How does Mei-yu precipitation respond to climate change?

ABSTRACT Mei-yu is an important weather phenomenon in the middle-lower Yangtze River valley (YRV) region. This study investigates the changes in the characteristics of Mei-yu under global warming and the potential reasons based on observation and reanalysis data during 1961–2022. Notable increasing long-term trends are detected in the number of days without rainfall (NDWOR), the intensity of rainfall events, and the frequency and intensity of extreme precipitation events (EPEs) in the YRV region during the Mei-yu period (15 June–10 July) over past decades. The increasing trend in NDWOR is attributed to decreased relative humidity over land surface and a longer time for the air to be replenished with moisture after rainfall events in a warming climate. The increasing trends in the intensity of rainfall events and frequency/intensity of EPEs are attributed to the strengthened transient water vapor convergence and convection in the atmosphere under global warming. Furthermore, the response of Mei-yu to 2°C of global warming with respect to the pre-industrial climate is analysed using CMIP6 models. The results suggest that the NDWOR, intensity of rainfall events and frequency of EPEs will increase in the YRV region during the Mei-yu period under the 2°C warming scenario, which implies a more challenging climate risk management in the future. Overall, the intensity of rainfall events during the Mei-yu period has the most significant response to climate change in observations and projections. The model results have a relatively large uncertainty.

Figure S1.Climatology during the Mei-yu period.(a) Climatology of water vapor transport (unit: kg m -1 s -1 , vector) and water vapor divergence (unit: 10 -5 kg m -2 s -1 , color), and (b) precipitation (unit: mm d -1 ) during June 15-July 10 during 1961-2022.The purple contour in (a) denotes the isoline of 5880 gpm in gepotential height, which represents the western Pacific subtropical high.The station data in Taiwan Province is not obtained in (b).

Figure S2 .
Figure S2.(a) Spatial distribution of long-term trends in NDWOR (unit: d/10a) during the Mei-yu period during1961-2022.(b) Time series of NDWOR (unit: d) in stations over the YRV region during the Mei-yu period.The "+" symbols in (a) denote where the trend is significant at the 90% confidence level based on the Student's t-test.The black and red lines in (b) represent time series of individual stations in the YRV region and their mean, respectively.The blue line in (b) represents the linear trend of the mean time series during 1961-2012.The station data in Taiwan Province is not obtained in (a).

Figure S3 .
Figure S3.Time series of factors associated with Mei-yu.Time series of areal mean (a) NDWOR (unit: d), (b) intensity of rainfall event (unit: mm d -1 ), (c) frequency of EPE (unit: d), (d) intensity of EPE (unit: mm d -1 ), (e) surface air temperature (unit: ℃), (f) surface q s (unit: g kg -1 ), (g) total column water vapor (unit: kg), and (h) surface RH (unit: %) in the YRV region during the Mei-yu period during (a-e, g) 1961-2022 and (f, h) 1961-2020.The red and blue lines in denote the long-term trend and the nine-year sliding average time series, respectively.

Figure S4 .
Figure S4.EEMD decomposition of time series of NDWOR.The (a) original time series, (b-e) IMFs and (f) trend components of areal mean NDWOR (unit: d) in the YRV region during the Mei-yu period based on EEMD.

Figure S5 .
Figure S5.EEMD decomposition of time series of intensity of rainfall events.The (a) original time series, (b-e) IMFs and (f) trend components of areal mean intensity of rainfall events (unit: mm d -1 ) in the YRV region during the Mei-yu period based on EEMD.

Figure S6 .
Figure S6.Long-term trend in total precipitation amount during the Mei-yu period.(a) Spatial distribution of long-term trend in total precipitation amount (unit: mm/10a) during the Mei-yu period during 1961-2022, and (b) time series of areal mean total precipitation amount (units: mm) in the YRV region during the Mei-yu period during 1961-2022.The "+" symbols in (a) denote where the trend is significant at the 90% confidence level based on the Student's t-test.The red and blue lines in (b) denote the long-term trend and the nine-year sliding average time series, respectively.The station data in Taiwan Province is not obtained in (a).

Figure S7 .
Figure S7.EEMD decomposition of time series of frequency of EPE.The (a) original time series, (b-e) IMFs and (f) trend components of areal mean frequency of EPE (unit: d) in the YRV region during the Mei-yu period based on EEMD.

Figure S8 .
Figure S8.EEMD decomposition of time series of intensity of EPE.The (a) original time series, (b-e) IMFs and (f) trend components of areal mean intensity of EPE (unit: mm d -1 ) in the YRV region during the Mei-yu period based on EEMD.

Figure S9 .
Figure S9.EEMD decomposition of time series of surface air temperature.The (a) original time series, (b-e) IMFs and (f) trend components of areal mean surface air temperature (unit: ℃) in the YRV region during the Mei-yu period based on EEMD.

Figure S10 .
Figure S10.EEMD decomposition of time series of surface q s .The (a) original time series, (b-e) IMFs and (f) trend components of areal mean surface q s (unit: g kg -1 ) in the YRV region during the Mei-yu period based on EEMD.

Figure S11 .
Figure S11.EEMD decomposition of time series of total column water vapor.The (a) original time series, (b-e) IMFs and (f) trend components of areal mean total column water vapor (unit: kg) over the YRV region during the Mei-yu period based on EEMD.

Figure S12 .
Figure S12.EEMD decomposition of time series of surface RH.The (a) original time series, (b-e) IMFs and (f) trend components of areal mean surface RH (unit: %) over the YRV region during the Mei-yu period based on EEMD.

Figure S13 .
Figure S13.Long-term trends in RH (unit: %/(10a)) at the 850-hPa pressure level during the Mei-yu period during 1961-2022 based on the (a) ERA5 and (b) NCEP reanalysis data.Stippling denotes where the trend is significant at the 95% confidence level based on the Student's t-test.

Figure S14 .
Figure S14.Long-term trends in wind speed and vertical velocity during the Mei-yu period.Spatial distribution of long-term trend in (a) surface wind speed (unit: (m s -1 )/10a), (b) 850-hPa wind speed (unit: (m s -1 )/10a), and (c) 500-hPa vertical velocity (unit: (Pa/s)/10a) during the Mei-yu period during (a) 1961-2017 and (b, c) 1961-2022.Time series of areal mean (d) surface wind speed (unit: m s -1 ), (e) 850-hPa wind speed (unit: m s -1 ), and (f) 500-hPa vertical velocity (unit: Pa s -1 ) over the YRV region during the Mei-yu period during (d) 1961-2017 and (e, f) 1961-2022.The "+" symbols in (a), (b), and (c) denote where the trend is significant at the 90% confidence level based on the Student's t-test.The red and blue lines in (d), (e), and (f) denote the long-term trend and nine-year sliding average time series, respectively.The station data in Taiwan Province is not obtained in (a).

Figure S16 .
Figure S16.Simulated response of Mei-yu factors to temperature rising in CMIP6 models eliminated in this study based on historical experiments (1961-2014).The left subplot represents the scatter plot of areal mean NDWOR (unit: d; Y-axis) vs surface air temperature (unit: ℃; X-axis) in the YRV region during the Mei-yu period.The middle subplot represents the scatter plot of areal mean intensity of rainfall events (unit: mm d -1 ; Y-axis) in the YRV region during the Mei-yu period vs global mean surface temperature (unit: ℃; X-axis) during summer.The right subplot represents the scatter plot of areal mean frequency of EPE (unit: d; Y-axis) in the YRV region during

Figure S17 .
Figure S17.Simulated response of Mei-yu variables to temperature rising in CMIP6 models selected in this study based on historical experiments (1961-2014).Scatter plot of areal mean (a, d, g) NDWOR (unit: d; Y-axis) vs surface air temperature (unit: ℃; X-axis) in the YRV region during the Mei-yu period, (b, e, h) areal mean intensity of rainfall events (unit: mm d -1 ; Y-axis) in the YRV region during the Mei-yu period vs global mean surface temperature (unit: ℃; X-axis) during summer, (c, f, i) areal mean frequency of EPE (unit: d; Y-axis) in the YRV region during the Mei-yu period vs global mean surface temperature (unit: ℃; X-axis) during summer.(a, b, c) EC-Earth3, (d, e, f) FGOALS-g3, and (g, h, i) ACCESS-CM2.The black lines denote linear regression.

Table S1 .
List of information for the different experiments of CMIP6 models.The models in bold indicate the preferred models.