Fig. 4: Changes in the frequency of (moist-)heat extreme events induced by irrigation expansion.

a–i Impacts of irrigation expansion (IE) on the frequency of the events in which 2-meter air temperature (T_2m: a, d, g), HUMIDEX (HU: b, e, h), and wet-bulb temperature (T_w: c, f, i) exceed their 99.9th percentile values of the first 30 years (1901–1930) in the simulations without irrigation expansion (1901irr) (shown in Fig. 3a–c). The spatial coverage include three regions: 130–60°W and 20–60°N (a–c), 20°W–50°E and 20–60°N (b–f), 50–120°E and 5–45°N (g–i). The location of these regions can be found in Fig. 3j–l. Impacts are quantified by probability ratio (PR: Eq. (5)) which is calculated by dividing the events frequencies in the new exp_period by those in the reference exp_period (see the lowest row in Table 1), and the values are the sixth root of the product of PR calculated from the outputs of six ESMs. Hatches indicate that signals (>1.2 or <1/1.2) are agreed by ≥5 of 6 models (results of individual models can be found in Fig. S22–27). Results of other metrics can be found in Fig. S7. In Fig. S4–6, the impacts of all forcings (ALL), all forcings except irrigation expansion (ALL-IE), and IE on the frequency of the events in which T_2m, HU, and T_w exceed their 99th, 99.5th, 99.9th percentile values during 1901–1930 in the simulations 1901irr are shown (results of other metrics can be found in Supplementary Figs. S37–42).