Abstract

Previous studies have shown that variations in extratropical cyclone activity significantly affect the frequency of extreme precipitation events over the Ohio Valley and northwestern United States. In this study, we examine the similarities and differences between the dynamics governing these events in these two regions. In the Ohio Valley, extreme precipitation events are associated with midlatitude synoptic-scale convergence northeast of cyclones and a southwestward oriented ridge near the Atlantic coast that drives strong water vapor transport from the Gulf of Mexico into the Ohio Valley. In the northwestern United States, extreme precipitation events are associated with a cyclonic and anticyclonic circulation pair aligned northwest to southeast, which together drive a long and strong moisture transport corridor from the lower latitude of the central Pacific Ocean toward the northwestern United States. Moisture budget analysis shows that moisture convergence due to dynamical convergence dominates in the Ohio Valley, whereas moisture advection dominates over the Pacific Northwest. Differences between the cases in the same region are examined by an empirical orthogonal function (EOF) analysis conducted on the vertically integrated moisture flux. Different EOFs highlight shifts in spatial location, orientation, and intensity of the moisture flux but demonstrate consistent roles of dynamics in the two regions. Composites based on these EOFs highlight the range of likely synoptic scenarios that can give rise to precipitation extremes over these two regions.

1. Introduction

Extratropical cyclones are a dominant driver for wintertime (December to February) extreme weather events over the midlatitudes (Ashley and Black 2008; Frankoski and DeGaetano 2011; Colle et al. 2013). The heavy precipitation and floods associated with them can cause tremendous loss to human society (Easterling et al. 2000; Rappaport 2000; Pall et al. 2011). Kunkel et al. (2012) subjectively assigned each daily extreme precipitation event a meteorological cause and showed that about 54% of extremes are near the fronts of extratropical cyclones, and about 24% are near the extratropical cyclone low pressure center. In northeastern United States, 60%–80% of 6-hourly precipitation extremes occur within the circulation of a cyclone in winter (Pfahl and Wernli 2012). Numerous case studies of hazardous weather for the winter over the United States demonstrate that these are mostly related to the passage of deep cyclones (e.g., Bosart 1981; Zhang et al. 2002; Cardone et al. 1996). Detailed statistical composites of extratropical cyclones have related the extreme precipitation to cyclone depths (Polly and Rossow 2016) or cyclone intensification (Rudeva and Gulev 2011). Over the U.S. west coast, many such events have been linked to atmospheric rivers (Ralph et al. 2006; Guan et al. 2010), which are also related to cyclones (Zhu and Newell 1994; Ralph et al. 2004). These extreme precipitation events not only are important for their weather impacts, but also provide for much of the winter snowpack that is critical for water resource (e.g., Eldardiry et al. 2019).

Apart from the strength of cyclones, modeling studies have shown that changes in moisture flux convergence are also important for modulating extreme precipitation (Meehl et al. 2005). Wong et al. (2018) studied the precipitation structure inside extratropical cyclones by decomposing the large-scale moisture flux convergence into two moisture tendency terms: moisture advection and moisture change due to dynamical convergence. Precipitation type and amount in different sectors of the cyclones are related to the relative contribution of the two terms to the total moisture flux convergence (Wong et al. 2018).

Besides the decomposition according to moisture transport mechanisms, the moisture budget can also be separated into contributions from different time scales. Based on the European Centre for Medium-Range Weather Forecasts interim reanalysis (ERA-Interim), during the cool season transient eddies converge moisture across much of the United States while the mean flow provides moisture to the northwest and dries the southwest (Seager et al. 2014). Under global warming, the CMIP5 models project drying for the southwest and wetting to the north, with changes in the mean flow moisture convergence being largely responsible across the west but intensified transient eddy moisture convergence wetting the northeast (Seager et al. 2014). Over the southwestern United States, the models’ projected spring drying is mainly caused by decreased mean moisture convergence, partially compensated by the increase in transient eddy moisture convergence (Ting et al. 2018). Hence future changes in precipitation depend on how the changes in the mean flow or the transient moisture convergence dominates one another.

Most previous studies focused on the meteorological cause for each extreme precipitation event. Ma and Chang (2017) quantified the response of extreme precipitation frequency against cyclone activity variations over the continental United States winter by winter. Associated with an overall increase in cyclone activity, this response mostly focuses on the Ohio Valley–Great Lakes region, showing much enhanced extreme precipitation rate in high cyclone activity winters, together with some signal in the northwestern United States Figure 1 shows that during winter with enhanced storm track activity, the frequency of extreme precipitation events is strongly enhanced over the aforementioned two regions (Fig. 1b). The apparent difference in spatial scales of the extreme precipitation regions associated with cyclone activity over these two regions is consistent with the findings of Touma et al. (2018).

Fig. 1.

(a) The difference in storm track activity {in terms of filtered SLP variance statistics, i.e., ECApp [see Eq. (6)]; hPa2} between the 10 high ECApp winters and the 10 low ECApp winters based on the MERRA-2 product. (b) Difference in the frequency of 99th-percentile extreme precipitation events (%; based on CPC daily gauge-based precipitation data) between high ECApp and low ECApp winters. We highlight the precipitation in two regions by superposing two boxes: the Great Lakes and Ohio Valley (36°–50°N, 95°–80°W) and the northwestern United States (44°–50°N, 125°–111°W). The dotted regions in (a) and the cross-hatched regions in (b) show significant difference at the 5% level based on the t test.

Fig. 1.

(a) The difference in storm track activity {in terms of filtered SLP variance statistics, i.e., ECApp [see Eq. (6)]; hPa2} between the 10 high ECApp winters and the 10 low ECApp winters based on the MERRA-2 product. (b) Difference in the frequency of 99th-percentile extreme precipitation events (%; based on CPC daily gauge-based precipitation data) between high ECApp and low ECApp winters. We highlight the precipitation in two regions by superposing two boxes: the Great Lakes and Ohio Valley (36°–50°N, 95°–80°W) and the northwestern United States (44°–50°N, 125°–111°W). The dotted regions in (a) and the cross-hatched regions in (b) show significant difference at the 5% level based on the t test.

While Ma and Chang (2017) found these associations, they did not explore the dynamics behind these relationships. This study extends the work of Ma and Chang (2017) and examines the physical mechanisms in terms of large-scale circulation patterns and the associated moisture transport that support the relationships between extreme precipitation and cyclone events, and whether the mechanisms that drive the Ohio Valley and northwestern U.S extreme precipitation are different. To systematically understand the difference between the two regions, we examine both the seasonal scale and daily scale. At the seasonal scale, we examine the following questions: How does the moisture budget look like in high cyclone activity winters and low cyclone activity winters? Which moisture transport mechanism, advection or dynamical convergence, contributes most to the total moisture flux convergence in different places? Are there changes in the moisture sources that are related to the precipitation extremes in the highlighted regions? At the daily scale, we investigate the followings: How frequent are the extreme precipitation events associated with cyclones? What does the circulation pattern look like and how is the moisture transported during extreme precipitation events? Our results suggest that there are significant differences between these two regions. The synoptic situations giving rise to individual extreme precipitation events are not the same. Variations among all these extreme precipitation days in both regions are examined through empirical orthogonal function (EOF) analysis of the moisture flux to provide a likely range of synoptic conditions associated with these extremes.

2. Data and methods

a. Data

The Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2), data are used to compute monthly mean moisture tendencies related to large-scale advection and dynamical convergence for diagnostics of the sources of precipitation. MERRA-2 is the latest atmospheric reanalysis of the modern satellite era produced by NASA’s Global Modeling and Assimilation Office (GMAO; Gelaro et al. 2017). It aims to provide improved accuracy in global water cycle variability (Bosilovich et al. 2017).

The moisture tendency terms are described in Wong et al. (2016), and are calculated from the hourly MERRA-2 products of vertically integrated moisture flux and total precipitable water on a 0.625° × 0.5° grid. Finite differencing is used for horizontal gradients and convergence. The final products are then interpolated onto 1.5° × 1.5° resolution by local area averaging. Based on the water vapor budget equation (e.g., Peixoto and Oort 1992)

 
PE+Qt=(QV)=QVVQ,
(1)

where

 
Q=1gρw0psqdp
(2)
 
V=1Q1gρw0psqwdp,
(3)

as mentioned in Wong et al. (2016), precipitation minus evaporation (PE) plus the tendency of total precipitable water (∂Q/∂t) is balanced by moisture flux convergence, which can be further decomposed into two terms: the tendency related to large-scale dynamical convergence (−Q∇ · V) and the tendency related to moisture advection (−V · ∇Q). The two-dimensional vector V is the vertically integrated velocity of horizontal winds w, weighted by the humidity profile in each column [Eq. (3)]. To further analyze the strength of anomalies at high and low frequency, we compute the monthly mean and 10-day high-pass transient components for both Q and V using a fast Fourier transform:

 
VQ¯V¯Q¯VQ¯
(4)
 
QV¯Q¯V¯QV¯,
(5)

where overbarred variables represent monthly mean and primed variables are 10-day high-pass transient components. Similarly, the transient products are first computed on the original grid and then interpolated to the 1.5° × 1.5° resolution for each month.

To directly compare with the above moisture budget terms, we use the 3-hourly sea level pressure (SLP) and hourly precipitation products (both at 0.5° × 0.625°) from MERRA-2 to compute the storm track activity, seasonal mean precipitation and extreme precipitation counts. Here we use the “PRECTOT” precipitation output from the GMAO atmospheric model, which is consistent with the assimilation results for more basic variables (like SLP, wind, etc.) and can close the model moisture budget when the analysis increment for water vapor is taken into account. Since our previous study (Ma and Chang 2017) used daily precipitation for extreme events, here we transform the hourly data into daily. To close the moisture budget [Eq. (1)], we also use the monthly mean surface evaporation flux (0.5° × 0.625°) from MERRA-2. The period for this analysis is between 1980 and 2011.

We have also compared GMAO precipitation and evaporation products with the ERA-Interim reanalysis precipitation and evaporation, and Climate Prediction Center (CPC) gauge-based precipitation data for many of the analyses presented in our previous study (Ma and Chang 2017) and made sure that they are highly consistent. For example, the extreme precipitation frequencies derived from MERRA-2 products are largely consistent with those derived from CPC data (cf. Fig. 2b with Fig. 1b), and the 10 high and 10 low storm track years calculated from MERRA-2 data (3-hourly SLP; 0.5° × 0.625°) are exactly the same set of years as selected by using ERA-Interim data (6-hourly SLP; 2.5° × 2.5° or 0.75° × 0.75°). Note that CPC (and not reanalysis) data are used to quantify extreme precipitation and select the extreme precipitation days. To further study the monthly mean tendency associated with large-scale dynamical divergence, we use the monthly mean dynamical divergence and specific humidity (0.75° × 0.75°) at 32 pressure levels from ERA-Interim.

Fig. 2.

Composite difference between 10 high ECApp winters and 10 low ECApp winters for the MERRA-2 data for (a) precipitation (mm day−1), (b) 99th-percentile extreme precipitation frequency, (c) evaporation (mm day−1), (d) moisture advection VQ¯ (mm day−1), (e) moisture convergence QV¯ (mm day−1), and (f) total moisture budget, which is the sum of (d) and (e). Also shown are the product of monthly mean terms: (g) V¯Q¯, (h) Q¯V¯, and (i) sum of (g) and (h), along with the covariance of 10-day high-pass transient terms: (j) VQ¯, (k) QV¯, and (l) sum of (j) and (k). The cross-hatched regions are significant at the level of 5%.

Fig. 2.

Composite difference between 10 high ECApp winters and 10 low ECApp winters for the MERRA-2 data for (a) precipitation (mm day−1), (b) 99th-percentile extreme precipitation frequency, (c) evaporation (mm day−1), (d) moisture advection VQ¯ (mm day−1), (e) moisture convergence QV¯ (mm day−1), and (f) total moisture budget, which is the sum of (d) and (e). Also shown are the product of monthly mean terms: (g) V¯Q¯, (h) Q¯V¯, and (i) sum of (g) and (h), along with the covariance of 10-day high-pass transient terms: (j) VQ¯, (k) QV¯, and (l) sum of (j) and (k). The cross-hatched regions are significant at the level of 5%.

For the extreme events, we define the extreme precipitation days based on the CPC gauge-based daily precipitation data (0.25° × 0.25°). To examine the circulation patterns on these extreme days, we use the 6-hourly SLP, 500-hPa geopotential height, vertically integrated divergence of moisture flux, and vertically integrated water vapor flux from ERA-Interim, all at a resolution of 0.75° × 0.75°, from 1979 to 2010. Figures 312 (described in detail below) are based on ERA-Interim data.

To calculate the cyclone frequency in winter, we use the cyclone track product as described in Chang and Yau (2016). The tracking is performed based on ERA-Interim’s 6-hourly SLP product (2.5° × 2.5°) for each winter, using the objective tracking algorithm of Hodges (1994). Here we use the filtered product, where the seasonal mean is first removed for each winter, and high-pass filtering is performed over the spatial anomalies with only wavenumbers 5–70 retained. Hence the large spatial scale and low-frequency temporal scale background is removed before the tracking algorithm is run. While previous studies (e.g., Neu et al. 2013; Raible et al. 2008) have suggested that cyclone tracks may be dependent on the tracking algorithm, statistics for strong cyclones are more robust. Here we are mainly concerned with the presence (or not) of significant cyclones during extreme precipitation events. All negative SLP anomaly centers are tracked, with tracks lasting less than 2 days or traveling less than 1000 km removed.

b. Methods

There are two families of methods that are widely used to study cyclone activity. The first directly studies cyclone trajectories using objective tracking algorithms (Hodges 1994, 1999; Hoskins and Hodges 2002), as mentioned in the last section. The second, on the other hand, studies the synoptic time scale variability using variance statistics (Blackmon et al. 1977, 1984a,b; Lau 1988; Wallace et al. 1988). The passage of a cyclone close to any location generates rapid pressure perturbations, circulation anomalies, and temperature advection. Thus, the use of temporal variance or covariance is an alternative way to represent cyclone activity. These statistics generally display two maxima spanning the North Pacific and North Atlantic Oceans (e.g., Chang et al. 2002). These centers coincide with the “storm track” region, introduced by earlier surveys of cyclone trajectories of individual cyclone centers (e.g., Petterssen 1956, 267–276). One should keep in mind that by using the filtered variance metrics to measure the storm track activity, the variance related to anticyclones is also included. However, anticyclones are usually slow moving and have pressure anomalies weaker than the cyclones (Hoskins and Hodges 2002), hence the variance is likely to be dominated by cyclones. In addition, our results will show that anticyclones are also important for generating extreme precipitation.

In this study, we use both methods but for different purposes. To validate the relationship between cyclone and extreme precipitation, during the extreme precipitation days, we count the number of times a tracked low pressure center is within 500 km of a grid point, which is equivalent to assuming that a cyclone has a radius of about 500 km, a threshold widely used in previous studies (e.g., Sinclair 1997; Grise et al. 2013). The total count for all composited events will be normalized as the average count per day. Since the temporal variance method has been shown to be a good representation that captures the essence of midlatitude cyclone activity, we also use it to show the interannual variability of cyclone activity. To be consistent with our previous study (Ma and Chang 2017), we use the 24-h difference filtered variance of SLP to measure extratropical cyclone activity (ECA; Wallace et al. 1988), referred to as ECApp:

 
ECApp=[SLP(t+24h)SLP(t)]2¯,
(6)

where the variance is taken for each winter. Similar metrics are widely used in previous work to quantify cyclone or storm track activity (Lau 1988; Feser et al. 2015; Alexander et al. 2005; Chang et al. 2012, 2016; etc.).

Our results are mainly shown through composite analyses. We summarize each composite by averaging or by counting extreme events and test their difference through a t test to compare their means or proportions.

According to Ma and Chang (2017), there are two regions that show strong variability of extreme precipitation frequency (99th percentile) associated with variations of ECApp over the continental United States: the Great Lakes–Ohio Valley (36°–50°N, 95°–80°W) and northwestern United States (44°–50°N, 125°–111°W) (marked by the rectangles in Fig. 1b). These two regions are also favored by teleconnections in winter, such as ENSO and PNA (e.g., Kunkel and Angel 1999; Montroy et al. 1998; Ning and Bradley 2015; Leathers et al. 1991). The extreme precipitation days (based on CPC precipitation data) for each region are selected in the following way: we put a box over each of the region mentioned above. Within each box, we only consider the grid points at which variations in extreme precipitation frequency are found to be significantly correlated with those in cyclone activity by Ma and Chang (2017). There are two reasons for only selecting the significant points: 1) The regions we are interested in and also highlighted by previous studies are not necessarily rectangles. 2) In certain locations, the precipitation is generated by orographic lifting, and hence only the grid points close to the mountains will have significant response. This can be important for the northwestern U.S. box, since the significant points in that box do not cover much of the area and occur mainly in two clusters: one being the coastal part of Washington State, and the other in western Montana and northern Idaho (Fig. 1b).

The daily accumulated precipitation amount of these selected grid points is averaged as a single time series. From this time series, we pick the days with the largest 5% of values and define them as the extreme precipitation days for this region. Since the daily data cover 32 years, about 90 winter days per year, the top 5% provides 145 extreme precipitation days, which are shown in Tables S1 and S2 in the online supplemental material for the Great Lakes–Ohio Valley region and northwestern United States, respectively.

To examine the variations among these 145 extreme precipitation days, we perform an EOF analysis on the column-integrated water vapor flux, concatenating the zonal and meridional component into a single field. The water vapor flux is chosen for the EOF analysis because it contains information regarding the moisture sources and sinks as well as transport pathways.

3. Moisture budget anomalies over seasonal time scale

In Ma and Chang (2017), ECApp is averaged over the continental part of the domain (60°–140°W, 25°–55°N) and the highest and lowest ⅓ (10 winters) are selected to form the high and low composites from the 32 years of data (1979–2010). Within each composite, we count the 99th-percentile precipitation events, which are defined over gridded daily precipitation data when the daily precipitation amount exceeds the local top first-percentile threshold. In the rare cases that the threshold is not meaningful (e.g., in very dry regions where the top first percentile of precipitation might be zero), the points are masked as missing. For any grid point, the nonmissing part is used if the fraction of missing data is less than 10% of its length; otherwise the point is not included in the analysis.

Figure 1 reproduces part of Fig. 4 of Ma and Chang (2017). Figure 1a shows the composite difference in ECApp, which is highly consistent with Fig. 1c of Ma and Chang (2017), except that MERRA-2 shows a slightly larger difference (by 5–10 hPa2) over the central United States. Figure 1b shows the difference in the frequency of 99th-percentile extreme precipitation events between the high and low composite based on CPC precipitation data, which highlights the Ohio Valley and the Pacific Northwest, where the absolute value of the difference is between 0.6% and 1.5%. Statistically, the probability of a 99th-percentile event happening is 1%. Thus, the response mentioned above is very strong, showing the significant impacts of storm track variations. Note that while we base our composites on the magnitude of ECApp, Ma and Chang (2017) showed that this “mode” is identical to the leading mode of covariability between ECApp and extreme precipitation frequency as found from a singular value decomposition analysis of the covariance between ECApp and extreme precipitation frequency over the continental United States.

Figure 2 shows the differences in MERRA-2 precipitation statistics and moisture budgets between high and low ECApp years. Results based on ERA-Interim data corresponding to Figs. 2d–i are qualitatively consistent and are shown in Fig. S1 of the online supplemental material for reference. Figure 2b shows the difference in 99th-percentile extreme precipitation frequency based on MERRA-2 precipitation, which highlights the positive response around the Great Lakes and the negative response over the Gulf of Mexico and the Atlantic east of Florida. Differences in extreme precipitation frequency from MERRA-2 (Fig. 2b) have patterns that are similar to those from CPC precipitation (Fig. 1b), with MERRA-2 having a weaker signal over the Ohio Valley that is more concentrated around the Great Lakes, and a weaker signal over the northwestern United States. Near the Gulf of Mexico coast, MERRA-2 shows a more pronounced negative signal. Figure 2a shows the difference in seasonal mean precipitation between the high and low composites. Similar to the response in extreme precipitation frequency (Fig. 2b), Fig. 2a shows a significantly positive signal over the Great Lakes and Ohio Valley, and a strongly negative signal over the Gulf of Mexico and subtropical Atlantic. A significantly positive signal can be seen in southwestern Canada spreading into northwestern United States. In summary, the response in extreme precipitation frequency is consistent between MERRA-2 and CPC, and the response in extreme precipitation frequency is consistent with the response in seasonal mean precipitation.

Figures 2d and 2e show the response of the moisture advection (VQ¯) and dynamical convergence (QV¯) contributions to the moisture budget, and Fig. 2f shows the sum of these two terms. The advection and convergence terms show opposite signs in many places: the coastal region of Washington and Oregon receives strong moisture advection from the Pacific, while there is strong divergence in these regions and the eastern Pacific. The sum is still positive for the coastal regions of Washington and Oregon, implying the advection is slightly stronger. From the central United States (Texas, Oklahoma, Kansas) to the Ohio Valley, there is strong convergence; the advection there is negative but much weaker. Close to the north shore of the Great Lakes, there is a small region where the advection is significantly positive while the convergence is much smaller. Along the coastal southeastern United States, the advection is negative while the moisture convergence is positive; there is significant cancellation between the two and their sum is a much weaker residual.

To close the moisture budget, we also show the composite difference for evaporation in Fig. 2c using the MERRA-2 product. Based on Eq. (1), for precipitation, the variability of remote evaporative source might also have significant contribution to our studied regions. In Fig. 2c, there is significantly more evaporation east of Cuba and equatorward of 28°N (0.6–0.9 mm day−1) and less evaporation along the eastern U.S. coast into the central Atlantic Ocean (from −0.6 to −1.2 mm day−1) in the high ECApp winters than the low ECApp winters. In the Gulf of Mexico, the evaporation is significantly reduced along the Texas coast and the south end of Texas, and not significantly changed in the remaining area. Along the coast of northwestern United States and western Canada, the difference in evaporation is significantly positive (0.3– 0.6 mm day−1). Over the continental United States, the difference is much smaller. The correlation coefficients between evaporation and area averaged ECApp over the United States are also significantly positive to the east of Cuba and near the west coast of Canada and northwestern United States, and not significant in the Gulf of Mexico (see Fig. S2c in the online supplemental material). The positive evaporation anomalies east of Cuba is in the path of the clockwise circulation of moisture transport anomalies (see discussions below), and hence it might have some positive contribution to the precipitation in Ohio Valley. Along the coast of the northwestern United States, the positive anomaly of evaporation is also in the path of the composite mean moisture transport (see discussions below), but the evaporation difference is only one-half of the values in subtropical Atlantic. Results based on ERA-Interim are consistent and are shown in supplemental Fig. S2. Over these regions, SST anomalies are negative [not shown, but see Fig. 13c of Ma and Chang (2017)], and thus enhanced evaporation is likely due to increase in surface wind. These results suggest that changes in evaporation may contribute, but further analyses using backward trajectories from the extreme precipitation regions will be needed to quantify the contributions from increased evaporation.

The sum of the advection and convergence terms (Fig. 2f) is consistent with the response of seasonal mean precipitation (Fig. 2a), especially after evaporation changes are taken into account (Fig. S3 in the online supplemental material). Over southwestern Canada and the northwestern United States, there are two parallel bow-shaped lines of enhanced values, both for the response of seasonal mean precipitation (Fig. 2a) and for the sum of moisture budgets’ response (Fig. 2f), one over Washington and Oregon and the other in northern Idaho and western Montana. These bow-shaped signals correspond to the two local maxima of extreme precipitation rate based on CPC’s results (Fig. 1c). From a topographic map (not shown), it can be seen that these two parallel bands correspond to the Cascade Range and the Rocky Mountains, respectively. With strong advection from the Pacific (Fig. 2d), air is forced to climb over the mountains resulting in orographic precipitation. This is probably why the northwestern United States can get extreme precipitation without positive convergence. Based on conservation of Ertel’s PV and mass, such upward motion is associated with divergence instead of convergence. Stronger airflow is expected to give rise to stronger divergence, which is consistent with strong moisture divergence seen in the northwestern United States (Fig. 2e).

As discussed in section 2 [Eqs. (4) and (5)], the monthly mean advection and convergence terms are decomposed into the products of seasonal means (Figs. 2g,h) and mean of transient covariance terms (Figs. 2j,k). Figures 2g and 2h show the response of mean moisture advection (V¯Q¯) and mean convergence (Q¯V¯), with Fig. 2i as the sum of them. Similarly, Figs. 2j and 2k show the response of transient moisture advection (VQ¯) and transient convergence (QV¯), with Fig. 2l as the sum of them. Over much of the United States and the eastern Pacific, the transient terms are much weaker than the mean terms, even though they are still significant. But over the Gulf of Mexico and along the coastal states nearby, the transient moisture advection (Fig. 2j) is dominant over the mean advection term (Fig. 2g), since the total advection’s response (Fig. 2d) shares the same sign as the transient advection’s response. Comparing with the transient advection, the transient convergence (Fig. 2k) is much weaker over the domain. The negative transient advection’s response in the southeastern United States (Fig. 2j) is largely cancelled by the positive response of mean advection (Fig. 2g) and mean convergence (Fig. 2h), leaving only the coastal region of Alabama and Georgia with negative values (Fig. 2f). Except for the southeastern United States, much of the United States is still dominated by the mean moisture budget terms.

Since we are showing the response to storm track variations, and the cyclone activity mostly involves disturbances with a period of 2–10 days, we might wonder why the mean terms dominate in many places. Our hypothesis is that even though cyclones are by themselves a high-frequency phenomenon, it is not necessary that their contribution to the moisture budget must stay in the corresponding high-frequency range. Stronger cyclone activity usually increases the seasonal mean dynamical convergence (given anticyclones are usually weaker than their nearby cyclones), hence enhancing the seasonal mean moisture convergence. Similarly, variations of cyclone activity contribute to the seasonal mean flow due to eddy momentum transports, thus affecting the mean moisture advection. On the other hand, cyclone activity is also affected by variations in the mean flow, with stronger flow (indicating stronger baroclinicity) and convergence east of quasi-stationary troughs being favorable conditions for enhanced cyclone activity (e.g., Chang et al. 2002). Consequently, detailed explanation of what determines whether the mean response or the transient response should dominate still needs further investigation.

Since in the Ohio Valley it is the response of the mean convergence term (Q¯V¯) that dominates, we further investigate whether it is the variations of seasonal mean precipitable water Q¯ or the variations of seasonal mean convergence (V¯) that is more important between the high and low storm track winters. We use the monthly mean divergence and specific humidity from ERA-Interim and integrated through 32 pressure levels to calculate V¯ and Q¯ for each winter. Then we decompose the variations of Q¯V¯ in the following way:

 
δ(Q¯V¯)δQ¯(V¯)Q¯δ(V¯),
(7)

where the δX means the high composite average of X minus the low composite average of X, and the non-δ term is approximated by the average of high and low composite.

Figures 3a and 3b compare these two terms. In most places including the Ohio Valley, the Gulf of Mexico, the subtropical Atlantic, and the eastern Pacific, the Q¯δ(V¯) term (Fig. 3b) is much larger than the δQ¯(V¯) term (Fig. 3a), which is close to zero almost everywhere. So, it is the variations of mean dynamical convergence δ(V¯) instead of the variations of mean precipitable water δQ¯ that dominates this term. This result further implies that the contribution of evaporation variations over the moisture source region might not be the dominant term for the precipitation in Ohio Valley. We have also shown the anomalies and average of precipitable water in Figs. 3c and 3d, separately. The anomalies in the precipitable water shows its maximum in southeastern United States (Fig. 3c), but the value represents only about a few percent of the mean precipitable water (Fig. 3d). On the other hand, in the Ohio Valley and the Great Lakes, the anomalies in the convergence is about 0.04 day−1 (Fig. 3e), but the averaged mean convergence is of the same order of magnitude. Thus the Q¯δ(V¯) term (Fig. 3b) dominates over the δQ¯(V¯) term (Fig. 3a) over these regions.

Fig. 3.

The interannual variations of mean moisture convergence δ(Q¯V¯) is decomposed and approximated by two terms: (a) the product between the variations of precipitable water δQ¯ and the averaged mean dynamical convergence V¯ (mm day−1) and (b) the product between the variations of mean dynamical convergence δ(V¯) and the averaged precipitable water Q¯ (mm day−1). (c) Composite difference for precipitable water between the 10 high ECApp winters and 10 low ECApp winters (mm). The stippled regions show significant difference at the 5% level. (d) Averaged precipitable water between both composites (mm). (e),(f) As in (c) and (d), but for the mean dynamical convergence (day−1). The regions where the surface pressure has ever been below 850 hPa has been masked out for the dynamical convergence and associated products.

Fig. 3.

The interannual variations of mean moisture convergence δ(Q¯V¯) is decomposed and approximated by two terms: (a) the product between the variations of precipitable water δQ¯ and the averaged mean dynamical convergence V¯ (mm day−1) and (b) the product between the variations of mean dynamical convergence δ(V¯) and the averaged precipitable water Q¯ (mm day−1). (c) Composite difference for precipitable water between the 10 high ECApp winters and 10 low ECApp winters (mm). The stippled regions show significant difference at the 5% level. (d) Averaged precipitable water between both composites (mm). (e),(f) As in (c) and (d), but for the mean dynamical convergence (day−1). The regions where the surface pressure has ever been below 850 hPa has been masked out for the dynamical convergence and associated products.

4. Conditions on extreme precipitation days

We have examined the response of extreme precipitation frequency and moisture budget to the variations of storm track activity at the seasonal time scale. Now we will examine how these extreme precipitation events are related to cyclones in the Ohio Valley and northwestern United States.

a. Large-scale conditions

1) Ohio valley

As described in section 2, we select 145 extreme precipitation days for the Ohio Valley and Great Lakes region. Figure 4 shows the composite average (Fig. 4, left column) and the composite anomalies which is the deviation from the climatology (Fig. 4, right column) for these days. During this extreme precipitation composite, the SLP is about 6–10 hPa lower than climatology within the box, and about 10 hPa higher over the North Atlantic (Figs. 4a,b). The 500-hPa geopotential height shows clearly Rossby wave–like troughs and ridges for the composite (Fig. 4c), with the box locating east of a trough and west of a ridge, a region that favors cyclogenesis and upward motion. In particular, the anomalous height field shows a zonally oriented synoptic-scale wave train pattern with amplitude over 100 m (Fig. 4d; see Wallace et al. 1988; Chang 1993). We have also examined the composite average for 1 day before and 1 day after the extreme precipitation days (not shown). Together they show clear eastward wave propagation. The cyclone count per day is strongly enhanced within much of the box extending southwestward and is significantly reduced over the North Atlantic (Figs. 4e,f). The cyclone count, SLP, and 500-hPa geopotential height pattern together confirm the relationship between the extreme precipitation events and midlatitude cyclones. Clearly, this relationship can be explained by typical midlatitude synoptic (Rossby) wave dynamics.

Fig. 4.

(a) SLP averaged within the 145 extreme precipitation days (based on CPC gauge-based daily precipitation data) defined for the box around the Great Lakes and Ohio Valley (hPa). (b) Difference between (a) and the climatology (hPa), where the stippled regions show significant difference at the 5% level. The remaining panels are presented in the same way as (a) and (b), but for (c),(d) 500-hPa geopotential height (m); (e),(f) cyclone density using 500-km radius (day−1); (g),(h) convergence of moisture flux, which is the sum of moisture convergence and advection (mm day−1); and (i),(j) column-integrated water vapor flux (kg m−1 s−1). All of the variables except precipitation are based on 6-hourly ERA-Interim data.

Fig. 4.

(a) SLP averaged within the 145 extreme precipitation days (based on CPC gauge-based daily precipitation data) defined for the box around the Great Lakes and Ohio Valley (hPa). (b) Difference between (a) and the climatology (hPa), where the stippled regions show significant difference at the 5% level. The remaining panels are presented in the same way as (a) and (b), but for (c),(d) 500-hPa geopotential height (m); (e),(f) cyclone density using 500-km radius (day−1); (g),(h) convergence of moisture flux, which is the sum of moisture convergence and advection (mm day−1); and (i),(j) column-integrated water vapor flux (kg m−1 s−1). All of the variables except precipitation are based on 6-hourly ERA-Interim data.

Figures 4g and 4h show the convergence of moisture flux in the extreme composite and its anomalies from the climatology. The maximum of moisture flux convergence is located near the maximum of cyclone count except displaced slightly eastward. In a typical cyclone, the poleward airstream of warm and moist air [the warm conveyor belt (WCB)], which runs ahead of the cold front and climbs over the warm front, is usually east of the low pressure center of the cyclone. Apart from the moisture convergence in the Ohio Valley, there is also a region of moisture flux divergence in Texas. This contrast is also captured by MERRA-2 at the seasonal scale (Fig. 2f). In addition, MERRA-2 also shows that the drying in Texas is mostly because of the negative advection (Fig. 2d), even though the moisture convergence is positive there (Fig. 2e). Finally, Figs. 4i and 4j show the moisture flux in the extreme composite and its anomalies. Most of the water vapor flux comes from the Gulf of Mexico, which is the main source of moisture for the eastern United States, but part of the flux also comes from Texas. In the extreme composite, the vapor flux is much stronger and more poleward than the climatology (not shown), but quickly turns into pure eastward transport in the North Atlantic, creating clockwise circulation anomalies in Fig. 4j, which is consistent with the higher SLP, positive 500-hPa geopotential height anomalies, and smaller cyclone count in the North Atlantic, shown in Figs. 4b, 4d, and 4f. In summary, a cyclone in the Ohio Valley, accompanied by an anticyclone to its east, provides a strong pressure gradient and southwesterly flow that give rise to strong moisture transport into the Ohio Valley, fueling these extreme precipitation events.

To further investigate the conditions favorable for producing extreme precipitation, we stratified the extreme precipitation days into days in which a cyclone center can be located within the high cyclone density region in Fig. 4e (area covered by the 0.2 contour; 99 cases or about ⅔ of the cases) and those in which a cyclone is not identified by the tracker over this region. The composite SLP and moisture flux anomalies for the cyclone cases are shown in Figs. 5a and 5b. These composites resemble Figs. 4b and 4j, except that the cyclonic anomaly is even slightly stronger. For the cases when extreme precipitation occurred but a cyclone is not identified by the tracker, the composites (Figs. 5c,d) still show a weak cyclonic anomaly resembling a trough (likely the location of a diffuse cyclone or a cold front) over the same region, but a much stronger and southwestward extended ridge over the western Atlantic. The ridge–trough couplet gives rise to a strong moisture flux anomaly over the southeastern United States, again providing very strong moisture transport into the Ohio Valley.

Fig. 5.

Anomalies of SLP and moisture flux averaged for the composite that are (a),(b) extreme precipitation days for the box and have at least one cyclone located within the 0.2 contour of Fig. 4e on that day (99 days); (c),(d) extreme precipitation days for the box but have no cyclone located within the 0.2 contour of Fig. 4e on that day (46 days); and (e),(f) nonextreme precipitation days but have at least one cyclone located within the 0.3 contour of Fig. 4e on that day with a daily maximum intensity at least 17 hPa (26 days). In this figure, for counting cyclones and finding the daily maximum intensity, we represent a cyclone as all of the grid points’ values 100 km from the tracked cyclone center.

Fig. 5.

Anomalies of SLP and moisture flux averaged for the composite that are (a),(b) extreme precipitation days for the box and have at least one cyclone located within the 0.2 contour of Fig. 4e on that day (99 days); (c),(d) extreme precipitation days for the box but have no cyclone located within the 0.2 contour of Fig. 4e on that day (46 days); and (e),(f) nonextreme precipitation days but have at least one cyclone located within the 0.3 contour of Fig. 4e on that day with a daily maximum intensity at least 17 hPa (26 days). In this figure, for counting cyclones and finding the daily maximum intensity, we represent a cyclone as all of the grid points’ values 100 km from the tracked cyclone center.

How are these conditions different from regular cyclone days? We have also examined cases in which a cyclone is located within the same region but no extreme precipitation occurred. It turns out that the intensity of the cyclones associated with extreme precipitation (median intensity 18 hPa) is higher than those that do not produce extreme precipitation (median intensity 12 hPa). Nevertheless, there are cases in which there is a strong cyclone without causing extreme precipitation. Composites based on such cases in which cyclones with intensity stronger than 17 hPa occurring within the 0.3 contour in Fig. 4e are shown in Figs. 5e and 5f. For these cases, a deep cyclone is found over the Ohio Valley, but the ridge to its east is much weaker and retreated northward. The absence of the southwestward ridge extension along the Atlantic coast means that the pressure gradient over the southeastern United States is much weaker despite there being a strong cyclone, thus the moisture flux anomaly over that region is also much weaker (Fig. 5f). These results show that apart from the existence of a cyclone over the Ohio Valley, the southwestward extended ridge near the Atlantic coast is also critical for the occurrence of extreme precipitation over the Ohio Valley.

2) Northwestern United States

For the northwestern United States (44°–50°N, 125°–111°W), composites for the extreme precipitation days and anomalies from climatology are shown in Fig. 6. As mentioned before, the grid points considered are only the significant ones shown in Fig. 1c, so the extreme precipitation mostly happens in two regions: the coastal region of Washington and Oregon, and northern Idaho and western Montana. This can be directly observed in Fig. 6g, which shows the convergence of moisture flux in the extreme composite, and Fig. 6h, which shows the difference from climatology. This feature is also captured by MERRA-2 in the seasonal mean precipitation’s response (Fig. 2a) and the response of the total moisture budgets (Fig. 2f).

Fig. 6.

As in Fig. 4, but for the extreme precipitation days defined for the box over the northwestern United States.

Fig. 6.

As in Fig. 4, but for the extreme precipitation days defined for the box over the northwestern United States.

Instead of a synoptic wave-like pattern, the northwestern U.S. composites show a pair of cyclonic and anticyclonic anomalies aligned northwest to southeast, regardless of SLP (Figs. 6a,b), 500-hPa geopotential height (Fig. 6d), and cyclone count (Fig. 6f). Comparing with the Ohio Valley, the SLP shows much larger spatial scale (Fig. 6a) and the 500-hPa geopotential height also shows much longer wavelength (Fig. 6c). The structure of the height anomalies (Fig. 6d) is distinctly different from those associated with zonally oriented synoptic-scale Rossby wave trains (see Wallace et al. 1988; Fig. 4). Unlike their Ohio Valley counterparts, cyclones associated with precipitation extremes in this region are located mostly to the north and northwest of the region with extreme precipitation. The most important feature is a long corridor of moisture transport (Figs. 6i,j) between the aforementioned dipole structures over the eastern Pacific resembling an atmospheric river. The composite vertically integrated water vapor flux value (over 400 kg m−1 s−1 over the eastern Pacific) is also consistent with the existence of an atmospheric river (e.g., Guan and Waliser 2015). Comparing Figs. 6a and 6b, it is clear that the long corridor for enhanced moisture transport is due to the superposition of the cyclonic and anticyclonic anomalies on top of the climatological Aleutian low and east Pacific subtropical high, giving rise to a long corridor of strong pressure gradient that provides the strong low-level flow for the enhanced moisture transport. When this strong moisture transport reaches the Cascade Range and the Rocky Mountains, it is forced to go upward by the mountains, producing heavy orographic precipitation. Thus, in this region, rising motion is not directly forced by the cyclones themselves, but is mostly due to the fixed orography. This explains the very different cyclone distributions relative to the location of the precipitation extremes in these two regions. Comparing with the preceding subsection, we see that from the moisture budget to weather patterns, the northwestern United States shows very different physical mechanisms from the Ohio Valley. These results also indicate that anticyclones are not always associated with sunny days. They do have positive influence on extreme precipitations, perhaps not locally but on the larger scale.

b. Moisture budget

In section 3, when we examined the winter mean moisture budget modulated by the strength of cyclone activity, we found that variations in moisture convergence dominate over the Ohio Valley, while changes in moisture advection dominate over the northwestern United States. Here we will examine whether similar differences can be found for the extreme precipitation days.

The interpretation of the moisture budget for individual days is not as straightforward as that for the seasonal mean. Over a season, the change in storage [third term on the left-hand side of Eq. (1)] averages out to be much smaller than all the other terms and can be neglected. However, over one day, the change in storage can be as large as the other terms and must be taken into account. Figure 7e shows the composite of this term for the extreme precipitation days for Ohio Valley. Between the beginning and end of the day, there is significant reduction in precipitable water to the southwest of the region, and significant increase to the east of the region, consistent with the northeastward propagation of the synoptic-scale system bringing warm and moist air northeastward toward the east coast of the United States, with cold dry air moving into the southern part of the United States behind a cold front.

Fig. 7.

Composite of moisture budget anomalies for 145 extreme precipitation days (mm day−1): (a),(b) moisture advection; (c),(d) dynamical moisture convergence; and (e),(f) change in storage for (left) the Ohio Valley and (right) the northwestern United States.

Fig. 7.

Composite of moisture budget anomalies for 145 extreme precipitation days (mm day−1): (a),(b) moisture advection; (c),(d) dynamical moisture convergence; and (e),(f) change in storage for (left) the Ohio Valley and (right) the northwestern United States.

The decomposition of the convergence of moisture flux anomaly (Fig. 4h) into moisture advection and dynamical moisture convergence is shown in Figs. 7a and 7c, respectively. Both exhibit positive contribution over the Ohio Valley. The contribution from the advection term appears to be larger than that from the convergence term. However, in comparing Figs. 7a and 7e, it is clear that a significant part of moisture advection is associated with the aforementioned northeastward displacement of the warm moist air and does not necessarily contribute to the heavy precipitation. Note that Figs. 7a,c and 4h appear to be much noisier than Fig. 7e. This could partly be due to the noisy distribution of precipitation, but it should be noted that the moisture flux terms are averaged over four 6-hourly instantaneous reanalysis snapshots, whereas the change in precipitation represents the integral of Q¯/t over a 24-h period and is thus expected to be smoother. Nevertheless, Fig. 7c shows that dynamical convergence clearly significantly contributes to the moisture budget for extreme precipitation events over the Ohio Valley.

Similar decomposition for northwestern United States is shown in the right panels of Fig. 7. Clearly, consistent with the monthly mean budget (Fig. 2), over this region advection (Fig. 7b) dominates, with the contribution from dynamical convergence (Fig. 7d) being opposite to the sign of the moisture flux convergence (Fig. 6f), similar to the seasonal moisture budget discussed above. In addition, in this region advection does not lead to northeastward movement of the warm and moist air mass, with little change in storage found (Fig. 7f). This is likely due to the moisture being largely exhausted by orographic precipitation after uplift by the mountains. Thus Fig. 7 confirms that for individual extreme precipitation events, moisture advection dominates over the northwestern United States, while dynamical moisture convergence contributes significantly over the Ohio Valley, qualitatively consistent with the conclusion reached above.

c. Variations in moisture transport

1) Ohio Valley

Figures 4 and 6 summarize the mean synoptic situation of the extreme precipitation cases and compare them with climatology. However, not all cases are the same, and we will examine some differences between these cases. For the Ohio Valley, an EOF analysis is conducted on the column-integrated water vapor flux from the 145 extreme days for the continental part of the United States by concatenating the zonal and meridional components of each case into a single vector. The leading three EOFs explain 28.0%, 16.4%, and 13.4% of the variance, respectively. EOF4 and higher modes only explain 7% or less of the variance each and will not be discussed. The value of the principal component for each day quantifies how strongly the moisture transport pattern on that particular extreme day resembles the sum of that EOF and the mean pattern. High and low composites shown in these figures are based on the top and bottom thirds of the principal component values. These composites provide a range of likely synoptic scenarios associated with the extreme precipitation days, and may be useful for identifying the potential for the occurrence of heavy precipitation. The differences between high and low composites that show the structure of the EOFs (as well as SLP composites) are shown in Figs. S4–S8 of the online supplemental material.

The composites for EOF1 generally show a Rossby wave phase shift between the high and low composites (Figs. 8a,b). In the high composite, the trough is located at about 110°W and the ridge at about 80°W, whereas in the low composite the trough is at around 97°W and the ridge at about 72°W. In the high composite, more cyclones occur toward the southwest of the box (Fig. 8a, shading) with the low pressure center located in Texas (online supplemental Fig. S4a), hence the warm conveyor belt (WCB) carries more moisture into the southern part of the box (Fig. 8c); in the low composite, cyclones are located farther northeast (Fig. 8b and Fig. S4b), resulting in more moisture convergence over the northeastern United States (Fig. 8d). The moisture fluxes for both cases are oriented southwest to northeast and are rather strong (over 500 kg m−1 s−1), with the main difference being an east–west shift in their location (Figs. 8e,f).

Fig. 8.

(left) High and (right) low composites for the EOF1 of vapor flux within the 145 extreme precipitation days defined for the box over the Great Lakes and Ohio Valley for (a),(b) 500-hPa geopotential height (contours; contour interval 100 m) and cyclone density (shading; day−1); (c),(d) convergence of moisture flux, which is the sum of moisture convergence and advection (mm day−1); and (e),(f) column-integrated water vapor flux (vectors; kg m−1 s−1) and the norm of the vectors (shading; kg m−1 s−1).

Fig. 8.

(left) High and (right) low composites for the EOF1 of vapor flux within the 145 extreme precipitation days defined for the box over the Great Lakes and Ohio Valley for (a),(b) 500-hPa geopotential height (contours; contour interval 100 m) and cyclone density (shading; day−1); (c),(d) convergence of moisture flux, which is the sum of moisture convergence and advection (mm day−1); and (e),(f) column-integrated water vapor flux (vectors; kg m−1 s−1) and the norm of the vectors (shading; kg m−1 s−1).

EOF2 contrasts stronger and weaker moisture flux, as can be seen from the 500-hPa geopotential height (Figs. 9a,b) and the norm of the moisture flux (Figs. 9e,f). In the low composite both the height gradient and wave amplitude at 500 hPa are stronger, implying stronger winds, which is consistent with stronger moisture flux, while the wave amplitude is weaker in the high composite. Even though the moisture flux flowing into the box is weaker in the high composite, the outgoing flux is even weaker (Fig. 9e), resulting in strong moisture convergence within the box (Fig. 9c). On the other hand, even though the low composite shows much stronger moisture flux, much of the moisture is carried through the Ohio Valley toward the Northeast (Fig. 9d). Thus, stronger incoming moisture flux does not necessarily imply stronger moisture convergence within the region.

Fig. 9.

As in Fig. 8, but for the EOF2 mode.

Fig. 9.

As in Fig. 8, but for the EOF2 mode.

EOF3 highlights the orientation of weather patterns (more meridional or more zonal). In the high composite, the wave amplitude is stronger, the wavelength is shorter (Fig. 10a), and the moisture flux is more poleward (Fig. 10e), whereas in the low composite the wave amplitude is weaker and the moisture flux is more zonal (Figs. 10b,f). Cyclones also expand more poleward in the high composite and spread eastward in the low composite (Figs. 10a,b). The moisture convergence is also affected by the different spatial orientations: in the high composite, it is concentrated within the box (Fig. 10c) whereas in the low composite it extends farther eastward (Fig. 10d).

Fig. 10.

As in Fig. 8, but for the EOF3 mode.

Fig. 10.

As in Fig. 8, but for the EOF3 mode.

In summary, Figs. 810 show that there is a range of synoptic situations that can lead to heavy precipitation over the Ohio Valley. The trough/ridge axes can occur over a range of longitudes (EOF1), the magnitude of moisture flux can vary (EOF2), and the orientation of the moisture transport can be more meridional or zonal (EOF3). Nevertheless, while there are differences between the cases, all groups display a Rossby wave train at 500 hPa with an amplified ridge near the U.S. Atlantic coast, and strong moisture flux flowing poleward across the Gulf Coast.

2) Northwestern United States

An EOF analysis has also been conducted for the northwestern U.S. extreme precipitation days, using a domain covering both the land and ocean (25°–60°N, 140°–100°W). For this case, the leading EOFs are even better separated: EOF1 explains 44.1% of the variance, and EOF2 19.3%. EOF3 and higher modes explain only 7.4% or less and are not discussed here. The EOF patterns are robust if we expand the domain farther (by 10°) into the Pacific.

EOF1 highlights the spatial orientation of the moisture transport. In the high composite, the 500-hPa geopotential height suggests strong southwesterly wind over the eastern Pacific (Fig. 11a). The strong moisture flux flows from the low latitudes of central Pacific into the box (Fig. 11e). Cyclones are located to the west of the box off the Washington–Oregon coast (Fig. 11a). In the low composite, the geopotential height contours are much more zonal, and the gradient is weaker over the Pacific (Fig. 11b). The comparatively weaker moisture flux flows eastward into the box at a higher latitude compared to the high composite (Figs. 11e,f). The cyclone density shows two local maxima, indicating a main occluded low in the Gulf of Alaska and a weak secondary cyclone near the U.S.–Canada border (Fig. 11b and online supplemental Fig. S7b). Both the high and low composites give rise to strong moisture convergence in the Northwest, with that in the high composite being slightly stronger (Figs. 11c,d and Fig. S7l); however, in the high composite strong moisture convergence extends southward into Northern California.

Fig. 11.

As in Fig. 8, but for the box over the northwestern United States.

Fig. 11.

As in Fig. 8, but for the box over the northwestern United States.

EOF2 highlights the strength of the moisture transport. The high composite shows strong moisture transport of over 600 kg m−1 s−1 similar to that of an atmospheric river (Fig. 12e), with strong cyclonic circulation to the northwest and anticyclonic circulation to the southeast of the moisture transport corridor (Fig. S8a), giving rise to strong moisture flux flowing northeastward from the lower latitudes of the central Pacific. However, in the low composite (Fig. 12f), the Pacific moisture transport corridor is much shorter in distance and the magnitude of the flux is much lower (only up to about 400 kg m−1 s−1), and is associated with partially inland weak cyclones located north of the moisture transport without an enhanced anticyclone to the south (Fig. 12b and Fig. S8b). Unlike the Ohio Valley cases, here stronger incoming moisture transport gives rise to stronger moisture convergence (Figs. 12c,d), likely related to the impacts of the mountain ranges in this region. In summary, Figs. 11 and 12 highlight the fact that there is also a range of synoptic scenarios that can give rise to heavy precipitation over northwestern United States.

Fig. 12.

As in Fig. 11, but for the EOF2 mode.

Fig. 12.

As in Fig. 11, but for the EOF2 mode.

5. Discussion and conclusions

Our previous study showed the significant impacts of storm track variations on extreme precipitation frequency in the Ohio Valley and the Pacific Northwest. Here we compare the physical mechanisms generating extreme precipitation within these two regions. By examining the extreme precipitation days and moisture budgets in those two regions, we find that the extreme precipitation days in the Ohio Valley display typical zonally oriented midlatitude synoptic (Rossby) wave train patterns at 500-hPa level and classic cyclone structure on the ground. Moisture mainly comes from the Gulf of Mexico between the cyclonic circulation over the central United States and the anticyclonic circulation close to the Atlantic coast. The low pressure center and fronts on the ground lead to convergence of air and upward motion to the northeast of the cyclone center. Our results also show that the existence of a southwestward extended ridge close to the Atlantic coast is a crucial ingredient leading to enhanced northward moisture transport from the Gulf. On the other hand, the extreme precipitation days in the northwestern United States show a cyclonic and anticyclonic circulation pair aligned northwest to southeast that is different from the signature of a zonally oriented synoptic-scale Rossby wave train. Strong northeastward moisture flux resembling an atmospheric river arises between such a dipole structure. As the flux from the Pacific meets the Cascade Range and the Rocky Mountains in the northwestern United States, the moisture is forced upward to form heavy orographic precipitation.

At the seasonal scale, to understand the moisture source for seasons with frequent extreme precipitation events, we decompose the total moisture flux convergence into two terms: moisture advection and moisture tendency due to dynamical convergence. We find that the dynamical convergence dominates the moisture budget in the Ohio Valley, whereas in the northwestern United States both are important but mostly cancel each other out, with the advection being slightly stronger. Evaporation is found to be enhanced over subtropical Atlantic east of Cuba and along the coast of northwestern United States, along the path of the enhanced moisture transport into the extreme precipitation regions. Future work will be needed to quantify the contribution from enhanced evaporation by directly computing moisture source following backward Lagrangian trajectories from regions of enhanced moisture convergence.

We further decompose each term into the mean and transient components. Within most of the domain, the mean terms dominate over the transient terms. The exceptions are over the Gulf of Mexico and nearby coastal states, where the transient advection term is comparable with the mean advection term and they mostly cancel each other. Over the Ohio Valley, we find that it is the variations of the mean dynamical convergence instead of the variations of precipitable water that contribute most to variations in mean moisture convergence.

Even though a cyclone is by itself a high-frequency transient phenomenon, it is not necessary that its contribution to the moisture budget must stay within the corresponding high-frequency range. Stronger and more frequent cyclones can increase the seasonal mean dynamical convergence, hence enhancing the seasonal mean moisture convergence. Similarly, variations in cyclone activity also contribute to changes in the seasonal mean flow, affecting the mean moisture advection. Our composites of the circulation during extreme precipitation days described above clearly demonstrate the importance of cyclone (and anticyclone) activity during these events. Nevertheless, what factors control whether mean or transient moisture transport dominates over a region should be further investigated.

Seasonal mean budgets and composites made from all extreme precipitation days highlight the average conditions for these events, but not all events are the same. We examine the differences between these events by conducting EOF analyses on the column-integrated vapor flux data. All three leading EOFs in the Ohio Valley show variations of midlatitude synoptic waves, with different wave phases and amplitudes, giving rise to different longitudinal location, orientation, and strength of the moisture flux. Nevertheless, all EOFs display a Rossby wave train at 500 hPa with an amplified ridge near the U.S. Atlantic coast, and strong moisture flux flowing poleward across the Gulf Coast. One interesting observation is that over this region, stronger incoming moisture flux may not necessarily give rise to stronger moisture flux convergence, since the latter depends also on the location of the synoptic weather system including the cyclone center and the fronts. For the northwestern United States, the two leading EOFs also highlight shifts in spatial location, orientation, and strength of moisture transport. But in this region, stronger incoming moisture flux generally leads to stronger moisture flux convergence, given the flux is met with the spatially fixed mountain ranges that force the incoming airstream to ascend and cool.

While some of our results are largely consistent with those of previous studies, this work extends previous studies by directly contrasting the differences between the dynamics governing extreme precipitation events in two different regions. Our results also highlight the importance of the southwestward extended ridge near the Atlantic coast for the Ohio Valley cases, and show that not only are the cyclones important, but the accompanying anticyclones are also important in channeling a strong moisture flux toward both regions.

In this study, we have examined conditions giving rise to winter extreme precipitation events in two regions in detail, and found that even though in both regions these events are due to cyclone and anticyclone activity, the detailed dynamics giving rise to these events are quite different. It would be of interest to examine similar events in other regions that may exhibit potentially different dynamics, for example the U.S. east coast and western Europe. In addition, given the importance of this topic, it will be of interest to see how well climate models simulate the different relationships between cyclone activity and precipitation extremes in the different regions, and whether there will be a significant change of these relationships between the current and future climates.

Acknowledgments

The authors thank three anonymous reviewers for comments that helped to improve the paper. We acknowledge the European Centre for Medium-Range Weather Forecasts (ECMWF), NASA Global Modeling and Assimilation Office (GMAO), and the Climate Prediction Center (CPC) for making the ERA-Interim (https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era-interim), MERRA-2 reanalysis products (https://gmao.gsfc.nasa.gov/reanalysis/MERRA-2/data_access/), and CPC gauged-based analysis of daily precipitation data (1948–2006: https://www.esrl.noaa.gov/psd/data/gridded/data.unified.daily.conus.html; 2007 onward: https://www.esrl.noaa.gov/psd/data/gridded/data.unified.daily.conus.rt.html) available to us, respectively. We thank Albert Yau for providing the tracked cyclone trajectory data. This work is supported by NASA Grant NNX16AG32G. The research carried out at the Jet Propulsion Laboratory, California Institute of Technology, was under a contract with the same NASA grant.

REFERENCES

REFERENCES
Alexander
,
L. V.
,
S. F. B.
Tett
, and
T.
Jonsson
,
2005
:
Recent observed changes in severe storms over the United Kingdom and Iceland
.
Geophys. Res. Lett.
,
32
,
L13704
, https://doi.org/10.1029/2005GL022371.
Ashley
,
W. S.
, and
A. W.
Black
,
2008
:
Fatalities associated with nonconvective high-wind events in the United States
.
J. Appl. Meteor. Climatol.
,
47
,
717
725
, https://doi.org/10.1175/2007JAMC1689.1.
Blackmon
,
M. L.
,
J. M.
Wallace
,
N.-C.
Lau
, and
S. L.
Mullen
,
1977
:
An observational study of the Northern Hemisphere wintertime circulation
.
J. Atmos. Sci.
,
34
,
1040
1053
, https://doi.org/10.1175/1520-0469(1977)034<1040:AOSOTN>2.0.CO;2.
Blackmon
,
M. L.
,
Y.-H.
Lee
, and
J. M.
Wallace
,
1984a
:
Horizontal structure of 500 mb height fluctuations with long, intermediate and short time scales
.
J. Atmos. Sci.
,
41
,
961
980
, https://doi.org/10.1175/1520-0469(1984)041<0961:HSOMHF>2.0.CO;2.
Blackmon
,
M. L.
,
Y.-H.
Lee
, and
H.-H.
Hsu
,
1984b
:
Time variation of 500 mb height fluctuations with long, intermediate and short time scales as deduced from lag-correlation statistics
.
J. Atmos. Sci.
,
41
,
981
991
, https://doi.org/10.1175/1520-0469(1984)041<0981:TVOMHF>2.0.CO;2.
Bosart
,
L. F.
,
1981
:
The Presidents’ Day snowstorm of 18–19 February 1979: A subsynoptic-scale event
.
Mon. Wea. Rev.
,
109
,
1542
1566
, https://doi.org/10.1175/1520-0493(1981)109%3C1542:TPDSOF%3E2.0.CO;2.
Bosilovich
,
M. G.
,
F. R.
Robertson
,
L.
Takacs
,
A.
Molod
, and
D.
Mocko
,
2017
:
Atmospheric water balance and variability in the MERRA-2 reanalysis
.
J. Climate
,
30
,
1177
1196
, https://doi.org/10.1175/JCLI-D-16-0338.1.
Cardone
,
V. J.
,
R. E.
Jensen
,
D. T.
Resio
,
V. R.
Swail
, and
A. T.
Cox
,
1996
:
Evaluation of contemporary ocean wave models in rare extreme events: The “Halloween storm” of October 1991 and the “storm of the century” of March 1993
.
J. Atmos. Oceanic Technol.
,
13
,
198
230
, https://doi.org/10.1175/1520-0426(1996)013<0198:EOCOWM>2.0.CO;2.
Chang
,
E. K. M.
,
1993
:
Downstream development of baroclinic waves as inferred from regression analysis
.
J. Atmos. Sci.
,
50
,
2038
2053
, https://doi.org/10.1175/1520-0469(1993)050<2038:DDOBWA>2.0.CO;2.
Chang
,
E. K. M.
, and
A. M. W.
Yau
,
2016
:
Northern Hemisphere winter storm track trends since 1959 derived from multiple reanalysis datasets
.
Climate Dyn.
,
47
,
1435
1454
, https://doi.org/10.1007/s00382-015-2911-8.
Chang
,
E. K. M.
,
S.
Lee
, and
K. L.
Swanson
,
2002
:
Storm track dynamics
.
J. Climate
,
15
,
2163
2183
, https://doi.org/10.1175/1520-0442(2002)015<02163:STD>2.0.CO;2.
Chang
,
E. K. M.
,
Y.
Guo
, and
X.
Xia
,
2012
:
CMIP5 multimodel ensemble projection of storm track change under global warming
.
J. Geophys. Res.
,
117
,
D23118
, https://doi.org/10.1029/2012JD018578.
Chang
,
E. K. M.
,
C.-G.
Ma
,
C.
Zheng
, and
A. M. W.
Yau
,
2016
:
Observed and projected decrease in Northern Hemisphere extratropical cyclone activity in summer and its impacts on maximum temperature
.
Geophys. Res. Lett.
,
43
,
2200
2208
, https://doi.org/10.1002/2016GL068172.
Colle
,
B. A.
,
Z.
Zhang
,
K. A.
Lombardo
,
E.
Chang
,
P.
Liu
, and
M.
Zhang
,
2013
:
Historical evaluation and future prediction of eastern North American and western Atlantic extratropical cyclones in the CMIP5 models during the cool season
.
J. Climate
,
26
,
6882
6903
, https://doi.org/10.1175/JCLI-D-12-00498.1.
Easterling
,
D. R.
,
G. A.
Meehl
,
C.
Parmesan
,
S. A.
Changnon
,
T. R.
Karl
, and
L. O.
Mearns
,
2000
:
Climate extremes: Observations, modeling, and impacts
.
Science
,
289
,
2068
2074
, https://doi.org/10.1126/science.289.5487.2068.
Eldardiry
,
H.
,
A.
Mahmood
,
X.
Chen
,
F.
Hossain
,
B.
Nijssen
, and
D. P.
Lettenmaier
,
2019
:
Atmospheric river–induced precipitation and snowpack during the western United States cold season
.
J. Hydrometeor.
,
20
,
613
630
, https://doi.org/10.1175/JHM-D-18-0228.1.
Feser
,
F.
,
M.
Barcikowska
,
O.
Krueger
,
F.
Schenk
,
R.
Weisse
, and
L.
Xia
,
2015
:
Storminess over the North Atlantic and northwestern Europe—A review
.
Quart. J. Roy. Meteor. Soc.
,
141
,
350
382
, https://doi.org/10.1002/qj.2364.
Frankoski
,
N. J.
, and
A. T.
DeGaetano
,
2011
:
An East Coast winter storm precipitation climatology
.
Int. J. Climatol.
,
31
,
802
814
, https://doi.org/10.1002/joc.2121.
Gelaro
,
R.
, and et al
,
2017
:
The Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2)
.
J. Climate
,
30
,
5419
5454
, https://doi.org/10.1175/JCLI-D-16-0758.1.
Grise
,
K. M.
,
S.-W.
Son
, and
J. R.
Gyakum
,
2013
:
Intraseasonal and interannual variability in North America storm tracks and its relationship to equatorial Pacific variability
.
Mon. Wea. Rev.
,
141
,
3610
3625
, https://doi.org/10.1175/MWR-D-12-00322.1.
Guan
,
B.
, and
D. E.
Waliser
,
2015
:
Detection of atmospheric rivers: Evaluation and application of an algorithm for global studies
.
J. Geophys. Res. Atmos.
,
120
,
12 514
12 535
, https://doi.org/10.1002/2015JD024257.
Guan
,
B.
,
N. P.
Molotch
,
D. E.
Waliser
,
E. J.
Fetzer
, and
P. J.
Neiman
,
2010
:
Extreme snowfall events linked to atmospheric rivers and surface air temperature via satellite measurements
.
Geophys. Res. Lett.
,
37
,
L20401
, https://doi.org/10.1029/2010GL044696.
Hodges
,
K. I.
,
1994
:
A general method for tracking analysis and its application to meteorological data
.
Mon. Wea. Rev.
,
122
,
2573
2586
, https://doi.org/10.1175/1520-0493(1994)122<2573:AGMFTA>2.0.CO;2.
Hodges
,
K. I.
,
1999
:
Adaptive constraints for feature tracking
.
Mon. Wea. Rev.
,
127
,
1362
1373
, https://doi.org/10.1175/1520-0493(1999)127%3C1362:ACFFT%3E2.0.CO;2.
Hoskins
,
B. J.
, and
K. I.
Hodges
,
2002
:
New perspectives on the Northern Hemisphere winter storm tracks
.
J. Atmos. Sci.
,
59
,
1041
1061
, https://doi.org/10.1175/1520-0469(2002)059<1041:NPOTNH>2.0.CO;2.
Kunkel
,
K. E.
, and
J. R.
Angel
,
1999
:
Relationship of ENSO to snowfall and related cyclone activity in the contiguous United States
.
J. Geophys. Res.
,
104
,
19 425
19 434
, https://doi.org/10.1029/1999JD900010.
Kunkel
,
K. E.
,
D. R.
Easterling
,
D. A. R.
Kristovich
,
B.
Gleason
,
L.
Stoecker
, and
R.
Smith
,
2012
:
Meteorological causes of the secular variations in observed extreme precipitation events for the conterminous United States
.
J. Hydrometeor.
,
13
,
1131
1141
, https://doi.org/10.1175/JHM-D-11-0108.1.
Lau
,
N.-C.
,
1988
:
Variability of the observed midlatitude storm tracks in relation to low-frequency changes in the circulation pattern
.
J. Atmos. Sci.
,
45
,
2718
2743
, https://doi.org/10.1175/1520-0469(1988)045<2718:VOTOMS>2.0.CO;2.
Leathers
,
D. J.
,
B.
Yarnal
, and
M. A.
Palecki
,
1991
:
The Pacific/North American teleconnection pattern and United States climate. Part I: Regional temperature and precipitation associations
.
J. Climate
,
4
,
517
528
, https://doi.org/10.1175/1520-0442(1991)004<0517:TPATPA>2.0.CO;2.
Ma
,
C. G.
, and
E. K.
Chang
,
2017
:
Impacts of storm track variations on winter time extreme weather events over the continental United States
.
J. Climate
,
30
,
4601
4624
, https://doi.org/10.1175/JCLI-D-16-0560.1.
Meehl
,
G. A.
,
J. M.
Arblaster
, and
C.
Tebaldi
,
2005
:
Understanding future patterns of increased precipitation intensity in climate model simulations
.
Geophys. Res. Lett.
,
32
,
L18719
, https://doi.org/10.1029/2005GL023680.
Montroy
,
D. L.
,
M. B.
Richman
, and
P. J.
Lamb
,
1998
:
Observed nonlinearities of monthly teleconnections between tropical Pacific sea surface temperature anomalies and central and eastern North American precipitation
.
J. Climate
,
11
,
1812
1835
, https://doi.org/10.1175/1520-0442(1998)011<1812:ONOMTB>2.0.CO;2.
Neu
,
U.
, and et al
,
2013
:
IMILAST: A community effort to intercompare extratropical cyclone detection and tracking algorithms
.
Bull. Amer. Meteor. Soc.
,
94
,
529
547
, https://doi.org/10.1175/BAMS-D-11-00154.1.
Ning
,
L.
, and
R. S.
Bradley
,
2015
:
Winter climate extremes over the northeastern United States and southeastern Canada and teleconnections with large-scale modes of climate variability
.
J. Climate
,
28
,
2475
2493
, https://doi.org/10.1175/JCLI-D-13-00750.1.
Pall
,
P.
,
T.
Aina
,
D. A.
Stone
,
P. A.
Stott
,
T.
Nozawa
,
A. G. J.
Hilberts
,
D.
Lohmann
, and
M. R.
Allen
,
2011
:
Anthropogenic greenhouse gas contribution to flood risk in England and Wales in autumn 2000
.
Nature
,
470
,
382
385
, https://doi.org/10.1038/nature09762.
Peixoto
,
J. P.
, and
A. H.
Oort
,
1992
: Physics of Climate. American Institute of Physics, 520 pp.
Petterssen
,
S.
,
1956
: Weather Analysis and Forecasting. Vol. 1. McGraw-Hill, 428 pp.
Pfahl
,
S.
, and
H.
Wernli
,
2012
:
Quantifying the relevance of cyclones for precipitation extremes
.
J. Climate
,
25
,
6770
6780
, https://doi.org/10.1175/JCLI-D-11-00705.1.
Polly
,
J. B.
, and
W. B.
Rossow
,
2016
:
Cloud radiative effects and precipitation in extratropical cyclones
.
J. Climate
,
29
,
6483
6507
, https://doi.org/10.1175/JCLI-D-15-0857.1.
Raible
,
C. C.
,
P. M.
Della-Marta
,
C.
Schwierz
,
H.
Wernli
, and
R.
Blender
,
2008
:
Northern Hemisphere extratropical cyclones: A comparison of detection and tracking methods and different reanalysis
.
Mon. Wea. Rev.
,
136
,
880
897
, https://doi.org/10.1175/2007MWR2143.1.
Ralph
,
F. M.
,
P. J.
Neiman
, and
G. A.
Wick
,
2004
:
Satellite and CALJET aircraft observations of atmospheric rivers over the eastern North Pacific Ocean during the winter of 1997/98
.
Mon. Wea Rev.
,
132
,
1721
1745
, https://doi.org/10.1175/1520-0493(2004)132%3C1721:SACAOO%3E2.0.CO;2.
Ralph
,
F. M.
,
P. J.
Neiman
,
G. A.
Wick
,
S. I.
Gutman
,
M. D.
Dettinger
,
D. R.
Cayan
, and
A. B.
White
,
2006
:
Flooding on California’s Russian River: Role of atmospheric rivers
.
Geophys. Res. Lett.
,
33
,
L13801
, https://doi.org/10.1029/2006GL026689.
Rappaport
,
E. N.
,
2000
:
Loss of life in the United States associated with recent Atlantic tropical cyclones
.
Bull. Amer. Meteor. Soc.
,
81
,
2065
2073
, https://doi.org/10.1175/1520-0477(2000)081<2065:LOLITU>2.3.CO;2.
Rudeva
,
I.
, and
S. K.
Gulev
,
2011
:
Composite analysis of North Atlantic extratropical cyclones in NCEP–NCAR reanalysis data
.
Mon. Wea. Rev.
,
139
,
1419
1446
, https://doi.org/10.1175/2010MWR3294.1.
Seager
,
R.
, and et al
,
2014
:
Dynamical and thermodynamical causes of large-scale changes in the hydrological cycle over North America in response to global warming
.
J. Climate
,
27
,
7921
7948
, https://doi.org/10.1175/JCLI-D-14-00153.1.
Sinclair
,
M. R.
,
1997
:
Objective identification of cyclones and their circulation intensity, and climatology
.
Wea. Forecasting
,
12
,
595
612
, https://doi.org/10.1175/1520-0434(1997)012<0595:OIOCAT>2.0.CO;2.
Ting
,
M.
,
R.
Seager
,
C.
Li
,
H.
Liu
, and
N.
Henderson
,
2018
:
Mechanism of future spring drying in the southwestern United States in CMIP5 models
.
J. Climate
,
31
,
4265
4279
, https://doi.org/10.1175/JCLI-D-17-0574.1.
Touma
,
D.
,
A. M.
Michalak
,
D. L.
Swain
, and
N. S.
Diffenbaugh
,
2018
:
Characterizing the spatial scales of extreme daily precipitation in the United States
.
J. Climate
,
31
,
8023
8037
, https://doi.org/10.1175/JCLI-D-18-0019.1.
Wallace
,
J. M.
,
G.-H.
Lim
, and
M. L.
Blackmon
,
1988
:
Relationship between cyclone tracks, anticyclone tracks, and baroclinic waveguides
.
J. Atmos. Sci.
,
45
,
439
462
, https://doi.org/10.1175/1520-0469(1988)045<0439:RBCTAT>2.0.CO;2.
Wong
,
S.
,
A. D.
Del Genio
,
T.
Wang
,
B.
Kahn
,
E. J.
Fetzer
, and
T. S.
L’Ecuyer
,
2016
:
Responses of tropical ocean clouds and precipitation to the large-scale circulation: Atmospheric-water-budget-related phase space and dynamical regimes
.
J. Climate
,
29
,
7127
7143
, https://doi.org/10.1175/JCLI-D-15-0712.1.
Wong
,
S.
,
C. M.
Naud
,
B. H.
Kahn
,
L.
Wu
, and
E. J.
Fetzer
,
2018
:
Coupling of precipitation and cloud structures in oceanic extratropical cyclones to large-scale moisture transport
.
J. Climate
,
31
,
9565
9584
, https://doi.org/10.1175/JCLI-D-18-0115.1.
Zhang
,
F.
,
C.
Snyder
, and
R.
Rotunno
,
2002
:
Mesoscale predictability of the “surprise” snowstorm of 24–25 January 2000
.
Mon. Wea. Rev.
,
130
,
1617
1632
, https://doi.org/10.1175/1520-0493(2002)130%3C1617:MPOTSS%3E2.0.CO;2.
Zhu
,
Y.
, and
R. E.
Newell
,
1994
:
Atmospheric rivers and bombs
.
Geophys. Res. Lett.
,
21
,
1999
2002
, https://doi.org/10.1029/94GL01710.
For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).