Full text
Supporting Information for “Tropical Humid Heat Stress Crosses a Critical Threshold in a Warming World” Subodh Kumar Saha1, Yashas Shivamurthy1, B. N. Goswami2 1Indian Institute of Tropical Meteorology, Dr Homi Bhabha Road, Pune, India 2Department of Physics, Gauhati University, Guwahati, India October 16, 2025, 9:41am
X - 2 : Figure S1. Global average exposure of a person, in hours, to the WBGT >45◦C and Humidex >45◦C at various relative humidity levels. Extreme heat exposure (shaded colour) to the global a) entire population, b) population living in rural areas, c) population living in urban areas for WBGT >45◦C. d)-f) are similar to a)-c) respectively, but for Humidex >45◦C. Exposure hours are estimated by calculating POPL, categorised in 5% relative humidity intervals (y-axis on the left) at each grid box, and then normalized (divided) by the number of people living in the grid box. October 16, 2025, 9:41am
:X - 3 Figure S2. Total working population (ages 25-60 years) and tendency. The working population of India (red line) and globally (black line) has been steadily increasing over the past seven decades, but the rate of growth (i.e. tendency) has been declining in the last decade (dotted lines). October 16, 2025, 9:41am
X - 4 : Figure S3. hPOPL (in person-days) using nine models from CMIP6 historical and piControl runs (1950-2014). a) Ensemble-mean global hPOPL with Humidex >45◦C for historical (blue line) and piControl (pink line). Light blue and light red shading indicate ensemble spread. The differences between average historical and piControl runs are shown by black line, with its scale on the right. b) Same as in a) but for India. October 16, 2025, 9:41am
:X - 5 Figure S4. Annual cycle of Humidex and rainfall in Delhi (77◦E, 28.5◦N). a) Average Humidex from 1940 to 1949 (dark red line) and Humidex (red line) and rainfall (green line) for the year 1940. b) Average Humidex over the recent decade (2010 to 1949; dark red line) and Humidex (red line) and rainfall (green line) for the year 2019. October 16, 2025, 9:41am
X - 6 : Figure S5. (a,b) Time series from 1950 to 2020 showing the nonlinear increasing trend and interannual variability of hPOPL (in person-hours) for global and the top 9 other affected countries for Humidex >45◦(solid colour lines, see legend). POPL with WBGT >33◦with its scale on right side (dashed lines). (c,d) Similar to (a,b) but for dPOPL for global (black) and several affected countries (colour lines, see legends) for Humidex >45◦. It may be noted that the maximum of global hPOPL (8.55 ×1011) is approximately 20 times larger than the maximum of global dPOPL (4.33 ×1010). October 16, 2025, 9:41am
:X - 7 Figure S6. Latitudinal distribution of zonal mean POPL and its impact on rural and urban populations. a) Zonal accumulated hPOPL based on WBGT >33◦C during the 1950s (19501959) and the recent decade (2011-2020) for the total population (black line), rural (green line) and urban population. b) Same as in a), but for dPOPL. c) and d) are same as a) and b) respectively, but for Humidex >45◦C. October 16, 2025, 9:41am
X - 8 : Figure S7. The spatial patterns of first three EOFs and PCs of annual mean D20 anomalies between 1950 and 2020. (a-c) The variance explained in the spatial patterns of EOF1, EOF2, and EOF3 (arbitrary units) is shown in the inset. (d-f) The first three principal components of the time series are normalized by their respective standard deviation (black lines). Anomaly correlations between PCs and hPOPL for India (red line) and globally (blue line) are indicated by the respective colors. October 16, 2025, 9:41am
:X - 9 Figure S8. Same as Figure 6, but using data of 71 years (1950-2020). Table S1. Description of CMIP6 models (piControl and historical) used in this study for the period 1950-2014. October 16, 2025, 9:41am
