Introduction

The influence of the thermal forcing of the Tibetan Plateau (TP) on the Asian summer monsoon (ASM) has been widely studied but remains controversial since the 2010s. Pioneering studies1,2,3,4,5,6 have indicated that the TP serves as a large heat source in boreal summer because high sensible surface heating could cause cyclonic circulation anomalies in the lower troposphere and could trigger intense vertical motion on the southern slope of the TP with notable latent heat release in the middle-upper troposphere. This phenomenon is referred to as the sensible heat-driven air pump (SHAP). However, a series of studies have also indicated that the elevated heating across the TP, especially above 3 km, is less important than that in the flat Indian monsoon region in summer, and its influence on the ASM is limited7,8,9,10. For example, Boos and Kuang7 revealed that the thermodynamic profiles obtained from twice-daily balloon soundings across the eastern part of the TP indicate that the elevated heating may be largely confined to a local dry boundary layer. Furthermore, Rajagopalan and Molnar10 reported that the correlations of surface heating across the TP with summer monsoon rainfall are insignificant during the main monsoon season. He et al.11,12 recently proposed that the sensible heat flux alone is not enough to explain the surface thermal forcing of the TP, whereas the surface potential vorticity (SPV) of the TP is a more suitable variable for representing the combined impact of the orographic and thermodynamic effects of the TP on the atmosphere. These studies have provided new ways to evaluate the influence of the TP on the ASM.

In numerical simulation studies, diverse monsoon responses have been obtained via the use of different models and experimental designs of the perturbed orography13,14,15,16,17,18,19. For example, Son et al.18 argued that the dynamic effect of the TP accounts for ~65% of the total East Asian summer precipitation, whereas the elevated heating and land–sea heat contrast account for only ~15% of the total summer precipitation, as determined via idealized Aqua–Planet sensitivity experiments. In a series of studies13,14 entailing the use of a fine-resolution model (25~9 km), it has been suggested that methods for modifying the topography and thermal forcing of the TP must be carefully implemented because the simulated precipitation responses are very sensitive to the experimental design. For example, Son et al.18 adopted an albedo that was raised to 0.6 over the TP: although a notable temperature anomaly is absent, vertical diffusion heating still occurs on both the slope and the elevated region of the TP, and the model responses cannot be attributed to dynamical forcing only. Moreover, Aqua–Planet experiments are insufficient for investigating the role of the topography in changes in the ASM, since state-of-the-art models still exhibit notable biases in reproducing the basic ASM pattern under realistic forcings. This has restricted the quantitative estimation of the relative roles of the thermal and dynamical forcing of the TP.

In the above studies, a common approach has been adopted whereby the topography or thermal properties in a given climate model are perturbed, and the climatic mean responses are generally investigated via multiple years of integration between sensitivity and control runs. This approach can be used to qualitatively and statistically understand the influence of the TP but is still not enough to determine the relative importance of TP forcing on ASM variations since these changes involve multiscale variabilities, which differ across years. For example, many precipitation processes of the ASM are related to the activities of intraseasonal oscillations, tropical cyclones, Mei-yu front cloud systems, etc. These activities could greatly vary in different years. Therefore, a new approach is necessary to better understand the influence of the thermodynamic forcing of the TP on the ASM.

Investigation of the effect of TP forcing on the prediction of ASM may provide a new way to understand the role of TP in shaping the monsoon system. From the perspective of sub-seasonal forecasting, the state-of-the-art prediction system shows considerable skill in predictions of the ASM in various aspects20,21. The analysis of subseasonal to seasonal prediction project (S2S) database shows that the predictability of East Asian summer monsoon (EASM) is ~7−14 days, whereas large differences persist within the S2S models20. The cause of the prediction biases have diverse sources. For example, Zhu et al.22 highlighted that the 10–30-day boreal summer intraseasonal oscillation (BSISO2) and its associated moisture convergence is important for improving extended-range forecasting of strong precipitation over south China. Other studies emphasize that the importance of the correct representation of regional and global climate systems are also crucial for improving forecasting skill, such as the positions of the South Asia high and the western Pacific high, tropical intraseasonal variability, El Niño‒Southern Oscillation (ENSO), North Atlantic Oscillation (NAO), and even Anthropogenic forcings23,24,25,26,27. Despite these studies, the role of TP forcing in EASM prediction has been rarely explored in previous literature.

Therefore, in this study, we investigated the impact of the elevated heating and dynamical forcings across the TP on EASM prediction via hindcast experiments for the extended-range scale. In this scale, boundary forcing becomes more important than the initial conditions, and this represents a new perspective to understand when and how the thermodynamic forcing of the TP becomes important for modulating the ASM. Moreover, although there are different perspectives on the relative importance of the thermal and dynamical forcings of the TP, the effect of elevated heating above 3 km is less than that of slope heating at the southern boundary of the TP, whether from a potential vorticity perspective3 or based on moist entropy theory8. However, both observational and simulation studies over the plateau part of the TP, which covers an area of 250 million square kilometers, remain highly challenging. The stations on the TP mainly cover the eastern part, whereas the elevated heat and moisture transport across the entire TP and their influences remain unclear.

Thus, in this article, we investigated the influence of elevated forcing across the TP, which is above 3 km, on a specific 30-day extended-range forecast by nudging heating and winds via reanalysis data during prediction to identify the role of the TP in generating extended-range forecasts for typical years. This study can not only be helpful for understanding the influence of elevated heating across the TP on changes in the EASM but also provide novel insight into how to improve the EASM precipitation prediction skill for an extended range.

The remainder of this paper is structured as follows: we use the SPV forcing to identify typical years and months when the surface forcing across the TP is especially strong or weak. Then, we analyze the thermal and dynamical structures of the surface and associated atmosphere over the TP and monsoon precipitation anomalies as well as the large-scale background of the global sea surface temperature (SST) pattern for these years. Finally, we perform a sensitivity experiment via a 30-day hindcast experiment to assess the sensitivity of precipitation prediction to the model biases of elevated heating, winds and global SST forcing.

Results

Climatological monthly evolution of the SPV across the Tibetan Plateau

The climate mean and interannual variability in the SPV from boreal spring to summer were investigated herein for the first time. Following the method of He et al.12, we calculated the SPV via Eq. (3), and the climatological mean of the SPV across the TP from March to August 1979–2022 is shown in Fig. 1. In March (Fig. 1a), the SPV was positive mainly over the middle and southeastern regions of the TP but was negative over other regions. From April to May (Fig. 1b, c), the SPV pattern was very similar to that in March, except that the intensity of the SPV gradually increased at the center of the TP. From June to August (Fig. 1d–f), almost the entire TP exhibited positive SPV values, except for the western part of the TP, where the SPV close to 77°E was negative. In particular, high positive SPV values occurred mainly in the western and middle parts of the TP in boreal summer (June–July–August), which indicates that the thermodynamic forcing of the TP may be greater and more important in these regions. In regard to the interannual variability (Fig. S1 in OSM), the standard deviation of the SPV exhibited strong signals in March and April and decreased from spring to summer. We were particularly interested in positive SPV values, which is associated with a large standard deviation, because this phenomenon could exert a pumping effect4 on the atmosphere, especially in summer. Thus, large values of the interannual standard deviation of the SPV across the TP are shown via black cross-shading in Fig. 1. Notably, the interannual variability in positive SPV values is especially large in June, which suggests that the thermodynamic forcing of the TP exhibits high variability in June over the past 40 years.

Fig. 1: Climatological monthly mean evolution of the SPV.
figure 1

Climate mean (1979–2022) SPV distribution (PVU, 1 PVU = 10−6 K m2 kg−1 s−1) from March to August. a March, b April, c May, d June, e July, f August. The black cross-shading denotes the standard deviation of the SPV where it is above 5 and where the SPV is positive.

Full size image

The surface sensible heat flux (SSHF) across the TP is recognized as the dominant surface heat source in spring and summer. However, as noted by He et al.12, the sensible heat flux cannot fully represent the thermodynamic forcing of the TP, and the SPV could serve as a better index as it exhibits a better relationship with the ASM than does the sensible heat flux. Here, we compared the spatial distributions of the SPV (Fig. 1) and the SSHF (Fig. S2) to better understand their monthly differences. Notably, from March to June, the spatial patterns of the SPV were very similar to those of the SSHF, with the high-value center located at the middle of the TP. Moreover, the SPV gradually becomes positive over the entire TP from March to June. One of the main differences from March to June is that the SSHF was largely positive across the entire TP, whereas the SPV exhibited considerable negative values in the western and eastern parts of the TP, especially in March. The other difference is that the SSHF indicated lower values in July and August than in June, whereas the SPV showed similar and even greater values in July and August than in June. We also compared the surface heating source Q1 with the SPV in Fig. S3, when changes in Q1 contribute to the change in SPV. The results showed that Q1 was similar to SPV from March to May but exhibited prominent differences from June to August. From this perspective, use of the SPV as an index for thermodynamic forcing of the TP instead of the SSHF and Q1 could provide a new understanding especially during the June-July-August period.

Based on the above analysis, the TP SPV exhibits overall positive signals from June to August. Therefore, to better understand the interannual variability in the TP SPV, we analyzed the climatological monthly mean anomaly (relative to the mean from 1979 to 2022) of the SPV in areas of the TP with a topography higher than 3 km, as shown in Fig. 2. Overall, the time series for all three months indicated a slightly increasing trend from 1979 to 2022, and the interannual variability clearly exhibited a larger amplitude in June (red line) than in July (blue line) and August (August line). We further investigated the concurrent and lead correlations between the TP SPV and East Asian summer precipitation in different months, and the results are provided in Table S1. The results indicated that the concurrent correlation between the SPV and precipitation in June exhibits the highest value of 0.46 compared with that of any other correlation. The lead correlations from June to August clearly decreased, but none passed the significance test. The above results suggest that the TP SPV could exert a greater impact on downstream precipitation in June than in the other two months.

Fig. 2: Interannual variability in the TP SPV anomaly.
figure 2

Time series of the interannual monthly mean TP SPV (PVU) anomalies (relative to the mean over the 1979–2022 period) where the topography is above 3 km for June, July, and August. The purple line denotes the regional mean of TP SPV where the standard deviation of interannual TP SPV is above 5 in June. The red circles denote the typical years of 1983, 1995, and 2022 (the TP SPV in June was especially weak or strong).

Full size image

In particular, when we consider only the TP SPV in instances with a standard deviation above 5 (purple line), the TP SPV is especially strong or weak in some extreme years. For example, in 1995 and 2022, the TP SPV in June showed two strong anomalies, both of which were above 6 PVU. In contrast, in 1983, the TP SPV in June exceeded -6 PVU, which is almost two times lower than that in the other months over the past 40 years. The TP SPV could greatly influence and be related with the surrounding climate system in these extreme years. Therefore, we investigated the local thermodynamics of TP forcing and its influence on downstream air flow and the associated large-scale background in 2022, 1995, and 1983. We focused on the thermodynamic structures over the TP and the changes in the EASM and the global SST to better understand the influence of subdaily perturbations in the thermodynamic forcing of the TP and global SST on extended-range predictions of the EASM in June.

Thermodynamical states over the TP and East Asia in June 2022, 1995 and 1983

In this section, we present the thermal and dynamical features over the TP, the spatial pattern of the EASM, and the associated global SST patterns in June for 2022, 1995, and 1983 to better understand the relationships between different TP SPV extremes and local and remote climate systems. The SPV anomalies in June (relative to the climate mean from 1979 to 2022) of the above three years are shown in Fig. 3. In 2022 (Fig. 3a), there were two positive SPV anomaly regions located in the western and middle to eastern parts of the TP. In contrast, in 1995 (Fig. 3b), the SPV anomaly was highly positive over the middle to northern parts of the TP but weakly positive over the eastern and southern parts of the TP and negative over the western part of the TP. These patterns suggest that although the SPV indicated high anomalies in both years (Fig. 2), the different anomaly patterns may exert different forcing influences on the atmosphere. Finally, in the extremely low SPV case of 1983 (Fig. 3c), the SPV anomaly was highly negative, mainly in the western and northern parts of the TP.

Fig. 3: Climatological SPV and circulation anomalies in June.
figure 3

The climatological anomalies (relative to 1979 to 2022) of a–c the SPV (PVU), d–f relative vorticity (shaded, 10–5 s–1) and near-surface wind (vectors, m s–1) for the years of a, d 2022, b, e 1995, and c, f 1983.

Full size image

Since the SPV is a comprehensive variable that accounts for both thermal and dynamical effects (Eq. (3)), we analyzed the vorticity and surface potential temperature tendency (\(\varDelta \theta\)) to better understand the role of dynamical or thermal condition changes for the contributions of the SPV anomaly in these years. First, the anomalies of the near-surface relative vorticity and wind vectors are shown in Fig. 3d–f. Notably, the vorticity anomalies only exhibited scattered strong values close to the boundary of the TP (the black contour of 3 km) and its northwestern regions in all three years, whereas the wind vectors did not show well-organized systems that could be associated with the formation of positive or negative SPV centers. However, convergence was observed in some areas in different years. In 2022 (Fig. 3d), weak convergence occurred over the western part of the TP, which corresponds with the positive SPV anomaly. In 1995 and 1983 (Fig. 3e, f), clear convergence was observed over the middle and southern parts of the TP. However, the convergence centers did not coincide with the strong positive/negative SPV anomalies.

According to Eq. (3), \(\varDelta \theta\) is an important term related to the control of the SPV from the perspective of thermal changes. The \(\varDelta \theta\) anomalies over the Asian continent in the three years are shown in Fig. S4a–c. The spatial patterns of \(\varDelta \theta\) were very similar to the SPV anomalies shown in Fig. 3a–c for each year, which indicates that the thermal contrast between land and air is the main contributor to the extreme SPV anomaly in these years. Notably, across the whole Asian continent, especially strong or weak \(\varDelta \theta\) anomalies were observed across the TP compared with those in other flat areas. We investigated the skin temperature anomalies in these years (Fig. S4d–f), which are closely connected with \(\varDelta \theta\). We observed skin temperature anomaly patterns over the TP that were similar to those of \(\varDelta \theta\) in each year. Moreover, the skin temperature indicated notable warming over the North Asian continent and East Asian land in 2022 (Fig. S4d), obvious warming over the Indian continent in 1995 (Fig. S4e), and considerable cooling over the North Asian continent with warming over the Indian continent in 1983 (Fig. S4f). These different surface thermal statuses could impact the atmosphere in these years at various degrees.

The surface SPV is closely related to elevated heating and circulation changes over the TP. To better understand the different thermal and dynamical conditions on the TP in the three years, we analyzed the vertical profiles of Q1 and winds in areas of the TP with an elevation above 3 km. The results are in Fig. S5. In regard to diabatic heating (Fig. S5a), the near-surface Q1 values in 2022 and 1995 were notably greater than the climate mean Q1 (black line). Moreover, in 2022, the vertical profile of Q1 was higher than the climate mean throughout the troposphere up to 100 hPa, whereas in 1995, Q1 was only higher than the climate mean below 400 hPa, which occurs in the boundary layers over the TP. These results suggested that the heating anomaly over the TP was greater in 2022 than in 1995, although the intensities of the SPV and Q1 were relatively close near the surface over the TP. In 1983, Q1 was slightly lower than the climate mean at the lower level but showed a cooling effect above 200 hPa in the upper troposphere.

The wind profiles also indicated distinct differences among these years. The zonal winds mainly exhibited notable differences in the westerly jet in the upper troposphere (Fig. S5). Above 200 hPa, the zonal wind exhibited the lowest velocity in 2022 compared with the climate mean and the values in 1995 and 1983, whereas the westerly wind was the strongest in 1983, with a maximum wind speed close to 30 m s–1, which is almost 1.5 times higher than that in 2022. Regarding the meridional winds (Fig. S5), although the wind speeds were lower than those of the zonal winds across the TP, the wind profile still indicated clear vertical shear in the troposphere. Compared with the climate mean meridional wind, the wind profile exhibited stronger southerlies in the upper troposphere in 2022 and 1983. In particular, the vertical shear exhibited a 5 m s–1 difference between the near-surface and 200-hPa layers, which is almost 5 times greater than the climatological value. In 1995, the meridional wind was very weak in the middle and lower troposphere and clearly northerly in the upper troposphere. The above analysis of zonal and meridional wind profiles suggests that Rossby waves may have more easily and more rapidly propagated from the TP to downstream regions in 1983 than in 2022 in the upper troposphere.

The spatial distributions of the monthly mean ASM and its anomalies in June are shown in Fig. 4. The precipitation patterns demonstrate two main rain belts over the tropical and subtropical regions in all three years (Fig. 4a–c). One is the rain belt from the tropical Indian Ocean to the marine continent, and the other is the rain belt from the Bay of Bengal (BOB) and the southern part of the TP to East Asia and the northwestern Pacific, accompanied by westerlies at 850 hPa. The precipitation over East Asia in these years exhibited notable differences. In 2022 (Fig. 4a), the precipitation center over southern China was stronger, whereas that over southern Japan was weaker. However, in 1995 (Fig. 4b), the precipitation center over East Asian land was located mainly in the middle and lower reaches of the Yangtze River and was much weaker than that at the center of southern Japan. In 1983 (Fig. 4c), the precipitation pattern was very similar to that in 1995, except that the center over land was much weaker. Compared with the climatology (Fig. 4d, e), the ASM was stronger over East Asian land, with two positive anomalies over northeastern China and southern China in 2022 (Fig. 4d). Moreover, precipitation exhibited meridional dipole anomalies at low latitudes with strong negative anomalies over the BOB, South China Sea (SCS), and tropical western Pacific, together with easterly wind anomalies, while positive anomalies occurred over the tropical Indian Ocean and southwestern marine continents. Compared with that in 2022, the meridional precipitation anomaly was much weaker in 1995 (Fig. 4e), and the precipitation over land only indicated a strong positive anomaly in South China. In 1983 (Fig. 4f), precipitation mainly exhibited a strong negative anomaly over subtropical regions, similar to that in 2022 (Fig. 4d).

Fig. 4: Spatial pattern of the ASM in June.
figure 4

Monthly mean precipitation (mm day–1) and 850-hPa wind (m s–1) and their anomalies (relative to the mean from 1979–2022) in June for a, d 2022, b, e 1995, and c, f 1983.

Full size image

The ASM is affected by multiscale forcings from nearby local regions to the global climate system, such as influences stemming from the ENSO, which is the largest global teleconnection pattern28,29,30,31,32. The typical years when the TP SPV showed extreme positive/negative anomalies indicated a climate background of strong El Niño episodes in 1983 and strong La Niña episodes in 2022. Global SST anomalies could play an important role in modulating both the thermodynamic forcing of the TP and the ASM. Thus, we examined the spatial patterns of global SST anomalies in June of the three selected years. The results are shown in Fig. 5. In 2022 (Fig. 5a), the strong La Niña event clearly led to a negative SST anomaly over the Eastern Pacific that was largely positive in all the other regions. In particular, the SST greatly increased over tropical marine continents, the Arabian Sea, and the northwestern Pacific. Although no strong ENSO episode occurred in 1995 (Fig. 5b), the SST pattern still indicated a weak La Niña status, with the SST decreasing over the Eastern Pacific. However, the SST in 1995 also decreased in the Indian and Pacific Oceans, which differs from that in 2022 (Fig. 5a). In 1983 (Fig. 5c), a typical El Niño event caused notable warming over the Eastern Pacific, and the SST in the Indian Ocean was relatively high but relatively low over tropical marine continents and the western Pacific.

Fig. 5: Spatial pattern of the SST anomaly in June.
figure 5

Monthly mean SST anomalies (relative to the mean from 1979–2022) in June of a 2022, b 1995, and c 1983.

Full size image

In the above two sections, we identify three years (2022, 1995, and 1983) when the TP SPV exhibits extreme positive/negative anomalies in June. In particular, ENSO backgrounds occur in two years: La Niña in 2022 and El Niño in 1983. Compared with those in 2022 and 1995, although the SPV exhibited a strong positive anomaly over the TP, diabatic heating over the TP greatly differed. Heating generated a notable influence from the bottom of the TP to the upper troposphere in 2022 but was relatively limited and shallow in 1995. The differences over the TP and the global SST could lead to varying impacts on East Asian precipitation prediction. To identify the influence of subdaily thermodynamic perturbations over the TP on extended-range forecasts of East Asian precipitation in June of these years, we performed a series of 30-days hindcast experiments.

Hindcast and sensitivity experiments for East Asian precipitation prediction

As indicated by the experimental design showed in Table 1, we first conducted a series of fully nudged experiments to identify the best-performing model for precipitation simulation in June of 2022, 1995, and 1983, with the experiments denoted as 2022_ALL, 1995_ALL, and 1983_ALL, respectively. In these experiments, the initial fields were derived from a long-term nudging simulation control run, and a time lag method was used to generate 48 ensemble members for each simulation. All the experiments were conducted from 1 June for 30 days of integration. In the ALL experiments, 6 hourly global T, U, V, and TS values from ERA5 data were employed for nudging in the model. Thus, the simulated circulations in the ALL experiments were very similar to those in the ERA5 reanalysis data, with very small biases (Fig. S7). The spatial pattern of the simulated East Asian monsoon in the ALL experiments is shown in Fig. S6 in OSM. In 2022 (Fig. S6a, d), the observed precipitation mainly encompassed three centers, with the strongest center located over South China. The model basically reproduced this pattern, except that the precipitation intensity was higher and that the centers over South China and the western Pacific coincided. The large-scale pattern of the simulated precipitation in 1995 (Fig. S6e) and 1983 (Fig. S6f) was very similar to that of the observations (Fig. S6b, c), but the intensity was overestimated (Fig. S7). The quantitative calculation results for the simulation skill in terms of the pattern correlation coefficient (PCC) and root mean square error (RMSE) are provided in Table 2. Because the winds were directly nudged in the model, the correlations were very high and are not provided in the table. We compared the precipitation simulation skills for both the East Asian region and the global region of the three ALL experiments. Overall, the PCC values, which are all above 0.8, indicated a very high skill over the three-year period in East Asia, and the RMSEs ranged from 2.8~3.5. At the global scale, the PCC reached ~0.8, and the RMSEs ranged from approximately 1.6~1.7. The results revealed that the model could better capture the precipitation distribution in East Asia than at the global scale but achieved a lower skill for capturing the precipitation intensity in East Asia under perfect circulation conditions.

Table 1 Experimental design of the hindcast and associated sensitivity experiments for June in the three typical years of 2022, 1995, and 1983. Each experiment contains 48 ensemble members with the time lag method employed for initialization and is integrated for 30 days from 00Z 1 June
Full size table
Table 2 Pattern correlation and root mean square error of precipitation simulation
Full size table

The above experiments confirmed that our nudging approach was successful and that the model could reproduce the precipitation pattern in June of the three years well if the circulations were correctly predicted. We then performed a series of sensitivity experiments to investigate the influences of subdaily thermodynamic perturbations over the TP and the global SST on 30-day prediction of precipitation in East Asia. As indicated in Table 1, the NO experiment is a free run experiment, which represents the current prediction ability of the model. The SST experiment is a nudging experiment in which only the global SST is nudged. This experiment is designed to better understand the role of the SST in extended-range prediction of East Asian precipitation. The TP_TUV experiment is a nudging experiment in which both the temperature and winds over the TP above 3000 m are nudged. This experiment was designed to investigate the importance of subdaily thermodynamic variations for prediction. Finally, the TP_TA and TP_UV experiments are two types of nudging experiments in which the temperature and winds in the model, respectively, are nudged to better understand their relative contributions and importance for prediction.

We first compared the relative contributions of the TP forcing and the global SST in 2022 (Fig. 6). The mean precipitation in June in the 2022_NO experiment (Fig. 6d) indicated that a clear rain belt extends from South China to the western Pacific. Compared with the observations (Fig. 6a), the predicted mean wind field was relatively close, but the precipitation intensity over East Asia was greatly underestimated. If the model biases mainly originated from the simulation of the SST, nudging the observed global SST throughout the prediction process should then largely correct the precipitation bias. However, in the 2022_SST experiment (Fig. 6b), the simulated rain belt over East Asia shifted east to the western Pacific, and the precipitation intensity remained lower than that in the 2022_NO experiment. Moreover, compared with the observations (Fig. 6a), nudging the SST even induced excess precipitation over the Arabian Sea, BOB, and SCS. We then evaluated the influence of nudging both the temperature and winds over the TP (Fig. 6e). Notably, in the 2022_TP_TUV experiment, the precipitation intensity was higher than that in the 2022_NO experiment, which is closer to the intensity in the observations (Fig. 6a). This result indicated that correctly simulating the thermodynamic states over the TP plays a very important role in predicting the downstream precipitation intensity in 2022. We further separated the influences of thermal and dynamic perturbations by nudging only T, or U, V over the TP in the model and compared the model responses (Fig. 6c, f). The results showed that thermal perturbation played a more important role in determining the overall pattern of the simulated rain belt, which covers South China and the western Pacific, and that dynamical perturbation was important for the occurrence of precipitation over South China. The PCCs and RMSEs of these experiments (Table 2) further supported our conclusion that the thermal forcing over the TP plays a more important role in predicting the EASM in June, with PCC values above 0.7 (2022_TP_TUV, 2022_TP_TA, and 2022_TP_UV), whereas the SST contributes little to EASM prediction in 2022, with a lower PCC value of 0.57. However, the SST is important for predicting the global precipitation pattern, with a PCC value of 0.74 and an RMSE of 1.89, which is better than those of all other sensitivity experiments.

Fig. 6: Hindcast experiments for 2022.
figure 6

June 2022 mean precipitation (mm day–1) and 850-hPa winds (m s–1): a 2022 OBS, b 2022_SST, c 2022_TP_TA, d 2022_NO, e 2022_TP_TUV, and f 2022_TP_UV.

Full size image

The observed precipitation pattern over East Asian land in 1995 (Fig. 7a) slightly differed from that in 2022. Notably, there were two precipitation centers. One was stronger and occurred in the middle and lower reaches of the Yangtze River, whereas the other was weaker and occurred over southern China. The model predictions (1995_NO, Fig. 7d) captured the precipitation center over southern China. However, the model failed to reproduce the strong center in the middle and lower reaches of the Yangtze River and seriously underestimated the precipitation intensity over the western Pacific, which is close to southern Japan. The experiments in which the SST was nudged (Fig. 7b) yielded very similar precipitation patterns to those in the 1995_NO experiment (Fig. 7d), which indicates that SST correction slightly influences the improvement in extended-range prediction of precipitation in East Asia in that year. Regarding the influence of thermodynamic forcing over the TP (Fig. 7e), the 1995_TP_TUV experiment revealed that TP forcing correction could lead to a more realistic precipitation pattern in extended-range prediction. The precipitation center in the middle and lower reaches of the Yangtze River could be captured, although the intensity was slightly lower. The precipitation intensity over southern China and southern Japan was also increased. However, when we nudged only T (1995_TP_TA, Fig. 7c) or U and V (1995_TP_UV, Fig. 7f) in the model, both simulations captured the two centers over southern China and southern Japan but failed to reproduce the precipitation center in the middle and lower reaches of the Yangtze River. These results indicated that the integrated perturbations of subdaily thermal and dynamical forcings over the TP are important for predicting precipitation in the middle and lower reaches of the Yangtze River in 1995. The calculated PCC and RMSE values (Table 2) indicated a lower skill in almost all the experiments for predicting East Asian precipitation in 1995 than in 2022, whereas only the skill obtained via nudging the SST was greater in 1995 than in 2022.

Fig. 7: Hindcast experiments for 1995.
figure 7

June 1995 mean precipitation (mm day–1) and 850-hPa winds (m s–1): a 1995 OBS, b 1995_SST, c 1995_TP_TA, d 1995_NO, e 1995_TP_TUV, and f 1995_TP_UV.

Full size image

The above two cases highlight the importance of TP forcing for extended-range prediction of the EASM, although the climate backgrounds are different: in both cases, the TP SPV exhibits strong positive anomalies, whereas diabatic heating over the TP is strong and deep in 2022 but relatively shallow and weak in 1995. The SST suggested a strong La Niña pattern in 2022 but suggested neutral conditions in 1995. However, in 1983, the climate background greatly differed from that in the above two years, with a negative pattern of the TP SPV anomaly and an El Niño type of the global SST pattern. Thus, the EASM responses to these forcings could also differ. The monsoon responses in 1983 are shown in Fig. 8. The observed precipitation intensities (Fig. 8a) over South Asia and East Asia were lower than those in 2022 (Fig. 6a) and 1995 (Fig. 7a). In regard to precipitation over East Asian land, the pattern in 1983 was relatively similar to that in 1995, with two distinct centers located over Central China and South China. However, in the predictions (1983_NO, Fig. 8d), the model still underestimated the precipitation intensity over southern China. When we nudged the global SST in the simulation (1983_SST, Fig. 8b), the precipitation center in the middle and lower reaches of the Yangtze River could be captured, although the intensity was slightly lower, which indicates that the SST plays a more important role in predicting precipitation in 1983 than in the other two years. The experiment in which the thermodynamic forcing of the TP was nudged (1983_TP_TUV, Fig. 8e) yielded a similar precipitation pattern to that obtained in the 1983_SST experiment, but the intensity was higher. A comparison of the individual effects of thermal (1983_TP_TA, Fig. 8c) and dynamic perturbations (1983_TP_UV, Fig. 8f) over the TP revealed that the thermal effect of the TP is more important for the model to predict the precipitation center in the middle and lower reaches of the Yangtze River, whereas the nudging of subdaily winds over the TP was more important for the model to predict the precipitation center over South China. The calculated PCC and RMSE values for precipitation prediction in the sensitivity experiments (Table 2) all revealed greater skill in 1983 than in the other two years, which suggests that in typical El Niño years, the monsoon response may be weaker and easier to predict than usual.

Fig. 8: Hindcast experiments for 1983.
figure 8

June 1983 mean precipitation (mm day–1) and 850-hPa winds (m s–1): a 1983 OBS, b 1983_SST, c 1983_TP_TA, d 1983_NO, e 1983_TP_TUV, and f 1983_TP_UV.

Full size image

The monthly mean precipitation responses for the three years showed varying prediction skills and regional differences. To better understand the relative contributions of different forcings and the reasons underlying the obtained prediction skills, it is necessary to investigate the daily evolution of circulation and precipitation in these experiments. The 500-hPa geopotential height is one of the most important variables that can serve as an index for evaluating the prediction of weather systems over East Asia. Usually, precipitation occurs in front of the trough at 500 hPa. Thus, we calculated the anomaly correlation coefficients (ACCs) of the daily zonal eddies at the 500-hPa geopotential height over East Asia (20°N–50°N, 105°E–140°E) in each experiment to better understand the prediction skill for the daily evolution of circulations, as shown in Fig. 9. The red dashed line denotes the 14-day forecasts, whereas the black dashed line denotes the threshold of skillful prediction with an ACC score of 0.6. In the case of 2022 (Fig. 9a), all the simulations showed similar capabilities from forecast days 1 to 14, while large differences occurred after the 14-day forecast period for the different hindcast experiments. This result indicates that the initial field greatly impacted precipitation over East Asia in 2022, whereas the influences of the SST and TP forcings were delayed for 14 days. Notably, there were still two stages in which the skill clearly decreased on forecast days 3 and 11, which suggests that factors other than the forcings analyzed herein control the prediction outcomes. After the 14-day forecast period, we found that the 2022_TP_TUV and 2022_TP_TA experiments achieved better skills than the other experiments throughout the integration process except from 21–23 days. The results emphasized that the thermal forcing of the TP is crucial for extended-range prediction of East Asian circulation, especially from weeks 2 to 4, at least in 2022.

Fig. 9: ACC for 500-hPa geopotential height prediction.
figure 9

Time series of the ACC of the regional mean (relative to 20°N–50°N, 105°E–140°E) geopotential height zonal eddies at 500 hPa for a 2022, b 1995 and c 1983. The black dashed line denotes a skill of 0.6, and the red dashed line denotes the day of the 14th day of prediction.

Full size image

In the case of 1995 (Fig. 9b), the prediction skills of the NO and sensitivity experiments were very close before day 9 of the forecast period and deviated thereafter. This time point is slightly earlier than that in 2022. Notably, this difference may have been caused by stronger westerlies in 1995 than in 2022 (Fig. S5b), and synoptic-scale waves from the TP could have propagated faster toward East Asia. Moreover, we observed that all the prediction experiments failed to reproduce the circulation pattern on day 7, which indicates that other important forcings were not considered in this study. The relative importance of the TP forcing and SST for prediction mainly demonstrated clear differences after day 10. In contrast to that for 2022, the prediction skill of all the experiments remained relatively high (above 0.6) throughout the prediction period after day 10 in 1995. The TP_TA experiment exhibited the highest skill, especially from days 20 to 27, whereas the TP_TUV experiment exhibited an extremely poor skill after day 25. The above results indicated that the subseasonal variation in the daily circulation over East Asia in 1995 could be effectively predicted by the model, while the influence of the thermodynamic forcing of the TP on circulation prediction was comparable to the influence of the global SST.

Finally, in the case of 1983 (Fig. 9c), the skill of all the sensitivity experiments decreased sharply on day 3, while the TP_TA experiment exhibited the highest skill, with a value close to 0.8 on this day. Moreover, the other experiments all exhibited a lower skill of 0.82 for the prediction of the first day, which indicates that the impact of the local synoptic process is important for the prediction in 1983. In contrast to 2022 and 1995, the prediction skills of all the sensitivity experiments did not differ substantially after day 3 until days 21 to 25. These results suggest that the skill for deterministic forecasts of East Asian circulation depends on various local and remote forcings. In any case, for predictions longer than 14 days, the TP_TUV experiment exhibited the highest skill from days 16 to 25, followed by the other experiments, which indicates that the influence of thermodynamic forcing over the TP remains important for predicting the circulation over East Asia in that year.

Along with circulation prediction, we examined the daily evolution of the predicted precipitation, and the results are shown in Fig. 10. The time series of the observed regional mean precipitation (20°N–50°N, 105°E–140°E) in June 2022 (Fig. 10a) revealed that 5 precipitation peaks occurred, whereas the ALL experiment, which was forced by ERA5 circulation data, captured almost all the precipitation peaks except for intensity overestimation on day 5 and intensity underestimation on day 10 and days 16 to 20, with a correlation value of 0.65. Both the original prediction (NO) and nudging SST (SST) experiments failed to reproduce the variability of the precipitation, with correlation values of –0.41 and 0.32, respectively. The primary reason is that both of these prediction experiments underestimated the precipitation intensity through June. In contrast, both the TP_TUV and TP_TA experiments exhibited relatively high correlations, while the TP_UV experiment exhibited a low skill. The results indicated that the thermal forcing of the TP is very important for predicting precipitation over East Asia in June 2022.

Fig. 10: Time series of precipitation predictions.
figure 10

Time series of the observed and simulated regional means (20°N–50°N, 105°E–140°E) of the daily precipitation (mm day–1) in June of a 2022, b 1995, and c 1983; the values on the right of the experimental name denote the correlations between the experimental data and the observations.

Full size image

In the case of 1995 (Fig. 10b), the NO experiment exhibited a low skill for capturing the precipitation variability, with a correlation value of –0.1. Moreover, both the TP_TUV and TP_TA experiments exhibited a very low precipitation prediction skill. In contrast, the TP_UV and SST experiments yielded relatively high correlation values of 0.31 and 0.24, respectively. Almost all the experiments failed to reproduce the precipitation variation before day 15 and achieved an improved prediction skill after day 15. In this case, the influence of the thermal forcing of the TP was smaller than that of the SST and dynamic perturbations on predicting precipitation in East Asia. This is possibly related to the fact that the heating layers over the TP were shallower in 1995 than they were in 2022 (Fig. S5 in OSM), and the thermal perturbations were not easily propagated downstream in 1995.

Finally, in the case of 1983 (Fig. 10c), when the SPV over the TP was negative with an El Niño background, both the NO and SST experiments exhibited low prediction skills, with values of 0.27 and 0.22, respectively. The correlation values of the TP_TUV, TP_TA, and TP_UV experiments were very close, but the TP_UV experiment exhibited the highest skill, with a value of 0.53. In this case, the prediction of East Asian precipitation was simple. All the correlations were positive, and most of the experiments could capture the precipitation variation except for overestimating the precipitation before day 5 and underestimating the precipitation peak after day 20. These results indicated that both thermodynamic forcing over the TP and the global SST played important roles in accurately predicting precipitation in East Asia in 1983.

We also summarize the experimental results in Table 3 to compare the relative role of the key factors, namely SST and TP thermodynamic forcing, on the extended-range prediction of East Asian precipitation. The results clearly show that, in 2022 alone, when TP SPV was strongly positive and developed vertically with strong overhead heating, the prediction of TP thermal forcing was especially important compared with all the other factors. It should be noted that although the nudging of SST was not effective for extended-range forecasting during this year with a La Niña background, SST might be important for shaping the extreme warmth over the TP this year over a longer timescale.

Table 3 Brief summary of the role of key factors in the extended-range forecast of East Asian precipitation for the specific three cases in this study
Full size table

Discussion

In this study, the influences of subdaily thermodynamic forcing over the TP on 30-day forecasts in early summer for three specifical years (2022, 1995, and 1983) were analyzed. In these years, the global SST anomalies exhibited typical patterns of La Niña, neutral, and El Niño events. Furthermore, the SPV over the TP was extremely positive (2022 and 1995) or negative (1983), which indicates that the surface thermodynamic forcing of the TP was especially strong or weak in these years. The hindcast experimental results indicated that the subdaily thermal perturbation over the TP from the surface to the troposphere aloft is very important for East Asian precipitation forecasts, especially after day 14 of the prediction period in 2022. In the other cases, the individual influences of thermal perturbations over the TP were not important for the prediction skill, which indicates that extended-range prediction of the EASM involves complex factors. In particular, the relative contributions of surface forcings such as the dynamical or thermal perturbations over the TP on EASM prediction could vary for deterministic forecasts of individual years.

To gain a physical image to elucidate why TP SPV forcing in June 2022 had a significant impact on the East Asian precipitation prediction, we additionally show the simulations of the 500 hPa vertical velocity and water vapor transport in Fig. S8 and Fig. S9 in OSM. By synthetic analysis of all the experimental results, we were able to generate a schematic diagram (shown in Fig. 11) that revealed the physical process. The correct simulation of TP SPV and associated overhead heating process could improve the accuracy of the overhead simulation of upper-level synoptic waves (Fig. 9). Moreover, the pumping effect of TP, which resulted from the strong SPV forcing, led to the correct simulation of lower-level water vapor transport in East Asian land. Both processes contributed to the correct prediction of strong vertical ascending motion and associated precipitation processes.

Fig. 11: Schematic diagram of how the correct TP SPV simulation improved the downstream precipitation prediction at an extended-range in 2022.
figure 11

The correct simulation of TP SPV and associated overhead heating in June 2022 improves the prediction of low-level water vapor transport and of middle- and upper-level synoptic waves, both of which favor the correct prediction of strong ascending motions related to strong precipitation.

Full size image

The results of this study also indicated that to improve the extended-range forecast skill for the EASM, improving the physical scheme of vertical heat transport over the TP in the numerical model would be effective for certain deterministic forecasts. The increase in the horizontal and vertical resolutions of the model and heat and water vapor transport over complex topography could also improve the overall simulation performance and the prediction of the Asian monsoon system. Moreover, we examined only 30-day forecasts in early summer. For longer time scales, such as subseasonal forecasts up to 90 days, the influence of thermal and dynamical forcings of the TP and other external forcings remain unclear. Previous studies have shown that the correct simulation of slowly varying components, such as ENSO and Indian Dipole Mode, is important for EASM prediction over a longer time scale33,34. Additionally, only the specific three years are investigated in this study. For the other years, the relative importance of TP forcing on the prediction remains unclear, although the TP SPV may not be strong. Therefore, further comprehensive investigations using diagnostic and numerical simulation must be implemented to better understand the diverse predictability of the EASM and the relative role of the thermodynamic forcing of the TP.

The results of this study also demonstrated the importance of enhancing observation studies in the plateau part of the TP, especially targeting vertical heat and energy transport in the western part of the TP, where observation sites are currently lacking35,36,37,38,39,40,41. The observation of heat and energy transport in the boundary layer over the western part of the TP could help us better understand the vertical heating status over different underlying surfaces of the TP and could benefit not only the understanding of the diurnal cycles of heat and momentum fluxes but also the development of boundary layer parameterizations in numerical models, which could improve the overall model performance and even promote both subseasonal prediction and climate modeling.

Materials and methods

Observational and reanalysis datasets

The following observational and reanalysis datasets are used in this study:

  1. (i)

    Monthly and daily mean precipitation datasets from the Global Precipitation Climatology Project (GPCP)42. The monthly datasets cover the period from 1979 to 2023 with a horizontal resolution of 2.5° × 2.5°, whereas the daily datasets cover the period from 1996 to 2023 with a horizontal resolution of 1.0° × 1.0°.

  2. (ii)

    Daily precipitation data from the Asian Precipitation–Highly Resolved Observational Data Integration Toward Evaluation of Water Resources (APHRODITE) project are also employed. This product covers the period from 1951 to 2015 and exhibits a horizontal resolution of 0.25° × 0.25° over the Asian monsoon region. The data are obtained by interpolating rain gauge observations obtained from meteorological and hydrological stations throughout the region43.

  3. (iii)

    The National Oceanic and Atmospheric Administration (NOAA) Daily Optimum Interpolation Sea Surface Temperature (OISST) was analyzed in this study. The dataset covers the period from 1981 to 2023 and exhibits a horizontal resolution of 0.25° × 0.25°44.

  4. (iv)

    The monthly and daily mean ECMWF Reanalysis v5 (ERA5) datasets from the European Center for Medium-Range Weather Forecasts (ECMWF)45 were used to analyze the multilevel air temperature, specific humidity, wind field, and surface variables such as the surface pressure and surface temperature. Six-hourly variables such as the multilevel air temperature, specific humidity, and wind field from ERA5 data were nudged in the climate system model to generate the initial field for the hindcast experiments. The same variables in the model were nudged in the sensitivity experiments with different aims.

Statistical methods

Pearson’s correlation analysis is a widely used statistical method to quantitatively measure the linear relationship between two variables with the same dimensions. The correlation coefficient can be calculated via Eq. (1), where r denotes the correlation coefficient, Xi denotes each value of the model variable in this study,\(\overline{X}\) denotes the mean model value, Yi denotes each value of the observed variable, \(\overline{Y}\) denotes the mean observation value, and n denotes the number of samples. Thus, the correlation coefficient is the quotient of the covariance and the standard deviation between the two variables. In this study, Pearson’s correlation coefficient was used to measure the simulation skill for the temporal evolution of the 500-hPa geopotential height zonal anomalies obtained in the hindcast experiments, as shown in Fig. 10, and to measure the variability in the simulated regional mean precipitation, as shown in Fig. 11. The correlation coefficient was also used to analyze the simulation skill for the monthly mean precipitation pattern of each experiment, as detailed in Table 2.

$${\rm{r}}=\frac{\mathop{\sum }\limits_{i=1}^{n}(Xi-\overline{X})(Yi-\overline{Y})}{\sqrt{\mathop{\sum }\limits_{i=1}^{n}(Xi-\overline{X}{)}^{2}}\sqrt{\mathop{\sum }\limits_{i=1}^{n}{(Yi-\overline{Y})}^{2}}}$$
(1)

The RMSE is widely employed to estimate the difference between the predicted and true values. The RMSE can be calculated via Eq. (2), where Xi denotes the value of each model variable, Yi denotes the value of each observed variable, and n denotes the number of variables. The RMSE was used to estimate the simulation ability for the regional mean precipitation of each sensitivity experiment, as indicated in Table 2.

$$RMSE=\sqrt{\frac{\mathop{\sum }\limits_{i=1}^{n}{[f(xi)-Yi]}^{2}}{n}}$$
(2)

Surface potential vorticity

The potential vorticity (PV) is a conserved variable under frictionless and adiabat motion. The PV system is widely used for analyzing the dynamics of atmospheric circulation. To better understand the thermodynamic forcing of the TP on the atmosphere, He et al.11,12 proposed a new index based on the SPV to quantitively measure the seasonal and interannual variations in the surface thermodynamic forcing of the TP via reanalysis data or climate model simulation. On the basis of their definition, the SPV in this study (\(P{V}_{\sigma }^{\ast }\)) can be calculated via Eq. (3).

$$P{V}_{\sigma }^{\ast }=-P{V}_{\sigma }=-\left[-\frac{g}{ps}(f+{\zeta }_{\sigma })\frac{\varDelta \theta }{\varDelta \sigma }\right]$$
(3)

where \(\sigma\) denotes the bottom level of the atmospheric model above the Earth’s surface in terrain-following coordinates; \({\zeta }_{\sigma }\) is the relative vorticity at the \(\sigma\) level; \(\varDelta \theta\) is the vertical surface potential temperature tendency at the bottom level of the atmospheric model; \(\varDelta \sigma\) is the \(\sigma\) difference between the Earth’s surface and the bottom level of the atmospheric model, which is ~0.996; and ps is the surface pressure. The detailed derivation can be found in He et al.12.

Apparent heat source Q 1

Diabatic heating Q and the apparent heat source Q1 in the atmosphere can be estimated following the methods of Yanai et al.46, as expressed in Eq. (4):

$$Q=\frac{{Q}_{1}}{{c}_{p}}={\left(\frac{p}{{p}_{0}}\right)}^{\kappa }\left(\frac{\partial \theta }{\partial t}+\overrightarrow{V}\cdot \nabla \theta +\omega \frac{\partial \theta }{\partial p}\right)$$
(4)

where \({c}_{p}\) is the specific heat of air at a constant pressure, \({p}_{0}\) is 1000 hPa, \(\kappa =R/{c}_{p}\), and \(\omega\) is the vertical velocity in the p-coordinate system. The heating profile of Q was determined in this study to obtain the atmospheric diabatic heating status over the TP in early boreal summer.

FGOALS-f2 climate system model

The finite-volume Flexible Global Ocean-Atmosphere‒Land System model version 2 (FGOALS-f2), developed by the State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics (LASG), Institute of Atmospheric Physics (IAP) of the Chinese Academy of Sciences (CAS)47, was used in this study for the hindcast experiments. The FGOALS-F2 model comprises five components: version 2.2 of the finite-volume atmospheric model (FAMIL)16,48,49; version 2 of the Parallel Ocean Program (POP2)50; version 4.0 of the Community Land Model (CLM4)51; version 4 of the Los Alamos sea ice model (CICE4)52; and version 7 of the coupled module from the National Center for Atmospheric Research (NCAR). The nominal horizontal resolution of the FGOALS-f2 model is 1° × 1°, and the number of vertical layers of the atmospheric model is 32, expressed in hybrid coordinates. The top of the model occurs at 2.16 hPa. In this study, we used the FGOALS-f2 model to perform hindcast experiments to explain the role of subdaily perturbations over the TP in downstream circulation and precipitation changes.

Experimental design

To better understand the effects of the thermal and dynamical perturbations over the TP on 1- to 30-day extended-range forecasts for the typical years of 2022, 1995, and 1983, as analyzed via observation diagnostics, we conducted a series of hindcast experiments to identify the roles of the subdaily air temperature, winds over the TP and global SST in predicting precipitation over East Asia. The main differences between the experiments are summarized in Table 1.

First, a control run was performed with a fully coupled run by using the FGOALS-f2 model, which is integrated from 1977 to 2022, and the 6-hourly multilevel air temperature, winds, and global surface temperature were nudged via ERA5 reanalysis data. The restart files of 00Z 1 June for 1983, 1995 and 2022 were saved to produce the initial field of the ensemble members for the individual forecasts for these years. Second, to apply the time lag method for generating the ensemble members for the hindcast experiments, we operated the model freely without nudging for two days from 00Z 1 June for 1983, 1995, and 2022 and saved the hourly restart files. Thus, we generated 48 initial fields for each of the ensemble members, as shown in the second column of Table 2. Third, a 30-day integration was performed under different nudging options to determine the influences of the TP and global SST on extended-range forecasting of East Asian precipitation in June of 1983, 1995, and 2022.

Here, we introduce the experimental settings for 2022 as an example, and the experimental settings are similar for the other two years. The first experimental group, namely, 2022_ALL, encompasses fully coupled runs that are integrated for 30 days while nudging the global air temperature, winds and surface temperature throughout the experiment. As documented above, 48 ensemble members were generated for the integrations from 00Z 1 June in 2022 in this group. This experimental group aimed to identify the best-performing in reproducing the evolution of precipitation and winds in June compared with the ERA5 data. The second experimental group, namely, 2022_NO, comprises fully coupled runs that are integrated for 30 days without nudging. This experimental group is the hindcast experiment, which represents the current forecast skill of the model. The third experimental group, namely, 2022_SST, comprises fully coupled runs in which only the global SST is nudged throughout the integration process. This experimental group aimed to investigate the role of the global SST in generating extended-range predictions of the development of the EASM in June. The fourth experimental group, namely, 2022_TP_TUV, encompasses fully coupled runs in which the air temperature and winds over areas of the TP where the topography remains above 3 km are nudged throughout the integration process. This experimental group aimed to investigate the effect of the thermal and dynamical subdaily variabilities over the TP on downstream extended-range predictions of the EASM in June. The fifth experimental group, namely, 2022_TP_TA, comprises fully coupled runs in which only the air temperature over areas of the TP where the topography is above 3 km is nudged throughout the integration. The sixth experimental group, namely, 2022_TP_UV, is the same as the 2022_TP_TA experimental group but only winds over the TP are nudged. These two experimental groups were used to determine the influence and relative role of thermal and dynamical perturbations over the TP in predicting the EASM in June for these years.