A revival of Indian summer monsoon rainfall since 2002

Transcription

1 In the format provided by the authors and unedited. SUPPLEMENTARY INFORMATION DOI: /NCLIMATE3348 A revival of Indian summer monsoon rainfall since 2002 Qinjian Jin and Chien Wang* Center for Global Change Science, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA. * wangc@mit.edu NATURE CLIMATE CHANGE Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

2 Table 1. CMIP5 simulations with number indices corresponding to Figure 4 in the main text. 1 ACCESS1-0_r1i1p1 19 CSIRO-Mk3-6-0_r3i1p1 37 inmcm4_r1i1p1 2 ACCESS1-3_r1i1p1 20 CSIRO-Mk3-6-0_r4i1p1 38 IPSL-CM5A-LR_r2i1p1 3 CCSM4_r1i1p1 21 CSIRO-Mk3-6-0_r5i1p1 39 IPSL-CM5A-LR_r3i1p1 4 CCSM4_r2i1p1 22 CSIRO-Mk3-6-0_r6i1p1 40 IPSL-CM5A-LR_r4i1p1 5 CCSM4_r3i1p1 23 CSIRO-Mk3-6-0_r7i1p1 41 IPSL-CM5A-MR_r1i1p1 6 CCSM4_r4i1p1 24 CSIRO-Mk3-6-0_r8i1p1 42 IPSL-CM5B-LR_r1i1p1 7 CCSM4_r5i1p1 25 CSIRO-Mk3-6-0_r9i1p1 43 MIROC-ESM-CHEM_r1i1p1 8 CCSM4_r6i1p1 26 GFDL-CM3_r1i1p1 44 MIROC-ESM_r1i1p1 9 CESM1-BGC_r1i1p1 27 GFDL-ESM2G_r1i1p1 45 MIROC5_r1i1p1 10 CESM1-CAM5_r1i1p1 28 GFDL-ESM2M_r1i1p1 46 MIROC5_r2i1p1 11 CESM1-CAM5_r2i1p1 29 GISS-E2-H-CC_r1i1p1 47 MIROC5_r3i1p1 12 CESM1-CAM5_r3i1p1 30 GISS-E2-H_r1i1p1 48 MPI-ESM-LR_r1i1p1 13 CMCC-CM_r1i1p1 31 GISS-E2-H_r1i1p2 49 MPI-ESM-LR_r2i1p1 14 CMCC-CMS_r1i1p1 32 GISS-E2-H_r1i1p3 50 MPI-ESM-LR_r3i1p1 15 CNRM-CM5_r1i1p1 33 GISS-E2-R-CC_r1i1p1 51 MPI-ESM-MR_r1i1p1 16 CSIRO-Mk3-6-0_r10i1p1 34 GISS-E2-R_r1i1p1 52 MRI-CGCM3_r1i1p1 17 CSIRO-Mk3-6-0_r1i1p1 35 GISS-E2-R_r1i1p2 53 NorESM1-M_r1i1p1 18 CSIRO-Mk3-6-0_r2i1p1 36 GISS-E2-R_r1i1p3 2

3 Figures Figure 1 Precipitation trends with 12-year moving window. This figure is created based on the time series in Figure 1a in the main text. Squares represent significant trends at 95% confidence level. 3

4 Figure 2 Time series of the ISM precipitation (mm day 1 ) and spatial patterns of its linear trends (mm day 1 decade 1 ). This figure is similar to Figure 1 in the main text, but for four individual months during. Data set is from the Climatic Research Unit (CRU). Black dots represent linear trends that are significant at 95% confidence level. 4

5 Figure 3 Time series of surface temperatures and spatial patterns of their linear trends during a. Time series of 5-year running mean temperatures ( C) area-averaged over Tropical Indian Ocean (blue rectangle in Supplementary Figure 3b) for four seasons. b e, Spatial patterns of linear trends of surface temperatures ( C decade 1) for boreal winter, spring, summer, and fall, respectively. Black dots represent linear trends that are significant at 95% confidence level. 5

6 Figure 4 Time series of surface temperatures and spatial patterns of their linear trends during a. Time series of 5-year running mean temperatures ( C) area-averaged over the Indian subcontinent (green rectangle in Supplementary Figure 3c) for four seasons. b e, Spatial patterns of linear trends of surface temperatures ( C decade 1 ) for boreal winter, spring, summer, and fall, respectively. Black dots represent linear trends that are significant at 95% confidence level. 6

7 Figure 5 Correlation between precipitation and pre-monsoon surface temperature during Precipitation is area-averaged over NCI during the ISM season. The black dots represent grid points that are significant above 95% level. Figure 6 Same as Supplementary Figure 5, but using temperature at 850 hpa from NCEP/NCAR reanalysis. 7

8 Figure 7 Precipitation trends (mm day 1 decade 1 ) and frequency distribution. a, The ISM precipitation trends from 3 observations and 53 CMIP5 historical runs during b and c, The same to a, but for the period of In b and c, the precipitation data is from CMIP5 historical runs during and from RCP45 and RCP85 runs during , respectively. d e, Frequency distribution of precipitation linear trends corresponding to a c, respectively. The solid filled bars are trends that are significant above 95% level, while the white blank bars are trends that are significant below 95% confidence level. The black stars are observational trends and the purple stars are the ensemble mean of the 53 CMIP5 runs. 8

9 Figure 8 The same as Supplementary Figure 6, but for land ocean surface temperature gradient in pre-monsoon season. The surface temperature gradient is defined as the temperature difference between the Indian subcontinent and the TIO. 9

10 Figure 9 Patterns of decadal trends corresponding to Figure 5 in the main context. (a) and (b) surface net and downward shortwave radiation (W m 2 decade 1 ). (c) and (d) total and low cloud area fraction (% decade 1 ). (e) and (f) Relative humidity (% decade 1 ) and specific humidity (kg kg 1 decade 1 ). (g) (i) Evaporation (E), negative precipitation ( P), and moisture flux convergence (MFC; mm day 1 decade 1 ). Trends are calculated in the pre-monsoon season from 2002 to Black dots represent linear trends that are significant at 95% level. 10

11 Figure 10 Same as Figure 5 in the main text, but for (a) evaporation and (b) wind speed at 10 m above the displacement height over the Arabian Sea (5 N 25 N, 50 E 75 E). (c) Aerosol optical depth for fine particles with radius smaller than 0.35 µm and (d) Ångström exponent for all particles over India (green rectangle in Supplementary Figure 2c). The values are 5-year running means of the area-averaged variables during the pre-monsoon season in The linear trends per decade are shown by the numbers in each panel with all p values smaller than

12 Figure 11 Same as Figure 9, but for (a) surface evaporation (mm day 1 decade 1 ), (b) wind speed (m s 1 decade 1 ), (c) aerosol optical depth for fine particles (decade 1 ), and (d) Ångström exponent for all particles (decade 1 ). 12