1. Introduction
It is well established that extreme precipitation frequency tends to increase as the atmosphere warms (Donat et al. 2013; IPCC 2014; O’Gorman 2015). Convection and strong extratropical cyclones (ETCs) tend to produce higher precipitation rates in a warmer climate (Bengtsson et al. 2009; Catto et al. 2019). Tropical cyclones (TCs) have the potential to strengthen to greater intensities, producing more rainfall and potentially higher wind speeds (Webster et al. 2005; Elsner et al. 2008; Knutson et al. 2010; Liu et al. 2019). Also, atmospheric rivers (ARs) are expected to produce more intense precipitation (Espinoza et al. 2018). These changes are consistent with the Clausius–Clapeyron scaling of moisture availability with temperature (e.g., Trenberth 1999; Allen and Ingram 2002; Pall et al. 2007).
The goal of this paper is to investigate changes in mid-Atlantic and Northeast U.S. extreme precipitation in a setting in which these moisture- and TC-related changes are particularly relevant: fall seasonal changes in large-scale extreme precipitation. Previous studies have identified fall as the season, or one of the seasons, with the largest increasing extreme precipitation trends over the Northeast (Huang et al. 2017, 2018; Howarth et al. 2019; Kunkel et al. 2020). Howarth et al. (2019) attribute this to increases in post-1996 tropical moisture– and TC-related events. Evidence also suggests that extreme precipitation events sustained by AR-like circulations may experience greater intensity increases than other weather events in a warming climate (Hatsuzuka et al. 2021).
During fall, the mid-Atlantic and Northeast lie at the confluence of tropical and extratropical influences: recurving TCs and tropical moisture from the south, and fronts and extratropical cyclones from the north (e.g., Kunkel et al. 2012). Extreme precipitation studies have typically been from the perspective of weather type attribution to TCs or ETCs, with trends attributed to frequency changes of these systems. But since atmospheric moisture content and transport is such an important and sensitive factor under climate change, it is useful also to investigate changes in extreme precipitation from a moisture perspective. This study will take this approach by quantifying the changes in integrated water vapor transport (IVT), moisture content, and wind speed within different extreme precipitation weather types, as well as presenting a range of IVT-related weather types for extreme events and extreme precipitation generally. Within this framework, changes in large-scale fall extreme precipitation will be linked to changes in season-mean atmospheric conditions, the synoptic characteristics of storms, and the distribution of weather types.
The paper will be structured as follows: section 2 contains data and methodology information, section 3 details the frequencies and trends of the weather types, and section 4 describes the precipitation means and trends associated with the weather types. Some particularly extreme events are described synoptically in section 5, and broader conclusions and discussion are given in section 6.
2. Data and methodology
a. Precipitation data
Rainfall data for the U.S. Northeast and mid-Atlantic come from the Global Historical Climate Network–Daily (GHCN-D) dataset (Menne et al. 2012; https://www.ncei.noaa.gov/products/land-based-station/global-historical-climatology-network-daily). To ensure that only high-quality data are used, the stations are filtered according to the following criteria: 1) the stations must have at least 95% daily precipitation data completeness between 1979 and 2019, inclusive (not counting missing values and quality flags), and 2) there must be precipitation observations during the first and last months of the time period (September 1979 and November 2019, respectively). This last requirement is to ensure that the start/end years are not artificially inflated or suppressed, which could disproportionately affect the trend calculation. Once these checks were satisfied, 314 stations remained in the Northeast and mid-Atlantic (Fig. 1). These stations are broadly distributed, with the lowest coverage occurring over high terrain in New York State. Repeating the precipitation and weather type analysis using an even more stringent completeness condition (99%) led to the same results and conclusions as presented in this paper.
Global Historical Climate Network–Daily (GHCN) stations that pass completeness checks (must have 95% completeness in 1979–2019; must have observations from September 1979 and November 2019).
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
At each station, the extreme precipitation threshold is defined as the 99th percentile of annual measurable (≥0.1 mm day−1) daily precipitation (Fig. 2a). On each day, extreme precipitation observations are summed across stations; days falling in the top 20% by this sum are referred to as “extreme precipitation (EP) days.” Filtering out isolated extreme values leaves mostly synoptic-scale or very intense mesoscale events. Using a sum of extreme precipitation also emphasizes precipitation intensity, whereas simply using the number of extreme-reporting stations would bias the ranking toward events that produce widespread but low-grade (∼50–80 mm day−1) extreme precipitation. EP day precipitation is relatively uniform across the study region, with the exception of far northern and western regions, where it is smaller (Fig. 2b). The mean precipitation on a given EP day is obtained by dividing the total precipitation by the number of reporting stations.
(a) The fall 99th percentile of daily precipitation across the Northeast United States. (b) Composite precipitation (all, not just extreme values) on the 169 extreme precipitation (EP) days. These EP days are defined as the days with total extreme precipitation (summed over all stations) above the 80th percentile.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
Sensitivity testing was performed on these two thresholds (station extreme precipitation threshold and EP day threshold) to ensure that our results are robust (see appendix B).
b. Tropical cyclone data
TC locations and wind speed were based on data from the International Best Track Archive for Climate Stewardship (IBTrACS; https://www.ncdc.noaa.gov/ibtracs/), taken at 6-hourly time steps (0000, 0600, 1200, and 1800 UTC). An EP day is considered to have an immediate TC influence (TC-dominant) if a TC center passes within 500 km of an extreme precipitation report, since this is the distance within which mean precipitation and heavy precipitation probability begin increasing (appendix A; Fig. A1). Other weather types may involve more distant TC influences, for instance if extreme IVT connects the extreme precipitation region to the circle of radius 500 km around a TC center.
c. Atmospheric rivers
Past studies have employed a wide range of AR definitions. Some have used satellite-observed integrated water vapor (IWV) values, specifying an AR as a region of >20 mm IWV that is >2000 km in length and <1000 km in width (Ralph et al. 2004; Neiman et al. 2008; Dettinger 2011). Others have used similar geometrical requirements with IVT, citing its close relation to precipitation over complex terrain (e.g., Junker et al. 2008; Neiman et al. 2002, 2013; Ralph et al. 2013).
This study uses a simple set of conditions based on a seasonal, local 95th percentile IVT threshold using data from the ECMWF (ERA5; Hersbach et al. 2020). These are the following:
-
ARs must be ≥2000 km in length.
-
ARs must be ≤1000 km in width.
-
IVT within 500 km of a TC center is not counted toward the AR requirements.
-
The region within 500 km of any IBTrACS system center cannot account for more than one-third of the area of an AR. This threshold was chosen based on examination of weather maps of the events involving both TCs and extended IVT fields.
Length is computed as the maximum pairwise geodesic distance between points, and width as the ratio of area and length. Six extreme IVT thresholds were tested: geographically varying 90th, 92.5th, 95th, 97.5th, and 99th percentile thresholds and a constant threshold (528.7 kg m−1 s−1) defined as the mean 95th percentile threshold over the Northeast. The results of this testing are shown in appendix B (Fig. A2). We ultimately chose to use the constant 95th percentile definition because it captures almost as many AR days as the lower thresholds while offering the advantage of being more physically intuitive.
An EP day is considered to have immediate AR influences (i.e., AR-dominant, unless already TC-dominant) if an AR passes near an extreme precipitation observation, as defined by the minimum pairwise distance between the ERA5 grid point nearest the station and an AR point. Here, “near” means within a box extending three latitude grid points and four longitude grid points ∼80–110 km) away from the near-station point—the distance within which heavy precipitation is maximized (appendix A; Fig. A2). The same method was used for attributing the day’s precipitation to extreme IVT.
d. Weather types
The method of defining weather types relies on proximity to extreme precipitation observations. If an AR intersects at least one extreme precipitation report, or if a TC passes within 500 km of an extreme precipitation report, then the AR and/or TC will be considered immediate influences on that day. On the other hand, if a distant TC is connected to the extreme precipitation region by a continuous band of extreme IVT, then the TC will be considered a remote influence on the extreme precipitation, as in the case of some predecessor rain events (PREs; e.g., Galarneau et al. 2010). In this way, extreme precipitation days are divided into groups and subgroups based largely on the characteristics of their moisture transport and TC activity. These weather types are described in more detail in appendix C.
Instead of focusing on a large set of weather types, which may be underpopulated, we will often aggregate these into three broad categories:
-
TC-dominant weather types (Pure TC, TC–AR combination, TC remnant–AR combination, TC remnants): A TC or TC remnants pass within 500 km of an extreme precipitation observation.
-
AR-dominant weather types (Pure AR, TC-linked AR, TC remnant-linked AR): The requirements for category 1 are not fulfilled, and an AR passes near an extreme precipitation observation.
-
Other weather types (Pure extreme IVT, TC-linked extreme IVT, TC remnant-linked extreme IVT, Unspecified): neither of the above categories is satisfied.
A schematic of these categories is shown in Fig. 3. Within the broader types, subtypes are typically defined by the proximity of TCs and ARs. For instance, a “TC-linked AR” day is an AR-dominant day where the AR is connected to extreme IVT within 500 km of a TC whose center does not lie within 500 km of the extreme precipitation reports. On the other hand, a “TC–AR combination” is both TC-dominant and AR-dominant: both influences are close to the extreme precipitation, either because a pre-existing AR is merging with a TC or because a TC is decaying into an AR. Figure 4 shows an example of the output of the classification algorithm. This October 1996 event saw TC Lili track well offshore south of a strong high. A cutoff low over the mid-Atlantic advected moisture from this system toward the coast and produced some of the most extreme precipitation totals on record. Based on the IVT field and TC track, 20 October 1996 has been classified as a TC-linked AR EP day. More broadly, however, it falls into the AR-dominant category.
Weather typing schematic showing the decision tree used to classify EP days. For example, if a TC center passes within 500 km of an extreme precipitation observation, then the day is considered TC-dominant. TC-dominant takes priority over AR-dominant, and AR-dominant takes priority over Other.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
Example event classification. Depicted are MSLP (contours), 850-hPa winds (arrows), IVT above 250 kg m−1 s−1 (gray shading), TCs (red stars), extreme precipitation observations within the U.S. Northeast and mid-Atlantic (colored dots), and detected ARs (blue shading). The automated weather type classification is shown in the upper left.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
The shortcomings of this approach are as follows: first, without a spatial extent condition, the dataset could include unusually extreme events that only span a few stations. This is only the case for one event: 1 September 2002, when over 350 mm of rain fell in a small area of Maryland. Second, weather types need only be present at one time step of a day to be included in the classification of a multiday event. Third, this weather typing does not account for ETCs or fronts as causes for ARs. Finally, the weather typing does not account for prior TC influences that are no longer listed as storms in IBTrACS. This last issue is partially addressed with a subjective “prior TC influence” analysis in section 4c.
3. Extreme precipitation days and events
a. Precipitation and frequency trends
Figure 5a shows spatially averaged time series of EP day frequency and precipitation. The average precipitation on EP days, shown in green, has not changed over the course of the 41-yr analysis period. EP day frequency, meanwhile, has increased (p = 0.057), with activity being sustained at a higher level since the early 2000s. This helps to explain the increasing trends shown in Fig. 5b: although the amount of precipitation on EP days has not changed, the frequency of EP days has, leading to significant increases in cumulative EP day precipitation. The few exceptions to the pattern of large, increasing trends occur along the eastern shoreline of the Great Lakes. This matches the results found in Howarth et al. (2019), using a similar subset of the GHCN dataset over New England and New York, and suggests that the immediate shorelines of Lakes Erie and Ontario are influenced by different factors than the rest of the mid-Atlantic and Northeast.
(a) Annual time series of extreme precipitation metrics. EP day frequency is shown in light blue bars, with trend line shown in dark blue and linear regression significance test results shown in the legend. Precipitation on each EP day is plotted as green dots, with the corresponding trend line shown in dark green. (b) Trends (% yr−1) in the fall seasonal accumulated precipitation due to EP days. Mann-Kendall test statistical significance is indicated by thick black (p ≤ 0.05) or thin gray (p ≤ 0.10) circles. Slopes are calculated using linear regression.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
That the frequency of EP days is increasing, however, does not necessarily mean that the frequency of EP “events”—defined as groupings of consecutive EP days—is increasing. Instead, EP event duration might account for the larger number of EP days, without a concurrent increase in the frequency of the events themselves. This distinction is discussed in the next section.
b. Sequentiality and the role of complex events
To assess whether EP day frequency trends are truly indicative of increasing event frequency, we group consecutive EP days into events and plot time series of the annual changes in the fall-mean variables F, R, and D above (Fig. 6). The total precipitation trend is strongly positive, with an approximate doubling of the fall seasonal total precipitation occurring on EP days (p < 0.025; Fig. 6a). The event frequency also exhibits a strong, statistically significant positive trend (Fig. 6b). However, event duration and daily precipitation rate seem to be relatively constant (Figs. 6c,d). Although these trends might vary within subsets of the EP days, the overall EP day precipitation increase is associated with increased frequency, not intensity or duration. This is likely due to two factors: first, extreme precipitation is already preselected for intensity, and second, changes in extreme precipitation intensity vary from station to station, averaging out to near zero (not shown).
Time series of EP event (a) seasonal total precipitation, (b) frequency, (c) duration in days, and (d) mean daily precipitation rate. All quantities are spatially averaged over the entire Northeast and mid-Atlantic study region. Trend results are depicted in the legend.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
Top 10 extreme precipitation events (defined as consecutive EP days), sorted by mean total precipitation. Columns give the mean total precipitation; daily mean precipitation above 1 mm (average number of extreme reports per day); duration; starting dates; and included weather types for each event.
Also shown are event duration, starting date, included weather types, and a measure of intensity given by the mean, rather than the sum, of the daily spatial average. Of the 10 events with the most total precipitation, all lasted at least two days, and four lasted at least three days. Seven of these 10 events involve a TC influence of some kind, with five being TC-dominant (TC passes within 500 km of extreme precipitation). Nine of these 10 events have occurred since 1999, and eight of those have occurred since 2004.
Within these TC-influenced events, there is a wide variety of different synoptic patterns and storm configurations. Some of these events are simple cases of landfalling TCs that degrade and dissipate; others involve TCs undergoing extratropical transition or merging with pre-existing ARs. Some may be similar to the PREs studied by Galarneau et al. (2010), in which the TC strengthens distant moisture fluxes before approaching the region itself, leading to a prolonged extreme event. Examples of such events are given in the online supplemental material as well as the data availability statement.
Furthermore, composites show that there are distinct precipitation distributions associated with the synoptic configurations captured by these weather types (Fig. 7). The “Pure” AR type, consisting of ARs with no current TC influence, is associated with enhanced precipitation across much of the study region. Pure TC EP days see very large totals in the south, where TCs are more likely, and TC remnants produce a maximum just north of the TC rainfall maximum. This leaves 31 days with more complex TC and AR influences: the “TC-linked AR,” “TC–AR combination,” and “TC remnant–AR combination” categories shown in Fig. 7. Because these subcategories have small sample sizes and are not obtained using rigorous synoptic attribution methods, they are not intended to individually provide meaningful results in this study. Still, taken together, it is notable that all three of these types produce the most precipitation in eastern Pennsylvania and northern New Jersey, extending into southern New England. This northern mid-Atlantic region is especially prone to very heavy rainfall from events with combined TC–AR influences within the EP days here described. Therefore, such events may be considered important for the risk and climatology of extreme precipitation in these areas.
Mean EP precipitation (mm day−1) associated with different weather types.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
4. Precipitation and synoptic trends of weather types
In this section, the precipitation contributions will be broken down by weather type. This will help to answer a central question of the paper: Which weather types are contributing to the anomalous increasing precipitation from fall EP days? Changes in synoptic properties of weather types will also be investigated in order to determine how the moisture and temperature properties of EP days have (or have not) changed, and how this might affect the aforementioned precipitation trends.
a. Mean time series
Figure 8 depicts the seasonal frequencies and precipitation totals of the different weather types described in the methodology, grouped broadly into AR-dominant, TC-dominant, and Other. Following each weather type name is the percentage of the precipitation mean and trend that it accounts for (note that trend percentage is based on linear regression slope and does not imply statistical significance). Results are as follows:
-
AR-dominant (49.7% mean, 35.3% trend): There are no statistically significant trends in AR-dominant weather type frequency and precipitation (Figs. 8a,b). Activity is initially low, until AR-dominant EP day frequency increases in the late 1980s. Since then, activity has been higher, with large interannual variability in precipitation totals.
-
TC-dominant (28.3% mean, 41.7% trend): The only statistically significant increases in TC-dominant EP day frequency and precipitation are in the TC remnant–AR combination category (n = 11; Figs. 8c,d). This category has 2 days prior to 2000 and 9 days since 2000, while the TC remnants category has 1 day prior to 2000 and 6 days since 2000. The TC-dominant grouping as a whole exhibits increased frequency after the mid- to late 1990s (p = 0.086).
-
Other (21.8% mean, 22.9% trend): Pure extreme IVT-type EP day frequency has increased significantly (p ≤ 0.05; Fig. 8e). Whether this is a function of subclassification-strength AR-like systems, or a ubiquitous increase in IVT strength, remains to be determined.
Time series of fall seasonal frequency and precipitation of different weather types on the 169 Northeast and mid-Atlantic EP days. (a),(b) Weather types defined primarily by the presence of an AR in the vicinity of extreme precipitation. (c),(d) Weather types defined primarily by the presence of a TC or TC remnants in the vicinity of extreme precipitation. (e),(f) Other weather types and unspecified events. Linear regression and Mann–Kendall trend results are shown in the legend, with significant trends shown as dashed lines.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
TC-dominant weather types account for 28.3% of fall EP day precipitation—similar to the value obtained by Agel et al. (2015), despite different extreme definitions and methodologies. Huang et al. (2018) also obtained a similar result to what we see here, except in a different context, concluding that TCs account for 41% of the 1996 Northeast annual extreme precipitation increase.
b. Composite overviews and synoptic trends
The precipitation time series depicted in Figs. 8b, 8d, and 8f are the result of averaging over the large spatial area encompassed by the mid-Atlantic and Northeast United States. In reality, precipitation trends from a given weather type may vary considerably over the ∼11° of latitude and ∼16° of longitude spanned by the study region. In this section, precipitation means and trends will be mapped geographically and synoptic composites will be shown for the AR-dominant, TC-dominant, and Other categories. Means and trends in IVT magnitude, 850-hPa wind speed, total column water vapor, and 1000–500-hPa thickness will be calculated to quantify the changing moisture transport properties of the days with the most extreme precipitation.
1) AR-dominant
AR-dominant EP days are characterized by mean 1800 UTC low pressure over the mid-Atlantic extending northward through western New England and into southeastern Canada (Fig. 9a). High pressure presides southeast of Newfoundland. Sandwiched between these two pressure anomalies, southerly flow brings warm, moist air northward, maximizing IVT off the mid-Atlantic and New England coasts. These events produce large precipitation totals over most of the region; there is little variation, aside from the noticeably lower values over far western regions of Pennsylvania and Virginia as well as all of West Virginia (Fig. 9b). However, trends in this type do exhibit a spatially coherent pattern, defined by widespread significant increasing trends only over northern regions (Fig. 9c). As these trends are in the seasonal total, they could be due to increased frequency of EP days of this type, increased intensity of EP day precipitation, or, more likely, some combination of both factors.
(a) 1800 UTC synoptic composite of AR-dominant EP days. Shown are 1000–500-hPa thickness (dam; dashed contours), MSLP (hPa; solid contours), IVT above 250 kg m−1 s−1 (gray shading), and regions with AR probability above 40% (blue overlay). (b) Mean daily precipitation (mm) on these days, and (c) trend (% yr−1) in season-total precipitation from these days. M-K trend significance is denoted by thick black (p ≤ 0.05) and thin gray (p ≤ 0.10) circles.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
Perhaps the most significant synoptic trend on AR-dominant EP days is a strengthening of the offshore ridge (Fig. 10a). This strengthened ridge is associated with enhanced IVT extending from the Caribbean northward into Atlantic Canada—a change that, in turn, arises from a combination of total column water vapor and low-level wind increases (Figs. 10b–d). Note that while the TCWV increases are sometimes significant, the 850-hPa wind speed increases are not. These changes could help explain the larger and more significant trends in northern regions: since IVT magnitude has increased offshore and over New England, these areas are more likely to see increased moisture flux convergence and precipitation, particularly over high terrain. On the other hand, the increases in AR-dominant EP day precipitation over New York State and western Pennsylvania are collocated with decreased IVT on these days, suggesting that these trends may be associated with other (non-AR) aspects of the extratropical cyclone.
Trends in (a) 1000–500-hPa thickness (dam yr−1), (b) 850-hPa wind speed (m s−1 yr−1), (c) total column water vapor (kg m−2 yr−1), and (d) IVT magnitude (kg m−1 s−1 yr−1) averaged over 0600–0000 UTC on AR-dominant EP days. Mean values are contoured and linear regression statistical significance is indicated by thick (p ≤ 0.05) and thin (p ≤ 0.10) white stippling.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
2) TC-dominant
TC-dominant EP days are characterized by composite low pressure over the coastal mid-Atlantic and eastern North Carolina, with enhanced IVT over the low and extending to the northeast along the coast (Fig. 11a). Once again, high pressure presides to the east, but this time the IVT is associated more with the low (TC) and does not have the characteristic filamentary AR structure. The upper-level pattern is also less amplified; 1000–500-hPa thickness shows a much smaller ridge than in the AR-dominant case. These days, unlike AR-dominant days, show a large precipitation asymmetry, with very high rainfall totals in the southeast and low totals in the northwest of the analysis region (Fig. 11b). This is also the pattern evinced by cumulative precipitation trends from these days: significant and increasing over southern and eastern areas, but not significant or increasing over northern New York State and all of Vermont (Fig. 11c). The significant increasing trends, which do span much of the mid-Atlantic and New England, are noticeably larger (as a percentage) than those in AR-dominant EP day precipitation.
(a) 1800 UTC synoptic composite of TC-dominant EP days. Shown are 1000–500-hPa thickness (dam; dashed contours), MSLP (hPa; solid contours), IVT above 250 kg m−1 s−1 (gray shading), and regions with AR probability above 40% (blue overlay). (b) Mean daily precipitation (mm) on these days, and (c) trend (% yr−1) in season-total precipitation from these days. M-K trend significance is denoted by thick black (p ≤ 0.05) and thin gray (p ≤ 0.10) circles.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
Synoptic changes associated with TC-dominant EP days are, in many ways, inverse to those on AR-dominant days: decreased 1000–500-hPa heights indicative of cooling to the north of New England (Fig. 12a), slowed 850-hPa winds over the Northeast (Fig. 12b), broad decreases in atmospheric moisture content (Fig. 12c), and large decreases in IVT magnitude over and north of the composite TC center (Fig. 12d). Interestingly, the thickness trends suggest that the flow associated with fall TC-dominant days over the eastern United States has become more baroclinic. But how does precipitation associated with these events increase despite large decreases in moisture associated with them? The answer likely lies in Fig. 8, which shows the frequency and precipitation breakdown of all the subtypes. TC-dominant EP days have become more common in the second half of the period, but most of the changes come in the form of TC remnants–related events. Since a larger portion of TC-dominant EP days in later years are related to TC remnants, synoptic trends are likely to be toward the characteristics of these TC remnants events—that is, weaker and having less intense IVT than TCs. This explains the observed drying.
Trends in (a) 1000–500-hPa thickness (dam yr−1), (b) 850-hPa wind speed (m s−1 yr−1), (c) total column water vapor (kg m−2 yr−1), and (d) IVT magnitude (kg m−1 s−1 yr−1) averaged over 0600–0000 UTC on TC-dominant EP days. Mean values are contoured and linear regression statistical significance is indicated by thick (p ≤ 0.05) and thin (p ≤ 0.10) white stippling.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
3) Other
“Other” EP days, consisting primarily of the “Pure extreme IVT” and “Unspecified” categories, are depicted in Fig. 13. As expected, they exhibit much weaker synoptic signals than those seen in the AR- and TC-dominant types. Still, the pattern is favorable for warm air advection into New England, with composite high pressure to the northeast, a weak low over the mid-Atlantic, and 250–350 kg m−1 s−1 IVT oriented meridionally in between the pressure anomalies. These days, like AR-dominant EP days, see precipitation broadly across the entire region, although at smaller values in this case. Trends in cumulative precipitation from these days are not significant, except at a few stations in the far southwest. However, the linear regression slope values are large; this is likely due to the Pure extreme IVT days, which exhibit significant increases in frequency (Fig. 8e).
(a) 1800 UTC synoptic composite of “Other” EP days. Shown are 1000–500-hPa thickness (dam; dashed contours), MSLP (hPa; solid contours), IVT above 250 kg m−1 s−1 (gray shading), and regions with AR probability above 40% (blue overlay). (b) Mean daily precipitation (mm) on these days, and (c) trend (% yr−1) in season-total precipitation from these days. M-K trend significance is denoted by thick black (p ≤ 0.05) and thin gray (p ≤ 0.10) circles.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
Of the three broad weather types, the Other category is marked by the most pronounced warming signatures: widespread significant 1000–500-hPa thickness increases and atmospheric moisture content (Figs. 14a,c). However, decreases in low-level wind speed, particularly in the southeast, closely mirror the IVT magnitude trend (Figs. 14b,d). This results in a lack of IVT trends (or nonsignificant negative trends) over New England. But significant increasing IVT trends do prevail over the mid-Atlantic region, where (nonsignificant) precipitation trends are positive. Note the anomalous neutral thickness changes over Newfoundland; this reflects the fact that these days are more AR-like during the first half of the analysis period, with a coastal low, strong high pressure to the northeast, and a weak trough. In the second half of the period, there is little to suggest low pressure, and almost no indication of a trough (see Fig. S8).
Trends in (a) 1000–500-hPa thickness (dam yr−1), (b) 850-hPa wind speed (m s−1 yr−1), (c) total column water vapor (kg m−2 yr−1), and (d) IVT magnitude (kg m−1 s−1 yr−1) averaged over 0600–0000 UTC on “Other” EP days. Mean values are contoured and linear regression statistical significance is indicated by thick (p ≤ 0.05) and thin (p ≤ 0.10) white stippling.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
4) All EP days
Figure 15 shows the same four panels as the previous trend figures, except now for all 169 EP days; this should capture the aggregate changes that have taken place in the days with the most extreme precipitation across the mid-Atlantic and Northeast. Over the 1979–2019 period, there has been significant warming and moistening of the atmosphere in many locations as depicted by 1000–500-hPa thickness and TCWV (Figs. 15a,c). This is what would be expected of the mean atmospheric state in a warming climate: that it should become progressively warmer and, since warmer air has a higher water vapor capacity, moister. Yet this has not translated to higher moisture transport over the mid-Atlantic and Northeast, at least in the fall season (Fig. 15d). This is because there have been significant decreases in low-level wind speed throughout New England and parts of the mid-Atlantic. Slowed wind speeds might lead to lower moisture transport values, but they could also produce longer-duration events if they are indicative of a slower progression of weather patterns. Previous research has proposed that the smaller meridional temperature gradient resulting from Arctic amplification will tend to increase waviness and slow zonal wind speeds in a warming climate (Francis and Vavrus 2012, 2015). However, evidence of this claim is contested (Barnes 2013; Barnes and Screen 2015).
Trends in (a) 1000–500-hPa thickness (dam yr−1), (b) 850-hPa wind speed (m s−1 yr−1), (c) total column water vapor (kg m−2 yr−1), and (d) IVT magnitude (kg m−1 s−1 yr−1) averaged over 0600–0000 UTC on all EP days. Mean values are contoured and linear regression statistical significance is indicated by thick (p ≤ 0.05) and thin (p ≤ 0.10) white stippling.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
c. Prior TC influences
One factor that is not fully examined in this weather typing is the incorporation of TC-related moisture into extratropical cyclones, ARs, or other extreme IVT events. In the classification scheme of Fig. 3, the TC remnants–related categories depend on IBTrACS having the system listed as a non-tropical storm. Once IBTrACS has stopped tracking the system—likely because the circulation has completely vanished—precipitation on subsequent days will not be classified as related to TC remnants.
Therefore, we wish to note this shortcoming, and perform a short, subjective analysis of events with TC influences of this nature. For each EP event, maps of the preceding 3 days are examined manually for TCs and TC remnants. If the TC-related IVT visibly becomes incorporated into the event IVT, or if extreme precipitation falls within the broader TC circulation, then the event is considered to be TC influenced. This applies to all days within the EP event.
Using this metric, 14 of the 65 pure AR EP days, 7 of the 34 pure extreme IVT EP days, and 3 of the unspecified EP days have preceding TC influences. These 24 days comprise 18 events, 14 of which have occurred since 2002. In light of these changes, all of the top 10 EP events by total precipitation have TC influences (Table 2). These modifications to the previous analysis show that widespread extreme precipitation—whether cumulative or daily—is very likely to be associated with a TC influence.
Top 10 extreme precipitation events (defined as consecutive EP days), sorted by mean total precipitation. Columns give the mean total precipitation, duration, starting dates, and included weather types for each event. As in Table 1, except now including the subjective “prior TC influence” weather type.
In fact, adding the “prior TC influences” weather type results in the near-elimination of the previously observed AR-dominant trends. Figure 16 recreates the trend analysis figures, but for EP days with any TC influence or no TC influence whatsoever. Averaged over all included GHCN stations, this “any TC influence” category—containing all EP days with any dominant, remote, or prior TC influences—has statistically significant increasing precipitation trends, with a notably higher activity level starting in ∼2002 (Fig. 16a). Meanwhile, the days with no TC influences exhibit no trend (Fig. 16b). The same is true for station trends (Figs. 16c,d): days with TC influences essentially reproduce the total EP day trend of Fig. 5, while days with no TC influences have smaller trends of varying signs and little statistical significance.
Trends (% yr−1) in fall season total precipitation on EP days with (a),(c) any TC influence and (b),(d) no TC influence. M-K test statistical significance is depicted by thick black (p ≤ 0.05) and thin gray (p ≤ 0.10) circles. Slopes are calculated using linear regression. Trends are shown only for stations with at least 5 years of nonzero totals.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
This preliminary analysis positions TCs as the dominant cause of increases in large-scale extreme precipitation in the mid-Atlantic and Northeast from 1979 to 2019. When looking only at days characterized by nearby TCs, the trends are not as statistically significant—potentially due to smaller sample size and low probability of direct TC hits in this part of the country. However, when all the TC-influenced categories are combined, they account for the overwhelming majority of the trend in cumulative precipitation from EP days, as well as nearly all of the most extreme events.
d. Climatology and remote TC influences
The changes presented thus far apply only to days with high extreme precipitation totals. Moreover, once a day is classified, all precipitation on that day is attributed to that weather type—a reasonable approximation for very extreme, large-scale events, but not for precipitation as a whole.
These concerns are addressed in Fig. 17, which depicts changes in precipitation where the weather typing scheme (appendix C) is applied individually to each station, over all days in the fall seasonal climatology. In this case, AR-dominant precipitation occurs whenever an AR passes within 3–4 ERA5 grid points of a GHCN station, and TC-dominant precipitation occurs when a TC center passes within 500 km. The new category (“Remote TC influences”) consists of all observations attributed to extreme IVT that is connected to a TC (or TC remnants) more than 500 km away—that is, precipitation associated with an extreme IVT linkage to a remote TC.
(a),(c),(e) Mean time series of station total (black) and extreme (top 5%; red) precipitation within the (a) AR-dominant, (c) TC-dominant, and (e) Remote TC influence categories. (b),(d),(f) Station trends (% yr−1) in season total precipitation from the corresponding categories. Here, “remote TC influence” is defined as any instance where extreme IVT connected to a distant TC passes near the station (see appendix C for more details). M-K test statistical significance is depicted by thick black (p ≤ 0.05) and thin gray (p ≤ 0.10) circles.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
When averaged over all stations, none of these three types have significant trends in total or extreme precipitation, although there is large interannual variability (Figs. 17a,c,e). However, when plotted on a station-by-station basis, the total precipitation trends and (nonsignificant) changes do exhibit the same patterns seen in the TC-dominant and AR-dominant EP day precipitation changes (e.g., Figs. 9, 11, and 13). AR-dominant precipitation has the same north–south asymmetry, but much less pronounced, with only slight increases over the northern part of the domain. TC-dominant precipitation, on the other hand, shows southern and coastal increases of nearly the same magnitude as seen in EP days, with larger increases west of the Appalachians and along the eastern seaboard. Despite the consistency of the spatial patterns, these two weather types have changed in different ways. AR precipitation extremes have increased more rapidly than totals, while TC precipitation as a whole has increased quickly.
Precipitation associated with remote TC influences, though smaller in sum, exhibits positive changes across much of the mid-Atlantic and Northeast (Fig. 17f). This “remote TC influences” category captures a range of events in which TCs (or TC remnants) coincide with moisture transport to areas distant from the TC center. These could be the PREs described in Galarneau et al. (2010); they could be TCs with unusually elongated moisture fields; or they could simply be TCs interacting and merging with pre-existing regions of intense extratropical moisture transport. Increased precipitation from such events indicates that TCs are interacting at a higher rate with midlatitude weather patterns to produce or enhance event precipitation outside of the normal TC sphere of influence. In this case, both the TC-dominant and the remote TC influence increases are likely due to the increased frequency of TCs beginning in the late 1990s; more TCs implies more precipitation directly related to TCs in southern and coastal regions, and more precipitation remotely influenced by TCs in northern and inland regions. Therefore, changes in TC activity—whether driven by interdecadal variability or climate change—are potentially impactful even for regions that do not experience the TC directly. This also raises the question of whether the temperature and moisture increases seen here and expected in the future will change the probability or nature of precipitation associated with remote TC influences.
5. Discussion and conclusions
In fall, a number of different weather types contribute to increasing large-scale extreme precipitation in the mid-Atlantic and Northeast United States. High totals are seen from TCs in the south, from TC remnants in the mid-Atlantic and southern coastal Northeast, and from ARs broadly across the analysis region. Each of the three broad categories—AR-dominant, TC-dominant, and Other—is associated with increasing trends in cumulative precipitation from EP days. For TC-dominant EP days, the trends are large and widespread; for AR-dominant EP days, they are moderate and restricted to northern areas, including most of New York and New England. For other EP days, the trends are large and positive across southern areas, but not statistically significant; however, the frequency of extreme IVT EP days has increased significantly. Within these broad weather types, subtypes involving combined TC and AR influences produce intense precipitation in the northern mid-Atlantic and southern New England. These regions may therefore be especially vulnerable to multifactorial weather events involving tropical moisture.
On the whole, fall EP days within the 1979–2019 time period have become both warmer and moister, based on 1000–500-hPa thickness as well as total column water vapor (TCWV). However, this does not translate to higher IVT, due to opposing wind speed tendencies. Furthermore, these trends are not consistent across all weather types, especially because of the confounding effect of EP day frequency changes. This is especially the case for TC-dominant EP days, which have become more common because of an increase in TC remnant-related events. Since TC remnants tend to be weaker than TCs, there is a significant decrease in moisture and IVT on these days, despite an increase in the overall frequency of TC-dominant EP days. The other two categories, meanwhile, are generally consistent with the overall EP day trend of increased thickness and atmospheric moisture content.
A preliminary analysis of days with prior TC influences reveals a number of pure AR, pure extreme IVT, and unspecified EP days that may have been influenced by TCs in ways not picked up by the classification scheme employed here. When these days are grouped together with all the other TC-influenced EP days, all of the EP events with the highest precipitation totals have TC influences, and TC influences account for nearly the entirety of observed cumulative EP day precipitation trends. Another type of TC-related precipitation that has increased in some areas is precipitation associated with remote TC influences, which occurs when near-station extreme IVT is linked to a distant TC or TC remnants. Remote TC influences have the potential to affect areas beyond the typical scope of tropical systems, and could pose an important hazard if northern TC activity and water vapor capacity are both increasing.
But how do these changes in extreme events relate to changes in the fall seasonal climatology, as well as changes that have been observed in previous studies? Figure 18 depicts trends in the season-mean MSLP, IVT, 850-hPa winds, 1000–500-hPa thickness, and total column water vapor. The dipole pattern of MSLP trends in Fig. 18a is indicative of pressure rises to the north and falls to the south of the mean NASH latitude. This corresponds well to the significant easterly 850-hPa wind trends (i.e., slowed westerlies) over the mid-Atlantic and New England (Fig. 18c). But the most widespread trends are related to atmospheric temperature and moisture content, with total column water vapor and 1000–500-hPa thickness increasing statistically significantly over the majority of the 120°–30°W, 12°–59°N domain (Figs. 18b,d). These changes are similar to those seen in the set of 169 EP days: a moistening atmosphere, likely the result of warming, coupled with slowed wind speeds. The wind trend is consistent with the results of Huang et al. (2018), who found that steering flows on days with extreme precipitation have weakened in the Northeast under the new (post-1990s) AMO phase, and that TC translation speed has also slowed. Moreover, in considering the relation between climatological trends and EP day changes, it is worth noting that climatological trends are likely to influence not only the synoptic characteristics, but also the frequency of EP days. That the atmosphere has become warmer and moister during this season likely contributes to the larger number of EP days in recent years (see Fig. 5), regardless of the specific weather types involved.
Synoptic changes in the fall mean climatology between 1979 and 2019. (a) MSLP (hPa; shading) and IVT (kg m−1 s−1; arrows) trends; IVT trends are shown only where at least one component trend is significant (p ≤ 0.05). (b)–(d) Means (contours) and trends (shading) of 1000–500-hPa thickness (dam), 850-hPa wind speed (m s−1), and total column water vapor, respectively. M-K test significance (p ≤ 0.05) is shown by white stipples.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
The higher frequency of TC-dominant EP days since ∼2000 can be explained by comparing IBTrACS best-track data between the 1979–99 and 2000–19 time periods. Figure 19 depicts TC tracks occurring on and before TC-dominant EP days. Despite being one year shorter, the more recent time period sees greatly elevated TC activity along the entire east coast (Figs. 19a,b). This change comes in two forms: 1) more numerous TCs and 2) TCs with more complex tracks, some involving loops and meanders near the coast. An increased occurrence of slow-moving TCs on these days is consistent with the slowed low-level winds reported here and slowed steering flows reported in the literature (Huang et al. 2018). Slowed steering flows have also been found in the western North Pacific (Chu et al. 2012; Liang et al. 2017) and globally (Kossin 2018). During the 2000–19 TC-dominant EP days, slowed, more onshore winds are also associated with the location and strength of oceanic high pressure, which is positioned north of its 1979–99 TC-dominant EP day location.
(a),(b) Western North Atlantic TC tracks in the 7 days leading up to and including the TC-dominant EP days, split up by time period. Tracks are color-coded by intensity and system classification. Note that only TCs associated with the EP day precipitation are included. (c),(d) Synoptic composites of the corresponding EP days, taken at 1800 UTC. Depicted are IVT (gray shading; kg m−1 s−1), mean sea level pressure (hPa; contours), and 850-hPa wind (m s−1; arrows).
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
The North Atlantic subtropical high is expected to strengthen and move west due to climate change (Li et al. 2012; Shaw and Voigt 2015; He et al. 2017). Because the east coast of the United States is located on the western flank of this high, its position and strength are important for meridional moisture transport and precipitation (e.g., Li et al. 2011). But other aspects of a warming climate might also affect the probability of extreme precipitation events. One such factor, Arctic amplification, is hypothesized to increase the frequency of “wavy” (highly amplified) jet stream configurations associated with enhanced meridional upper-level flow and slower zonal progression (Francis and Vavrus 2012, 2015). Historical and future trends in such patterns are uncertain (Barnes 2013; Barnes and Screen 2015). But if true, this trend, along with a potential increase in the frequency and intensity of extratropical transition (Michaelis and Lackmann 2019), raises an interesting question for future work: Will high-impact multifactorial events become more common or intense due to an increased occurrence of slow-moving, tropical moisture-linked weather patterns?
In this study, we have provided a detailed account of the fall days producing the most extreme precipitation in the mid-Atlantic and Northeast United States. Following previous studies that show large fall extreme precipitation trends, results here indicate increased frequency of EP days associated with several factors: TCs, ARs, and other extreme IVT. The TC increases are the largest and most widespread, and the AR and extreme IVT-related trends may be due to events involving preceding TC influences. Synoptic changes are also present in the set of EP days: the environment is warmer with higher moisture content, but low-level wind speeds are slowed or unchanged.
The most extreme long-duration events typically involve AR or extreme IVT features in addition to TCs. Within the broad TC-, AR-, and IVT-based categories, subcategories are formed based on the IVT linkages between different features. These subcategories display different characteristic precipitation distributions. For instance, direct TC hits produce the most rainfall in the southern mid-Atlantic, while TCs that are decaying and/or interacting with AR features produce more rainfall in northern mid-Atlantic and southern Northeast. Such distinctions inform the risk of extreme precipitation occurring in different regions with different types of synoptic setups.
Acknowledgments.
This research was supported by the National Science Foundation Partnership for International Research and Education Program between the United States and Taiwan, OISE-1545917, awarded to the University at Albany, SUNY.
Data availability statement.
TC track data used in this study are freely available at https://www.ncdc.noaa.gov/ibtracs/. ERA5 data are freely available at https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era5. Precipitation data are freely available at https://www.ncei.noaa.gov/products/land-based-station/global-historical-climatology-network-daily. Scripts used in the creation of this paper have been made available at https://github.com/lexihenny/northeast_mid_atlantic_us_precipitation. Datasets produced and maps of all EP days are available at https://www.dropbox.com/sh/yakwj9g70iijw0i/AAAqubByF2EsYwHD8lQODIoKa?dl=0.
APPENDIX A
TC and AR Distance Thresholds
The 500-km (TC) and ∼100-km (AR, extreme IVT) distance criteria are determined by plotting precipitation as a function of distance to each of these weather events. Figure A1a shows station precipitation as a function of distance to the nearest tropical or non-tropical IBTrACS system. For both tropical and non-tropical systems, mean precipitation begins increasing more rapidly within ∼500 km; outside of this radius, the graph is relatively flat. The same is true for the fraction of >50 mm day−1 “heavy” precipitation, shown in Fig. A1b.
Testing of rainfall sensitivity to TCs and other IBTrACS systems. (a) Daily precipitation at each station vs the minimum distance to a TC (red) or non-tropical (black) IBTrACS system center on that day. Mean precipitation values for these categories in each 10-km bin are plotted as lines. (b) The fraction of precipitation observations above 50 mm day−1 for each 10-km bin.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
Figure A2 repeats this analysis for ARs, with 0.5° instead of 0.25° AR resolution to save resources. Because IVT correlates well with precipitation over complex terrain (Junker et al. 2008), mountainous regions in California and Taiwan experience reliably intense precipitation with the passage of an AR. However, in the mid-Atlantic and Northeast, the distance–precipitation relationship is less clear. There is a broadly linear increase in mean precipitation, measurable precipitation fraction, and heavy (>50 mm day−1) precipitation fraction extending out to and possibly beyond 500 km. Intuitively, it makes sense to have a smaller distance threshold; precipitation should not be attributed to an AR if the closest AR that day was 500 km away. We choose ∼100 km, since mean and especially heavy precipitation are maximized within this radius. We approximate this as a box extending 4 latitude grid points and 5 longitude grid points away from the ERA5 grid point nearest the precipitation observation.
Testing of rainfall sensitivity to detected ARs. (a) Daily precipitation at each station vs the minimum distance to an AR on that day. Mean precipitation values for these categories in each 10-km bin are plotted as lines. (b) The fraction of observations having measurable precipitation for each 10-km bin. (c) The fraction of observations above 50 mm day−1 for each bin.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
APPENDIX B
Extreme Precipitation and IVT Threshold Sensitivity Testing
To ensure that trends are not wholly dependent on the choice of threshold, cumulative EP day precipitation trends were computed for a range of station extreme thresholds (95% and 99%) and regional extreme sum percentiles (80%, 90%, and 95%) in Table B1. All trends were positive, with five of six falling between 1.3% and 2% yr−1. Although only the 99/80 combination used in this paper was statistically significant, three of the other combinations have linear regression p values between 0.12 and 0.14 as well as similar trend magnitudes.
Precipitation threshold sensitivity testing. The first column shows the percentile of measurable precipitation used as an extreme threshold at each station. The second column shows the percentile of summed extreme precipitation used to define extreme precipitation days (EP days). The remaining columns detail the number of days in the resulting sample, the percent trend in seasonal-total precipitation on EP days, and statistical significance results from linear regression and Mann–Kendall (M-K) significance testing.
Likewise, we tested spatially varying 90%, 92.5%, 95%, 97.5%, and 99% IVT thresholds, as well as a constant 95% threshold. The number of AR-dominant (see section 2d) EP days decreases as the threshold percentile increases, with the constant 95% having more days than the variable 95% (Fig. B1a). But trends in the frequency and precipitation of AR-dominant EP days remain nearly constant, the only exception being the small sample size 99% threshold (Fig. B1b).
IVT threshold sensitivity testing. (a) The percentage of (bars) EP days and (circles) EP day precipitation that are associated with AR-dominant weather types. (b) The linear regression trend in (bars) the annual frequency of AR-dominant EP days and (circles) the annual rate of AR-dominant EP day precipitation.
Citation: Journal of Climate 35, 20; 10.1175/JCLI-D-21-0953.1
APPENDIX C
Weather Type Definitions
For the automated portion of the study, the weather types are as follows:
-
Pure TC: TC passes within 500 km of an extreme precipitation observations; no AR involved
-
Pure AR: AR passes within ∼100-km vicinity of extreme precipitation observation; no TC involved
-
TC remnants: Same as pure TC, but with a non-tropical IBTrACS system
-
TC–AR combination: AR passes within ∼100 km and TC passes within 500 km of extreme precipitation observations; not necessarily the same observation
-
TC remnant–AR combination: Same as the TC–AR combination, but with a non-tropical IBTrACS system
-
TC-linked AR: AR passes within ∼100 km of an extreme precipitation observation and is connected to extreme IVT within 500 km of a TC center (but not within 500 km of extreme precipitation)
-
TC-linked extreme IVT: Same as TC-linked AR, but with a non-AR connected region of extreme IVT
-
TC remnant-linked AR: Same as TC-linked AR, but with a non-tropical IBTrACS system
-
TC remnant-linked extreme IVT: Same as TC-linked IVT, but with a non-tropical IBTrACS system
-
Pure extreme IVT: Extreme IVT passes within ∼100 km of extreme precipitation observation; no TC or AR involved
-
Unspecified: does not meet any of the above criteria
Figure 3 depicts these conditions as a decision tree. These categories are not meant to all be well populated or to serve as objective synoptic categories; characterizing the remote influence of TCs or TC remnants would require involved techniques such as PV surgery or trajectory analysis. Rather, the goal is to provide classifications that can be recombined as appropriate to obtain larger, more statistically significant groupings while retaining some information about the nature of the events.
REFERENCES
Agel, L., M. Barlow, J. Qian, F. Colby, E. Douglas, and T. Eichler, 2015: Climatology of daily precipitation and extreme precipitation events in the northeast United States. J. Hydrometeor., 16, 2537–2557, https://doi.org/10.1175/JHM-D-14-0147.1.
Allen, M., and W. Ingram, 2002: Constraints on future changes in climate and the hydrologic cycle. Nature, 419, 228–232, https://doi.org/10.1038/nature01092.
Barnes, E. A., 2013: Revisiting the evidence linking Arctic amplification to extreme weather in midlatitudes. Geophys. Res. Lett., 40, 4734–4739, https://doi.org/10.1002/grl.50880.
Barnes, E. A., and J. A. Screen, 2015: The impact of Arctic warming on the midlatitude jet-stream: Can it? Has it? Will it? Wiley Interdiscip. Rev.: Climate Change, 6, 277–286, https://doi.org/10.1002/wcc.337.
Bengtsson, L., K. I. Hodges, and N. Keenlyside, 2009: Will extratropical storms intensify in a warmer climate? J. Climate, 22, 2276–2301, https://doi.org/10.1175/2008JCLI2678.1.
Catto, J. L., and Coauthors, 2019: The future of midlatitude cyclones. Curr. Climate Change Rep., 5, 407–420, https://doi.org/10.1007/s40641-019-00149-4.
Chu, P.-S., J.-H. Kim, and Y. R. Chen, 2012: Have steering flows in the western North Pacific and the South China Sea changed over the last 50 years? Geophys. Res. Lett., 39, L10704, https://doi.org/10.1029/2012GL051709.
Dettinger, M., 2011: Climate change, atmospheric rivers, and floods in California—A multimodel analysis of storm frequency and magnitude changes. J. Amer. Water Resour. Assoc., 47, 514–523, https://doi.org/10.1111/j.1752-1688.2011.00546.x.
Donat, M. G., and Coauthors, 2013: Updated analyses of temperature and precipitation extreme indices since the beginning of the twentieth century: The HadEX2 dataset. J. Geophys. Res. Atmos., 118, 2098–2118, https://doi.org/10.1002/jgrd.50150.
Elsner, J. B., and Coauthors, 2008: The increasing intensity of the strongest tropical cyclones. Nature, 455, 92–95, https://doi.org/10.1038/nature07234.
Espinoza, V., D. E. Waliser, B. Guan, D. A. Lavers, and F. M. Ralph, 2018: Global analysis of climate change projection effects on atmospheric rivers. Geophys. Res. Lett., 45, 4299–4308, https://doi.org/10.1029/2017GL076968.
Francis, J. A., and S. J. Vavrus, 2012: Evidence linking Arctic amplification to extreme weather in mid-latitudes. Geophys. Res. Lett., 39, L06801, https://doi.org/10.1029/2012GL051000.
Francis, J. A., and S. J. Vavrus, 2015: Evidence for a wavier jet stream in response to rapid Arctic warming. Environ. Res. Lett., 10, 014005, https://doi.org/10.1088/1748-9326/10/1/014005.
Galarneau, T. J., Jr., L. F. Bosart, and R. S. Schumacher, 2010: Predecessor rain events ahead of tropical cyclones. Mon. Wea. Rev., 138, 3272–3297, https://doi.org/10.1175/2010MWR3243.1.
Hatsuzuka, D., T. Sato, and Y. Higuchi, 2021: Sharp rises in large-scale, long-duration precipitation extremes with higher temperatures over Japan. npj Climate Atmos. Sci., 4, 29, https://doi.org/10.1038/s41612-021-00184-9.
He, C., B. Wu, L. Zou, and T. Zhou, 2017: Responses of the summertime subtropical anticyclone to global warming. J. Climate, 30, 6465–6479, https://doi.org/10.1175/JCLI-D-16-0529.1.
Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 1999–2049, https://doi.org/10.1002/qj.3803.
Howarth, M. E., C. D. Thorncroft, and L. F. Bosart, 2019: Changes in extreme precipitation in the northeast United States: 1979–2014. J. Hydrometeor., 20, 673–689, https://doi.org/10.1175/JHM-D-18-0155.1.
Huang, H., J. M. Winter, E. C. Osterberg, R. M. Horton, and B. Beckage, 2017: Total and extreme precipitation changes over the northeastern United States. J. Hydrometeor., 18, 1783–1798, https://doi.org/10.1175/JHM-D-16-0195.1.
Huang, H., J. M. Winter, and E. C. Osterberg, 2018: Mechanisms of abrupt extreme precipitation change over the northeastern United States. J. Geophys. Res. Atmos., 123, 7179–7192, https://doi.org/10.1029/2017JD028136.
IPCC, 2014: Climate Change 2014: Synthesis Report. R. K. Pachauri and L. A. Meyer, Eds., IPCC, 151 pp.
Junker, N. W., R. H. Grumm, R. Hart, L. F. Bosart, K. M. Bell, and F. J. Pereira, 2008: Use of normalized anomaly fields to anticipate extreme rainfall in the mountains of northern California. Wea. Forecasting, 23, 336–356, https://doi.org/10.1175/2007WAF2007013.1.
Knutson, T., and Coauthors, 2010: Tropical cyclones and climate change. Nat. Geosci., 3, 157–163, https://doi.org/10.1038/ngeo779.
Kossin, J. P., 2018: A global slowdown of tropical-cyclone translation speed. Nature, 558, 104–107, https://doi.org/10.1038/s41586-018-0158-3.
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.
Kunkel, K. E., T. R. Karl, M. F. Squires, X. Yin, S. T. Stegall, and D. R. Easterling, 2020: Precipitation extremes: Trends and relationships with average precipitation and precipitable water in the contiguous United States. J. Appl. Meteor. Climatol., 59, 125–142, https://doi.org/10.1175/JAMC-D-19-0185.1.
Li, W., L. Li, R. Fu, Y. Deng, and H. Wang, 2011: Changes to the North Atlantic subtropical high and its role in the intensification of summer rainfall variability in the southeastern United States. J. Climate, 24, 1499–1506, https://doi.org/10.1175/2010JCLI3829.1.
Li, W., and Coauthors, 2012: Intensification of Northern Hemisphere subtropical highs in a warming climate. Nat. Geosci., 5, 830–834, https://doi.org/10.1038/ngeo1590.
Liang, A., L. Oey, S. Huang, and S. Chou, 2017: Long-term trends of typhoon-induced rainfall over Taiwan: In situ evidence of poleward shift of typhoons in western North Pacific in recent decades. J. Geophys. Res. Atmos., 122, 2750–2765, https://doi.org/10.1002/2017JD026446.
Liu, M., and Coauthors, 2019: Causes of large projected increases in hurricane precipitation rates with global warming. npj Climate Atmos. Sci., 2, 38, https://doi.org/10.1038/s41612-019-0095-3.
Menne, M. J., I. Durre, R. S. Vose, B. E. Gleason, and T. G. Houston, 2012: An overview of the Global Historical Climatology Network–Daily database. J. Atmos. Oceanic Technol., 29, 897–910, https://doi.org/10.1175/JTECH-D-11-00103.1.
Michaelis, A. C., and G. M. Lackmann, 2019: Climatological changes in the extratropical transition of tropical cyclones in high-resolution global simulations. J. Climate, 32, 8733–8753, https://doi.org/10.1175/JCLI-D-19-0259.1.
Neiman, P. J., F. M. Ralph, A. B. White, D. E. Kingsmill, and P. O. G. Perrson, 2002: The statistical relationship between upslope flow and rainfall in California’s coastal mountains: Observations during CALJET. Mon. Wea. Rev., 130, 1468–1492, https://doi.org/10.1175/1520-0493(2002)130<1468:TSRBUF>2.0.CO;2.
Neiman, P. J., F. M. Ralph, G. A. Wick, J. D. Lundquist, and M. D. Dettinger, 2008: Meteorological characteristics and overland precipitation impacts of atmospheric rivers affecting the West Coast of North America based on eight years of SSM/I satellite observations. J. Hydrometeor., 9, 22–47, https://doi.org/10.1175/2007JHM855.1.
O’Gorman, P. A., 2015: Precipitation extremes under climate change. Curr. Climate Change Rep., 1, 49–59, https://doi.org/10.1007/s40641-015-0009-3.
Pall, P., M. R. Allen, and D. A. Stone, 2007: Testing the Clausius–Clapeyron constraint on changes in extreme precipitation under CO2 warming. Climate Dyn., 28, 351–363, https://doi.org/10.1007/s00382-006-0180-2.
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<1721:SACAOO>2.0.CO;2.
Ralph, F. M., T. Coleman, P. J. Neiman, R. J. Zamora, and M. D. Dettinger, 2013: Observed impacts of duration and seasonality of atmospheric-river landfalls on soil moisture and runoff in coastal Northern California. J. Hydrometeor., 14, 443–459, https://doi.org/10.1175/JHM-D-12-076.1.
Shaw, T. A., and A. Voigt, 2015: Tug of war on the summertime circulation between radiative forcing and sea surface warming. Nat. Geosci., 8, 560–566, https://doi.org/10.1038/ngeo2449.
Trenberth, K. E., 1999: Conceptual framework for changes of extremes of the hydrological cycle with climate change. Climatic Change, 42, 327–339, https://doi.org/10.1023/A:1005488920935.
Webster, P. J., and Coauthors, 2005: Changes in tropical cyclone number, duration, and intensity in a warming environment. Science, 309, 1844–1846, https://doi.org/10.1126/science.1116448.
