Robust and Uncertain Sea-Level Pressure Patterns 2 over Summertime East Asia 3 in the CMIP6 Multi-Model Future Projections

Robust and uncertain sea-level pressure patterns over summertime East Asia in the future 28 global warming projections and their causes are studied by applying the inter-model empirical 29 orthogonal function (EOF) analysis to the multi-model experiments in the sixth phase of the 30 Coupled Model Intercomparison Project (CMIP6) and focusing common features with the previous 31 CMIP5 analysis. The ensemble average and the first to third EOF modes associated with future 32 pressure changes are similar to the corresponding ones from CMIP5. The first and second modes 33 represent strengthened and weakened high pressure systems in subtropical and northern East Asia, 34 respectively. The third mode is the reverse anomaly of the climatological pressure pattern over 35 summertime East Asia, indicating weakened southerly monsoon winds. The second mode pattern 36 makes positive contributions to almost all the CMIP6 future pressure changes, representing a robust 37 future projection pattern. The robust mode is the result of surface warming over the northern 38 continents and neighboring seas that is stronger than the global average. The first and third modes 39 are considered to be uncertain (but major) patterns in the ensemble projections because the signs of 40 their contributions to the future changes are dependent on the model used. Suppressed vertical 41 motion over the equatorial (northern) Indian Ocean caused by the vertically stabilized atmosphere 42 under the global warming scenario is the source of the first (third) mode, together with the counter 43 vertical motion anomaly over the equatorial (northern) Pacific. The above characteristics of the 44 modes are essentially similar to those identified in the CMIP5 analysis while different sea surface temperature anomalies are related to the secondary structures of the modes. Some uncertainties in the future projections can be attributed to the systematic differences in the model climatology of the present-day precipitation, which determines the distribution of the suppressed vertical motion under 48 the future warmer climate.


Introduction
Future changes in the CMIP6 ensemble mean sea-level pressure (dslpMEAN) and the present-148 day climatology of mean sea-level pressure (slpMEAN) are shown in Fig. 1a. The dslpMEAN and 149 slpMEAN from CMIP5 are also shown in Fig. 1b for comparison. The dslpMEAN and slpMEAN 150 from CMIP6 are fairly similar to those from CMIP5 (hereafter referred to as dslpMEAN_CMIP5 and 151 slpMEAN_CMIP5) over Asia and the Indo-Pacific region, including East Asia. In both CMIP6 and 152 CMIP5, dslpMEAN is characterized by lower pressure over northern Asia and higher pressure over 153 the tropical ocean than the present day.

154
The first EOF mode (dslpEOF1) explains 65.6% of the total multi-model variance of the future 155 sea-level pressure anomalies (dslp) over the East Asian EOF domain (Fig. 2a). The inter-model 156 correlation between dslp and dslpEOF1 resolution coefficients (dslpCOR1:see the Appendix) 157 represents the strengthened Pacific high-pressure system expanding over the subtropical Pacific and 158 along the continental coast from South Asia to East Asia. The dslpEOF2 (Fig. 2b) mode represents 159 12.7% of the total variance. The spatial pattern of dslpCOR2 shows a low-pressure anomaly over 160 northern East Asia and a high-pressure anomaly over the tropical oceans. The dslpEOF3 (Fig. 2c) 161 pattern is roughly reverse to the summertime climatological distribution of sea-level pressure over 162 East Asia, indicating weakening of the southerly East Asian monsoon wind. The dslpEOF4 to 163 dslpEOF6 modes (Fig. 2d-f) have tripolar anomalies over East Asia that explain <5% of the total 164 variance; they show high-pressure anomalies over northern and tropical East Asia, the Okhotsk High 165 anomaly, and a high-pressure anomaly over Japan, respectively.
Resolution coefficients of dslpEOF1-6 by the dslpEOF1-5 from the previous CMIP5 analysis 167 (hereafter referred to as dslpEOF1-5_CMIP5: see the Appendix) are shown in Table 2. The dslpEOF1 168 to dslpEOF3 modes are similar to the corresponding modes from the CMIP5 (dslpEOF1_CMIP5 to 169 dslpEOF3_CMIP5), and share >75% of the variance each other. Each variance of the dslpEOF4 and 170 dslpEOF6 modes is broadly divided into the dslpEOF4_CMIP5 and dslpEOF5_CMIP5 modes. Note where the double parentheses mean a calculation of the area-weighting inner product over the East on dslpEOF1 will be given in the next subsection.

228
The dtasCOR3 pattern shows some positive SST anomalies in the subtropical northwestern

247
The correlation between dslpEOFs and the present-day precipitation (prCORs) in Fig. 5a-d will 248 be discussed later. (du200MEAN) and its present-day climatology (u200MEAN). The du200MEAN is similar to that of 293 CMIP5, except that the future decrease in the East Asian jet stream is found in lower latitudes.

294
The distribution of du200COR1 ( Fig. 7b) is also similar to that of the CMIP5, but its magnitude 295 is significantly weaker, especially in East Asia. The significant tropical westerly anomalies between   the relatively dry (wet) present-day climatology (Fig. 5d). We suggest that the mechanism for 396 dslpEOF3 is basically the same as that for the summertime stationary waves produced by the monsoon 397 diabatic heating over South Asia only (Ting 1994).

398
Major differences from the CMIP5 analysis are observed in the SST anomalies related to the 399 dslpEOFs. However, their major characteristics, including the basic structures and sources, are not 400 affected, although the SST anomalies are related to the secondary structures of the dslpEOFs.

402
The results regarding the robust pattern from the land warming and the uncertain patterns from 403 the vertical motion anomalies over the oceans are reasonable because, in general, the warming process 404 over land is determined relatively simply by modeling of the land surface energy budget, whereas the 405 vertical motion process over the oceans involves much more complicated modeling, such as ocean 406 circulation, atmospheric convection, and SST in the ocean surface flux budget.

407
The suppressed vertical motion anomalies or changes by the vertically stabilized atmosphere 408 under global warming are closely related to the present-day precipitation climatology in the model 409 simulations (w500CORs in Fig. 5 and prCORs in Fig. 6). This may lead to the possibility that the 410 uncertainty associated with the dslpEOFs could be reduced by comparing the modeled and observed 411 precipitation climatology.

412
The higher modes of the dslpEOFs have fine structures, which are not necessarily the same as the higher modes from the CMIP5 analysis (  in the case of k=3 and f=tas, the notations such as "dslpEOF3", "dtasMEAN", "dtasCOR3" and "SD3" 493 are used instead of "dslpEOF.3.i", "dtasMEAN.i", "dtasCOR.3.i" and "SD.3". The same statistical 494 variables but from the CMIP5 case are denoted such as "dslpEOF3_CMIP5", "dtasMEAN_CMIP5",