Variability in the western North Pacific summer monsoon in 140-year-long AGCM hindcast experiments: SST impact on the cyclonic anomaly around 1890s–1930s

Abstract

In this study, we examine the long-term variability in the western North Pacific (Philippine Sea) summer monsoon (WNPSM) from the 1880s to the present. To determine the sea surface temperature (SST)-induced atmospheric variability, atmospheric general circulation model (AGCM) ensemble simulations are conducted with the observed SST. For the AGCM experiments, we use two convection schemes and SST datasets to examine the uncertainty in the convective parameterization and robustness of the simulated atmospheric response. In the AGCM simulation (T42; approximately 300-km horizontal resolution, 3-member ensemble in each), interdecadal variability can be seen as a cyclonic anomaly in the 1890s–1930s. This cyclonic anomaly is robust to different convection schemes and SST datasets. The observed rainfall over the land demonstrates consistent features. In our AGCM ensemble simulations, sensitivity experiments with several SST patterns show that the interbasin effect (i.e., contribution of tropical Indian Ocean and Atlantic) played a key role in the WNPSM cyclonic anomaly during the 1890s–1930s. Further, idealized SST experiments show that reduction in SST over the tropical Indian Ocean and Atlantic causes the WNPSM cyclonic anomaly. Additionally, using high-resolution AGCM experiments (T119; approximately 100-km horizontal resolution, 4-member ensemble in each), we investigate the increase in tropical cyclone (TC) activity over the western North Pacific during the 1890s–1930s. However, variability of the TC activity is sensitive to the convective parameterization.

Appendices

Appendix 1. Robustness of atmospheric general circulation model response using idealized sea surface temperature experiments

In Sect. 4, we suggested that the interbasin effect (i.e., the contribution of tropical Indian Ocean (TIO) and Atlantic) played a key role in the WNPSM cyclonic anomaly in the 1890s–1930s through a specific mechanism (Figs. 5 and 6). To confirm this hypothesis, we conducted additional AGCM experiments using idealized SST anomalies. Figure 11 shows the SST patterns in each AGCM experiment. Approximately 4 K SST reduction was added to monthly SST climatology over TIO in a TIO run (Fig. 11a). Conversely, an SST reduction was added over the tropical Atlantic in the TATL run (Fig. 11d). Moreover, a CLM run was performed as a reference, in which SST was given by the monthly climatology of 1981–2019 HadISST. TIO, TATL, and CLM experiments were AMIP-like run for 20 years, and we used the last 19 years (as a 19-member ensemble).

Fig. 11

a-c SST anomaly and AGCM response in TIO runs; a SST (prescribed in each AGCM experiment) difference from climatology, b-c Simulated z850 (shaded) / uv850 (vector) / rainfall (contour) difference between TIO and CLM runs in JJA (TIO minus CLM); b CTL-TIO (Emanuel convection scheme), and c SP-TIO (Spectral convection scheme). d-f Same as (a-c) but for TATL runs. In vector, larger than 2 m/s is only drawn and thick vector is larger than 5 m/s. In contour, ± 2, 4, 6, 8, 10 mm/d is contoured

Figure 11 also shows the simulated z850 difference between TIO/TATL and CLM in JJA. In TIO-CLM (Fig. 11b and c), cyclonic (negative z850) anomalies appeared over WNP. In TIO, a Rossby-wave-like response was observed as a positive z850 anomaly. This feature—cyclonic anomaly over WNP that appeared in the east of the cold Matsuno-Gill response (Matsuno 1966; Gill 1980)—is consistent with previous studies (Li et al. 2008; **e et al. 2009). In TATL-CLM (Fig. 11e and f), the cyclonic (negative z850) anomaly appeared over WNP. Notably, the cyclonic anomaly over WNP had a significance of > 90%. Similar to the z850 response, the active WNPSM is observed in the rainfall response (Fig. 11). In TIO and TATL runs, a significant (> 90%) rainfall increase was observed over WNP. These results show that the AGCM sensitivity experiments support the hypothesis of the trans-basin impact, and the results explain that the two basins remotely impacted the activeness of WNPSM.

In a recent study, Wang et al. (2021) proposed that the cyclonic/anticyclonic anomaly in the WNPSM is regarded as an intrinsic mode under dynamically unstable conditions of the ASM. To examine whether the summer ASM meets the dynamically unstable condition as suggested by Wang et al. (2021), we computed barotropic conversion at 850 hPa. The barotropic conversion is derived from the equation of eddy kinetic energy (EKE; $(\stackrel{-}{{u^{\prime}}^{2}}+\stackrel{-}{{v^{\prime}}^{2}})/2$) as follows;

$${(\partial }_{t}+\overline{{\varvec{u}} }\bullet \nabla )(\stackrel{-}{{u^{\prime}}^{2}}+\stackrel{-}{{v^{\prime}}^{2}})/2=-\nabla \bullet (\overline{{\varvec{u} }^{\prime}\Phi ^{\prime}})-R(\stackrel{-}{\omega ^{\prime}T^{\prime}})/p+CK$$

(1)

$$CK=\{(\overline{{{v }^{\prime}}^{2}}-\overline{{{u }^{\prime}}^{2}})/2\}({\partial }_{x}\overline{u }-{\partial }_{y}\overline{v })-\stackrel{-}{u^{\prime}v^{\prime}}({\partial }_{y}\overline{u }+{\partial }_{x}\overline{v })$$

(2)

where $\overline{X }$ (X’) is mean (eddy) field of X. The mean field is defined as the JJA mean of CLM, while the eddy field is defined as the JJA mean of TIO/TATL minus CLM. In this equation, CK refers to the barotropic energy conversion term. A positive CK indicates that EKE is being converted from the mean flow. We utilized idealized AGCM runs detailed in Appendix-A. In both TIO-SST and TATL responses (Fig. 12a and b), the energy source of barotropic energy conversion is evident in the southern flank of the WNP anticyclone. This finding supports the notion that the WNP cyclone/anticyclone anomalies are part of the dynamical mode of the ASM basic state.

Fig. 12

Barotropic conversion rate at 850 hPa (× 10^–4 m²/s³). a CTL-TIO run, and b CTL-TATL run. Perturbation fields are calculated from difference of JJA climatology (TIO minus CLM), while background fields are defined as JJA climatology of CLM. Vector shows horizontal wind at 850 hPa. In vector, larger than 2 m/s is only drawn and thick vector is larger than 5 m/s

Appendix 2. Discussion of long-term WNPSM trend

To confirm the effect of trends, Fig. 13a displays detrended WNP z850 variability. It illustrates interdecadal variability. For instance, in the CTL (black lines), positive (negative) phases occur in 1890–99, 1909–18, 1930–41, 1970–77, and 1995–2012 (with negative phases in 1900–08, 1955–69, and 1990–94). The ER case (red lines) exhibits similar variability to CTL after 1920, but significant differences arise in certain periods (e.g., 1891–1900, 1915–20) before 1920. This disparity between CTL and ER underscores the uncertainty associated with observed SST between HadISST and ERSST. The SP case (green lines) differs from CTL and ER in some epochs, indicating sensitivity of AGCM response to different convection schemes. For instance, 1900–05, 1930–39, and 1950–69 significantly differ from CTL and ER. Nonetheless, in the raw WNP z850 variability (without detrending, Fig. 13b), WNP variability can be categorized into three epochs: a negative (cyclonic anomaly) phase in the 1880-1930s, a neutral phase in the 1950-70 s, and an increasing positive (anticyclonic anomaly) trend after the 1980s.

Fig. 13

a Similar to Fig. 1a but detrended WNP z850 variability. b Similar to Fig. 1a but horizontal lines show average of 1890–1934, 1935–1979, and 1980–2016

Figure 14 illustrates the trend of z850, rainfall, and uv850 over the WNP in AGCM experiments. The trend in AGCM response reveals an anticyclonic z850 pattern over the WNP in CTL, SP, and ER runs (Fig. 14a-c). The rainfall pattern also exhibits a decrease (increase) in the south (north) of the anticyclone in CTL and SP runs, but it is less clear in the ER run. Additionally, the reanalysis (grey line in Fig. 1a) similarly indicates a negative (cyclonic anomaly) phase in the 1880-1930s, a neutral phase in the 1950-70 s, and an increasing positive (anticyclonic anomaly) trend after the 1980s.

Fig. 14

a-c Similar to Fig. 2c-2e but for trend. d-f Similar to (a-c) but for difference between 1890–1934 and 1935–1979 (1890–1934 minus 1935–1979). g-i Similar to (a-c) but for difference between 1980–2016 and 1935–1979 (1980–2016 minus 1935–1979). In vector, larger than 0.2 m/s is only drawn and thick vector is larger than 1 m/s. In contour, ± 0.3, 0.6, 0.9, 1.2, 1.5 mm/d is contoured

The difference between 1980–2016 and 1890–1934 resembles the trend pattern (Figure not shown). However, it's worth noting that the z850 anomaly patterns over the WNP do not exhibit a mirror image between 1890–1934 (negative z850) and 1980–2016 (positive z850) from Fig. 14. The positive z850 anomaly in 1980–2016 (WNP subtropical high) appears to be shifted eastward (Fig. 14g-i) compared to the negative anomaly in 1890–1934 (Fig. 14d-f). Additionally, Fig. 13b indicates that the difference between 1890–1934/1935–1979 is clear and robust, whereas the difference between 1935–1979/1980–2016 is unclear and sensitive to each AGCM experiment. Taking these observations into account, we primarily focused on the "interdecadal change in each epoch" rather than the "trend."

Similar to Fig. 6, we attempted to disaggregate AGCM responses to each basin SST forcing. Figure 15 illustrates the z850 difference between 1980–2016 and 1935–1979 in each sensitivity experiment (TROP, TP, TP + TIO runs) and the AGCM response to each basin SST anomaly as differences of these three experiments (referred to as "TIO, TIO + TATL, TATL"). The total response to the tropical forcing (TROP run; Fig. 15a) can be subdivided into forcings over the Pacific (TP run; Fig. 15c), Atlantic (TATL effect; Fig. 15d), and Indian Ocean (TIO effect; Fig. 15e). Different from the 1890–1934 case (Fig. 6), the anticyclonic response over the WNP (around 20–40°N, 150–180°E) is mainly formed by tropical Pacific SST forcing. In the western part (around 10–30°N, 120–150°E), the Atlantic and Indian Ocean SST forcing forms the anticyclonic response, canceling out the strong cyclonic response of the Pacific SST forcing. These results also support the observation that the z850 anomaly patterns over the WNP are not a mirror image between 1890–1934 (negative z850) and 1980–2016 (positive z850).

Fig. 15

Similar to Fig. 6 but for simulated z850 difference between 1980–2016 and 1935–1979