East Asian Monsoon Forcing And North Atlantic Subtropical High Modulation of Summer Great Plains Low-Level jet

Dynamic inuences on summertime seasonal United States rainfall variability are not well understood. A major cause of moisture transport is the Great Plains low-level jet (LLJ). Using observations and a dry atmospheric general circulation model, this study explored the distinct and combined impacts of two prominent atmospheric teleconnections – the East Asian monsoon (EAM) and North Atlantic subtropical high (NASH) – on the Great Plains LLJ in the summer. Separately, a strong EAM and strong western NASH are linked to a strengthened LLJ and positive rainfall anomalies in the Plains/Midwest. Overall, NASH variability is more important for considering the LLJ impacts, but strong EAM events amplify western NASH-related Great Plains LLJ strengthening and associated rainfall signals. This occurs when the EAM-forced Rossby wave pattern over North America constructively interferes with low-level wind eld, providing upper-level support for the LLJ and increasing mid- to upper-level divergence.


Introduction
Continental United States (CONUS) summer rainfall variability remains an important eld of research due to its implications for human health and the economy. Unfortunately, current subseasonal-to-seasonal forecasts for summer precipitation have relatively low skill (Becker et  greater Great Plains moisture transport compared to eastern NASH events. The NASH has shifted or extended west more frequently in recent decades. It is projected that trend will continue in a warming climate (Li et  Understanding the primary forcing mechanisms for these planetary waves, such as the EAM, and how they develop over North America with in uence from the NASH circulation, is essential. Therefore, this study will concentrate on atmospheric teleconnections active in the June-July-August (JJA) season, particularly the EAM's and NASH's relationship with the Great Plains LLJ.
Despite the considerable literature on the EAM and NASH and their distinct in uence on CONUS rainfall variability, there is little to no exploration into how these teleconnections interact. Because the Great Plains LLJ is a key driver of summer precipitation, this study will investigate the Great Plains LLJ response to the EAM forcing and consider how the NASH modulates that response. Simple dry atmospheric general circulation models (AGCMs) have been successful in reproducing the dynamics and variability of quasi-stationary/planetary wave activity from diabatic heating related to monsoons (Zhu and Li 2016; Zhu and Li 2018; Lopez et al. 2019; Malloy and Kirtman 2020, submitted to Climate Dynamics). We will use a simple dry nonlinear AGCM to understand the large-scale responses and modulation of the Great Plains LLJ on seasonal-to-interannual timescales. Section 2 will describe the datasets, details of the nonlinear AGCM and the experiments, and the relevant analysis methods. Section 3 will present the results as follows: The observed responses of the EAM and NASH and their interactions will be quanti ed. Then, this paper will examine the AGCM's EAM-forced response of the Great Plains LLJ. Finally, we will evaluate how NASH modulates the EAM-forced response. Section 4 will serve as a summary and re ection of the results in the context of previous literature and future work needed.

Data And Methods a. Observational dataset
Pressure-level meridional wind, zonal wind, temperature, and geopotential height were taken from the European Centre for Medium-Range Weather Forecasts (ECMWF) fth-generation reanalysis (ERA5). ERA5 atmospheric data is provided on a 0.25° latitude/longitude grid (Hersbach et al. 2020). U.S. precipitation data were taken from the CPC Uni ed Gauge-based Analysis, provided on a 0.25°l atitude/longitude grid (Chen et al. 2008;). This study used the June through August monthly data between 1979-2019 to serve as observations.

b. Model and Experiments
The model in this study is a dry, baroclinic, and nonlinear AGCM, i.e. it includes the full primitive equations of divergence, vorticity, temperature and surface pressure. It is a spectral model with Rhomboidal truncation at R42 -approximately 1.7° latitude by 2.8° longitude -with 26 vertical levels. The vertical levels are analogous to the Community Atmospheric Model, version 4 (CAM4), which uses hybrid sigma-pressure coordinate system. The AGCM is adapted from Brenner et al. (1984) to remove moist processes. Newtonian cooling is speci ed throughout the troposphere with enhanced damping near the surface. Rayleigh friction is speci ed at the lower levels and mimic realistic land-sea frictional contrasts to generate climatological features, such as the NASH, monsoonal systems, and the Great Plains LLJ. Realistic topography is also an important aspect of this model as the large-scale Great Plains LLJ requires topographical modulation of stationary ow ( The surface temperature climatology for JJA is input as background state for the model. This climatology was calculated from ERA5 data and interpolated to the model's grid. Each experiment was integrated forward for 900 days with the JJA background state to estimate the steady-state response for both seasonal and interannual analysis. Analysis excludes the rst 100 days to assure that there is no contamination from the spin-up period.
This AGCM is used for both unforced and forced experiments. The unforced experiment, or control (hereby CTRL) run, is evaluated to compare climatology with observations. It is also compared to the EAM-forced runs to understand NASH modulation of the Great Plains LLJ, divergence, and circulation response in the model. The strong EAM experiment applies a constant diabatic heating via Gaussian bubble with a maximum of 2 K day −1 centered at 30°N, 120°E and 300 hPa (see Supp.

c. Analysis Methods
To investigate the separate and combined roles of the EAM and NASH in both observations and the AGCM, we calculated difference composites of 900-hPa meridional wind (V900) anomalies, 250-hPa geopotential height (Z250) anomalies, and rainfall anomalies. This means that anomalies are averaged for upper tercile events, and then subtracted from anomalies averaged from lower tercile events. We chose a composite analysis to highlight any nonlinearities in responses as weak and strong events may not yield equal and opposite LLJ anomalies. The EAM index is de ned by 200-hPa zonal wind (U200) The western NASH index is de ned as follows: Z850(15-28°N, 50-85°W). A variation of this intensity index was used by Li et al. (2012) and Ferreira and Rickenbach (2020) in evaluating Z850 anomaly elds associated with the Great Plains LLJ strengthening, but this index highlights northern NASH variability, which impacts Plains/Midwest rainfall variability to a greater extent. Overall, the index distinguishes between strong western NASH events, with the western ridge over North America, and weak western NASH events, with the western ridge remaining over the Atlantic. All indices are standardized before anomalies are composited. We also composited the 1560 geopotential meter (1560-gpm) lines for observations corresponding to the strong and weak events to signify the NASH extent (Li et al. 2011) for the samples.
In addition, these composites are organized by a secondary condition, e.g. western NASH-related anomalies are further differentiated by strong (upper tercile) or weak (lower tercile) EAM events before averaging. To assess the signi cance of these difference composites, we performed a two-sided Wilcoxon rank-sum test. This test is preferred because it does not assume a Gaussian distribution, but it compares two samples' population mean ranks by considering if their distributions are the same.
To understand potential processes associated with these difference composites in observations and CTRL experiment, we included composites of V anomaly pro les at 30°N, the latitude where the Great Plains LLJ and its related V900 anomalies are located. These composites separate by weak/strong western NASH and weak/strong EAM. This aids in visualizing the interactions in the vertical.
Finally, we assessed NASH's in uence on the EAM-forced responses in the dry AGCM using difference of the composites, i.e. strong -weak EAM response during strong western NASH events minus strongweak EAM response during weak western NASH events. This determines whether the dry AGCM can simulate the correct tendency of the response by NASH modulation. Anomalies are calculated by subtracting the climatology from the CTRL experiment, and weak/strong western NASH events are based on the lower/upper quintile thresholds calculated from the CTRL experiment. In contrast, during a strong EAM, the Great Plains LLJ strengthening is greater (>2 m s −1 ) and penetrates further into the U.S. (bottom left). This is related to more extreme wet anomalies (>2 mm day −1 ) stretching from the Plains to the Northeast U.S. The NASH-related Z250 anomalies are different between weak and strong EAM events (middle, bottom left), particularly over East Asia, North America, and the North Atlantic. The north-south orientation of anomalous trough-ridge pattern over CONUS during strong EAM events signals a negative PNA and enhanced meridional transport (Harding and Snyder 2015; Malloy and Kirtman 2020).
We considered the reverse analysis as well by taking strong -weak EAM difference composites of V900 anomalies, further separated into weak or strong western NASH events (Fig. 2). A strengthened EAM is associated with a ~0.5 m s −1 strengthening of the Great Plains LLJ (top left), though is further east from the Rockies than the climatological Great Plains LLJ location and the NASH-related LLJ strengthening. To further understand processes between the strong and weak events, Fig. 3 shows the vertical pro le of V anomalies at 30°N. Rows differentiate between NASH strength, and columns differentiate between EAM strength. The EAM-related ow can be discerned east of the Himalayas (100-120°E) by the northerlies (left column) or southerlies (right column), which signals whether there is low-level divergence or convergence over the EAM region, respectively. The Great Plains LLJ is found between -100 and -90°W, with northerlies coinciding with a weak western NASH (top row) and southerlies coinciding with a strong western NASH (bottom row). During weak EAM and weak western NASH events (top left), as well as during strong EAM and strong western NASH events (bottom right), the LLJ-related winds are of the same sign as the upper-level ow. This suggests that when the EAM and western NASH are both weak or strong, their related circulation patterns are in constructive interference, i.e. the low-and upper-level ow are in alignment to promote enhanced precipitation patterns. While this mechanism is supported by previous literature (Harding and Snyder 2015; Mallakpour and Villarini 2016), we want to explore whether a simple dry AGCM can reproduce this interference between the EAM, NASH, and Great Plains LLJ.

b. Control Experiment Climatology and Biases
Before analyzing the dry AGCM responses, we evaluated the climatological biases of the model and its ability to produce realistic dynamic responses (e.g. quasi-stationary Rossby waves). Zonally-asymmetric components (represented by * ) of time-mean circulation (represented by − ) -also known as stationary waves -are useful for understanding the production and maintenance of Rossby waves. Seasonally, stationary waves describe preferred locations of meridional uxes of heat and moisture, affecting hydroclimate. We compared the stationary waves in ERA5 and the CTRL experiment (no heating forcing) from the dry AGCM (Fig. 4). In observations, . These ridges generally appear in the CTRL experiment (bottom middle), but the NASH is weaker and further north. These biases may have implications for discerning NASH in uences, e.g. related anomalies that are higher in latitude than observations. Nevertheless, the basic circulation and Great Plains LLJ features follow NASH location and extent.
The Great Plains LLJ climatology can be compared in Fig. 5. The ERA5 time-mean V900 shows a strong (~8 m s −1 ) Great Plains LLJ feature (left). Despite the climatological core being about 5° northward from observational estimates, the location in CTRL experiment is close to the Rockies. The magnitude of the climatological core is 3 m s −1 , which is weaker than the observations. However, the objective of the study is to analyze large-scale dynamical differences between forced experiments, not to represent thermodynamics, diurnally-varying radiative processes nor mesoscale physics; therefore, this simulated Great Plains LLJ is within reason given that the model has relatively coarse resolution and lacks moist processes and associated land-atmosphere feedbacks.

d. Strong -Weak EAM-forced Experiment Analysis
An advantage of this experiment setup with the dry nonlinear AGCM is that one can assess the effect of sub-sampling 90-day (or one single season) means during the 900-day experiment. Fig. 6 demonstrates the internal variability of 90-day V900 means for this experiment; for V900 responses, the σ values are relatively large on the northern and southern edges of the climatological Great Plains LLJ region, and substantial off both North American coasts. This suggests that uctuations in V900 are primarily in the north-south direction. The Z250 and Z850 responses indicate relatively higher σ values along the approximate climatological jet stream latitude and along the boundaries of climatological subtropical highs, respectively (Fig. 7). This likely means that uctuations in upper and lower heights are linked to East Asian jet variations and shifts in the subtropical highs, respectively. By dividing the time-mean difference by this standard deviation, we assess the robustness (or statistical signi cance) of the longterm response on seasonal-to-interannual timescales.
The EAM-forced V900 response is summarized in Fig. 8, indicating a 0.5-1 m s −1 strengthening -a ~25% magnitude increase compared to Fig. 5 (right) -in the Great Plains LLJ. This strengthening is con ned to the northern side of the jet (right), which differs from the strong -weak EAM difference composite in Fig.  2 (top left). Overall, this forced response is considered robust on the seasonal timescale in the Great Plains and over the Gulf of Mexico, seen by the positive (negative) difference values that exceed 1σ (-1σ).
The EAM Z250 time-mean response shows zonally-oriented troughs and ridges that stretch from the EAM region and over the North Paci c (Fig. 9, left), with an anomalous trough-ridge pattern oriented west-east over North America, similar to the observed pattern (cf. Fig. 2, top right). The anomalous trough over western North America is typically associated with Great Plains LLJ strengthening (Harding and Snyder 2015; Mallakpour and Villarini 2016; Malloy and Kirtman 2020). The EAM Z850 time-mean response (right) presents anomalous ridging over eastern North America and the high-latitude North Atlantic and anomalous troughing over the mid-latitude Atlantic, which could signal an increased variability of the NASH in the west-east direction. The Z250 and Z850 responses are mostly robust except for the high latitudes and the eastern North Paci c/Alaska region.
To get a sense of the response in the vertical, we assessed the latitudinally averaged cross-section of strong -weak EAM V (left; meridional velocity) and divergence (right) response in the general region where the downstream wave response travels (35- 45°N; Fig. 10). The response is mostly equivalent barotropic except for over Gulf of Alaska/eastern North Paci c. However, the most statistically signi cant ΔV values are located over the EAM region as well as North America, including the upper-level trough and ridge from Fig. 9 (left). This corresponds to the anomalous divergence on the leeside of the Rockies (right). Despite the robust differences in this region, there is still substantial internal variability over the mid-latitude Paci c and/or the upper levels.
Lastly, we evaluated the in uence of NASH on the EAM responses, visualized by taking the strong -weak EAM responses during strong western NASH events and subtracting by the strong -weak EAM responses during weak western NASH events (Fig. 11), done for both observations (top row) and the dry AGCM (bottom row). The climatological biases of the dry AGCM are apparent, with Great Plains LLJ strengthening ~10° northward from the observational strengthening (left column). However, by considering these biases and comparing the NASH-modulated strong -weak EAM response (shaded contours) with the original strong -weak EAM response (no NASH considered, solid black contours), it is evident that strong western NASH modulation is comparable between observations and the dry AGCM, i.e. a strong western NASH ampli es the Great Plains LLJ strengthening signal, especially on the side closest to the Rockies. The dry AGCM generally simulates enhanced 500-250-hPa layer-averaged divergence in the Plains associated with the enhanced precipitation anomalies from observations (middle column). This suggests that dry dynamics in the AGCM may be su cient to produce basic NASH-related modulation of EAM-forced patterns, such that the tendencies are correct (see Supp. Fig. 2 for full strongweak EAM response separated by western NASH strength as in Fig. 2). Z250 patterns outside North America compare well to observations, but there are discrepancies in the dry AGCM representation of NASH modulation of Z250 over North America (right column) that may limit its representation.
NASH modulation is further demonstrated by taking vertical pro les of V where the Great Plains LLJ strengthening occurs (30°N for observations and 40°N for dry AGCM; Fig. 12). NASH modulation of circulation is notably similar to observations and the dry AGCM except over North Atlantic. Over the region of interest that affects the Great Plains LLJ, the dry AGCM presents alignment of positive V values from the low to upper levels (bottom), though not as vertically stacked as presented in observations (top) or Fig. 3. Overall, the dry AGCM simulates NASH modulation of EAM-forced responses to a reasonable degree, including the ampli cation of Great Plains LLJ strengthening and related divergence during strong EAM and strong western NASH events.

Summary And Discussion
Seasonal forecasts of CONUS precipitation during the summer have relatively low skill, and there is little consensus on the driving causes of rainfall variability on this timescale. We suggested that examining large-scale Great Plains LLJ responses in a dry nonlinear AGCM will aid in discerning dynamic causes and variability of pluvial events. First, we compared observational analysis of the NASH and EAM teleconnections and their interactions. Then we analyzed and compared Great Plains LLJ responses from EAM experiments in a dry AGCM and explored whether NASH modulation of EAM circulation responses can be reproduced with simple dry dynamics.
Results from the ERA5 conditional difference composites ( Fig. 1 and Fig. 2) suggested that the strength of the western NASH or EAM matters when considering Great Plains LLJ impacts. Strong western NASHrelated Great Plains LLJ strengthening and associated wet anomalies were greater during strong EAM events. However, EAM-related Great Plains LLJ responses were more dependent on the NASH location: during weak western NASH events, the strong -weak EAM response is a weakened Great Plains LLJ, and the LLJ response during strong western NASH events is not statistically signi cant. Pro les of V anomalies revealed that strong (weak) EAM and strong (weak) western NASH events were linked to inphase lower-and upper-level circulation patterns, providing enhanced upper-level support for the Great Plains LLJ (Fig. 3).
The strong -weak EAM responses were largely captured by the dry AGCM, including an elongated wave structure over the North Paci c and anomalous trough over western North America (Fig. 9) comparable to observations (cf. Fig. 2, top right). This promoted robust Great Plains LLJ strengthening (Fig. 8). In addition, the dry AGCM simulated the ampli cation of the EAM-forced LLJ and mid-to upper-level divergence during a strong western NASH due to constructive interference of low-and upper-level wind patterns (Figs. 10 and 11), shedding light on the major dynamic causes of Great Plains LLJ strengthening and its impacts.
Despite the AGCM capturing many of the dynamical processes behind EAM responses and NASH modulation, there were climatological biases in the AGCM that help explain some of the discrepancies between observations and the model's EAM-NASH-LLJ relationships. For example, in the upper levels, the model had a northward-shifted jet stream due to increased horizontal height gradients further north (cf. Fig. 4). Accordingly, the AGCM's Great Plains LLJ climatological core (Fig. 5)  , it is beyond the scope of this study to diagnose and disentangle true causal relationships between the EAM, NASH, Indian monsoon, and CGT. Future work will be needed to understand these inter-relationships and how they contribute to rainfall variability over Asia, North America, and Europe.
A future study should expand on NASH's role by forcing vorticity anomalies over the western NASH region with the AGCM or investigating other sources of North Atlantic Rossby wave activity (e.g. NAO; Weaver and Nigam 2008). In addition, the seasonal transition from early summer to late summer may also change the relationships between the NASH, EAM, and Great Plains LLJ. Simple AGCMs have the potential to isolate circulation responses from distinct forcing and evaluate the predictability of summer hydroclimate features. This research serves as a preliminary step for understanding more complex models and assessing the predictability of atmospheric dynamics in the summer on the more "elusive" long-range timescale. Figure 1 ERA5 strong -weak western NASH difference composites (top row) with no EAM condition considered, (middle row) only during weak EAM events, and (bottom row) only during strong EAM events. Difference composites of (left) V900 anomalies, (middle) CPC gauge-based precipitation anomalies, and (right) Z250 anomalies, with purple and green contours denoting the 1560-gpm line for strong and weak composites, respectively. Sample sizes for the composites and the percentage of total events the samples represent are annotated on top left of each row. Stippling indicates anomalies signi cant at 90% con dence level based on the Wilcoxin rank-sum test.

Figure 2
Similar format as Fig. 1       Standard deviation of the 90-day ΔV900 (Δ = strong EAM experiment -weak EAM experiment) moving mean.    Subtraction difference between strong EAM experiment and weak EAM experiment 35-45°N meridionally averaged time-mean pro le of (left) V and (right) divergence. The 1σ (solid black) and -1σ (dashed black) contours are overlaid as in Fig. 7.

Figure 11
Modulation of EAM response by NASH: (top row) Subtraction difference between ERA5 strong -weak EAM composites during strong western NASH and weak western NASH, i.e. Fig. 2 bottom row -middle row. (bottom row) Subtraction difference between strong EAM experiment -weak EAM experiment, i.e. strong -weak EAM response during strong western NASH minus strong -weak EAM response during weak western NASH, with 500-250-hPa layer-averaged divergence anomalies instead of precipitation anomalies. Strong -weak EAM V900 anomalies without NASH condition in Great Plains LLJ region are overlaid in left column for reference.

Figure 12
Modulation of vertical pro le of EAM response by NASH: (top) Subtraction difference between ERA5 strong -weak EAM 30°N V composites during strong western NASH and weak western NASH.
(bottom) Subtraction difference between strong EAM experiment -weak EAM experiment, i.e. strongweak EAM 40°N V response during strong western NASH minus strong -weak EAM 40°N V response during weak western NASH. Each panel annotates the approximate location of the Great Plains LLJ, and a thin vertical dotted line from LLJ is displayed to visualize upper-level support.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.