
It went pretty well; their predictions were more accurate than the other standard drought prediction methods, except for predicting extreme drought, which their model under-predicted. This is possibly because there wasn’t a lot of data points in the previous 30 years with extreme drought, so as extreme weather becomes more normal under climate change, their model will probably get more accurate. They also found that the model is more accurate for arid locations, as flooding messes up the model.
As the extreme, unpredictable realities of climate change start to affect everyone in the next decade or so, this drought prediction model will likely be very useful. Predicting the extremes as best we can is going to become an essential tool for preventing massive crop failures as well as loss of human lives.