Fig. 5: The trend of the contribution of human-induced climate change on the probability of regional extreme fires during the fire season of 2002–2015.
From: Growing human-induced climate change fingerprint in regional weekly fire extremes
a For individual regions (x-axis), the y-axis shows the linear trend value of \(FA{R}_{FWI}^{MedM}\) over fire dates, where \(FA{R}_{FWI}^{MedM}\) is the multimodel median of the effect of human-induced climate change in FWI on the predicted probability of an individual extreme fire event (grey bar shows the ±1 standard error of the regression line; see ‘Methods’). The black triangles indicate regions where over 75% of climate models agree on the sign of trend. b The annual impact of human-induced climate change in FWI on the probability of extreme fire events, obtained by first averaging FARFWI across all fires of all regions for each year given one climate model and then getting multimodel median—the shading indicates ±1 standard deviation of annual global FAR from multiple climate models. The trend is illustrated by the linear regression line, with the numbers in the panel showing the slope of the regression line and its p-value. The black triangle indicates that over 75% of climate models agree on the sign of trend. c–f Similar to panel b but for the annual contribution of human-induced climate change on the probability of extreme fire events based on four FWI counterfactuals where the trend of temperature (c, FART), precipitation (d, FARP), relative humidity (e, FARRH), and wind speed (f, FARW) was removed once at a time (see ‘Methods’).
