You are on page 1of 22

A SCALE TO CHARACTERIZE

THE STRENGTH AND IMPACTS


OF ATMOSPHERIC RIVERS
F. Martin R alph, Jonathan J. Rutz, Jason M. Cordeira, Michael Dettinger, Michael Anderson,
David Reynolds, L awrence J. Schick, and Chris Smallcomb

A scale for atmospheric river intensity and potential impacts is introduced, enhancing
situational awareness and forecast communication.

A
tmospheric rivers (ARs) have emerged as a et al. 2016). Recently, the linkages between ARs and
subject of broad interest not only in the other key phenomena such as the warm conveyor belt
scientific community, but also with water man- and tropical moisture exports have been elucidated
agers, emergency managers, media, the public, and (e.g., Dettinger et al. 2015). The recently released
policy makers. The role of ARs in creating extreme Fourth National Climate Assessment now includes
precipitation, flooding, drought, and other impacts ARs as a type of extreme storm (along with tropical
is well established (Table 1). Major field experiments, storms, severe convection, and winter storms) and
such as the 6-yr-long interagency CalWater program highlights increasing AR occurrence and inten-
of field studies, have been conducted on ARs (Ralph sity as a climate change risk (Wuebbles et al. 2017).

AFFILIATIONS: R alph —Center for Western Weather and Reynolds —Cooperative Institute for Research in the Environmen-
Water Extremes, Scripps Institution of Oceanography, Univer- tal Sciences, University of Colorado Boulder, Boulder, Colorado;
sity of California, San Diego, La Jolla, California; Rutz—Western Schick*—Seattle District, U.S. Army Corps of Engineers, Seattle,
Region Headquarters, NOAA/National Weather Service, Salt Lake Washington; Smallcomb —NOAA/National Weather Service, Reno,
City, Utah; Cordeira—Department of Atmospheric Science and Nevada
Chemistry, Plymouth State University, Plymouth, New Hampshire, * Retired.
and Center for Western Weather and Water Extremes, Scripps CORRESPONDING AUTHOR: F. Martin Ralph, mralph@ucsd.edu
Institution of Oceanography, University of California, San Diego, La
The abstract for this article can be found in this issue, following the table
Jolla, California; Dettinger—Water Cycle Branch, U.S. Geological
of contents.
Survey, Carson City, Nevada, and Center for Western Weather and
DOI:10.1175/BAMS-D-18-0023.1
Water Extremes, Scripps Institution of Oceanography, University
of California, San Diego, La Jolla, California, and Global Water In final form 14 September 2018
Center, University of Nevada, Reno, Nevada; Anderson —Cali- ©2019 American Meteorological Society
For information regarding reuse of this content and general copyright
fornia Department of Water Resources, Sacramento, California; information, consult the AMS Copyright Policy.

AMERICAN METEOROLOGICAL SOCIETY FEBRUARY 2019 | 269


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
Table 1. Selected scientific findings and impacts associated with ARs.
ARs on the West Coast… Quantitative finding References
Cause the heaviest rains 92% of West Coast’s heaviest 3-day rain events fed by ARs Ralph and Dettinger (2012)
Bring warmest storms Average >50% more precipitation and 2.5°C warmer Dettinger (2004); Backes et al.
(less snow, more rain) than other storms in Sierra Nevada (2015); Guan et al. (2016)
Ralph et al. (2006); Neiman
40%–90% of major floods in West Coast rivers have
Cause West Coast floods et al. (2011); Konrad and
been fed by ARs
Dettinger (2017)
Cause storm surges in coastal 15%–50% of annual sea level maxima are associated with
Khouakhi and Villarini (2016)
areas AR-related cyclones
20%–50% of extreme coastal-wind episodes associated
Yield extreme coastal winds Waliser and Guan (2017)
with ARs
Breach levees 81% of Central Valley levee breaks happened during ARs Florsheim and Dettinger (2015)
Cause landslides, debris flows, ARs cause 68% of postfire debris flows in Southern Oakley et al. (2017); Young et al.
and avalanches California (2017); Hatchett et al. (2017)
Bring cycles of wet and dry Account for 85% of multiyear precipitation variance in
Dettinger and Cayan (2014)
years California
Fill reservoirs and provide 30%–50% of California rain, snow, and streamflow from Guan et al. (2010); Dettinger
water supplies ARs et al. (2011)
End West Coast droughts 40%–75% of droughts on West Coast ended by an AR Dettinger (2013)
Sustain wetlands, floodplains, 77% of ecologically significant inundations of Yolo Bypass
Florsheim and Dettinger (2015)
and fisheries floodplain, Sacramento River, initiated by ARs
Statistically significant relations found between summer
Water deserts and forests far
normalized difference vegetation index (greenness) and Albano et al. (2017)
inland, modulate wildfire risks
areas burned in parts of interior Southwest
Freshen estuaries but some- Mar 2011 ARs freshened San Francisco Bay by 60%,
Cheng et al. (2016)
times threaten estuarine fauna resulting in wild oyster kill rate of 97%–100%
Modify banks and bottom sedi-
More invertebrate densities and diversity after major AR
ments, modulating aquatic fauna Herbst and Cooper (2010)
flooding; 10 times more in predisturbed settings
in mountain streams

Community interest in the subject led to the devel- weaker and stronger ARs—many of these requests
opment of a definition of “atmospheric river” for the came directly from operational staff with the Na-
Glossary of Meteorology (American Meteorological tional Weather Service, who must communicate
Society 2018; Ralph et al. 2018), and an International potential impacts to the public.
Atmospheric Rivers Conference (IARC) brought The term atmospheric river has entered the scien-
together over 100 people representing AR work on tific and public lexicon (American Meteorological
six continents (Ralph et al. 2017a). Society 2018; Ralph et al. 2018), but while the me-
The anomalously wet 2016/17 cool season over the teorological community often focuses on AR-related
western United States highlighted the need for greater hazards (a minority of storms), information regarding
distinction between the majority of ARs that are pri- AR-related benefits (most storms) is often lacking.
marily beneficial and the minority of ARs that are pri- Yet in California, which has the largest interannual
marily hazardous. ARs of varying intensity affected variability in precipitation of any state (Dettinger
Northern California during this time, contributing to et al. 2011), ARs contribute 25%–50% of annual
reservoir storage, snowpack, and drought relief, but precipitation (including critical snowpack) in just a
also cumulatively leading to the Oroville Dam crisis few days each year (Dettinger et al. 2011; Ralph et al.
during February 2017 (Fig. 1). One particularly strong 2013; Rutz et al. 2014). Furthermore, the top 5% wet-
AR in January 2017 produced extreme precipitation test days each year, most of which are attributable
over the Feather River basin (Fig. 2), in which the to ARs, are responsible for 85% of the interannual
Oroville Dam is located. The active weather pattern variability in precipitation over Northern California
over the western United States during this timeframe (Dettinger and Cayan 2014). Therefore, the presence
led to numerous requests for differentiation between or absence of a few AR events can “make or break”

270 | FEBRUARY 2019


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
Fig. 1. Photographs illustrating the wide range of water levels at Oroville Dam and associated challenges faced
in managing water in an environment of extremes, for which variability is driven by atmospheric rivers. Figures
are all public domain and accessed via the California Department of Water Resources (CADWR) with photo
credits: (a) Zack Cunningham/CADWR, 23 Oct 2015; (b) Paul Hames/CADWR, 23 Jun 2005; (c) Dale Kolke/
CADWR, 15 Feb 2017; and (d) Dale Kolke/CADWR, 27 Feb 2017.

precipitation over the course of the water year.1 of similar importance in other key areas globally, for
Many water managers recognize the benefits of ARs example, Europe (Lavers et al. 2011; Ramos et al. 2015;
in terms of water supply (J. Jasperse 2018, Sonoma Eiras-Barca et al. 2018) and South America (Viale
County Water Agency, personal communication; and Nuñez 2011).
G. Woodside 2018, Orange County Water District, Currently, there is no concise method for convey-
personal communication). In recognition of the key ing the spectrum of benefits and hazards faced by
role ARs play in Californian hydroclimate, the state communities during a particular AR event. This
passed legislation in 2015 creating the “Atmospheric paper, written by hydrometeorological scientists,
River Research, Mitigation and Climate Forecast- weather forecasters, and users of weather infor-
ing”2 program, which aims to develop methods to mation, introduces a scale for characterizing the
better characterize and communicate information strength and impacts of ARs. This scale is intended
about ARs to policy makers, decision-makers, and to serve the western United States, and other regions
the public. While much has been learned about ARs with significant AR climatologies, in the same way
on the U.S. West Coast, they have been found to be that scales for hurricanes (Elsner and Kara 1999),
tornadoes (Fujita 1981), and nor’easters (Kocin and
1
The term water year is used extensively in the western United Uccellini 2004) have served other parts of the country.
States based on the annual cycle of precipitation and runoff. The paper first describes the characteristics of ARs
For example, water year 2016 started on 1 October 2015 and and the selection of vertically integrated water vapor
ended on 30 September 2016. transport (IVT) and AR event duration as key metrics
2
For more information see https://leginfo.legislature.ca.gov by which to gauge AR strength. The AR scale is then
/faces/billTextClient.xhtml?bill_id=201520160SB758. defined, followed by a discussion of the frequencies

AMERICAN METEOROLOGICAL SOCIETY FEBRUARY 2019 | 271


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
and impacts associated with ARs of varying strength, What variable to use to measure ARs? Historically, both
as defined by the scale, over the western United States. vertically integrated water vapor (IWV; also known as
The implications of this study are then summarized, precipitable water) and IVT (see appendix for calcula-
and a forecast example is presented. tion) have been used to define the spatial extent and
intensity of ARs. Initially, AR-related studies used sat-
IDENTIFYING THE CHARACTERISTICS ellite-based observations of IWV as a proxy to identify
OF ATMOSPHERIC RIVERS ON WHICH ARs, since observations of column wind speed are not
TO BASE A SCALE. What is an atmospheric river? available everywhere (Ralph et al. 2004; Neiman et al.
Development of a scale for ARs requires specification 2008). However, IVT is less dependent on surface eleva-
of what constitutes an AR. There now exists a formal tion, more directly related to precipitation outcomes
definition in the Glossary of Meteorology that can (e.g., Moore et al. 2012; Rutz et al. 2014; Oakley et al.
form the basis (American Meteorological Society 2017), and is used in most current studies as the basis for
2018). As defined in the glossary, “an atmospheric identifying ARs (e.g., Cordeira et al. 2017; Young et al.
river is a long, narrow and transient corridor of strong 2017; Waliser and Guan 2017; Dettinger et al. 2018).
horizontal water vapor transport that is typically as- Furthermore, numerical weather prediction models
sociated with a low-level jet stream ahead of the cold predict IVT more skillfully than precipitation itself,
front of an extratropical cyclone” (Zhu and Newell offering an advantage in forecasting (Lavers et al. 2016).
1998; Ralph et al. 2004, 2006, 2017b; Bao et al. 2006; Hence, IVT is also used here in defining the AR scale.
Stohl et al. 2008; Warner et al. 2012; Cordeira et al.
2013; Sodemann and Stohl 2013; Dacre et al. 2015). What is the range of AR intensities in terms of IVT? Since
Some ARs entrain water vapor from the tropics (e.g., IVT is chosen to define the AR scale, it is useful to
Stohl et al. 2008; Ralph et al. 2011; Cordeira et al. 2013; examine the wide range of IVT magnitudes associated
Sodemann and Stohl 2013), but this is not a trait of all with ARs. In many ARDT strategies, the term atmo-
ARs. Atmospheric rivers frequently lead to heavy pre- spheric river is, by definition, restricted to features with
cipitation where they are forced upward, for example, IVT ≥ 250 kg m–1 s–1 (and IWV ≥ 2.0 cm). However,
by mountains or by ascent in the warm conveyor belt. the IVT values in ARs cover a wide range, with re-
Horizontal water vapor transport in the midlatitudes analyses and observations of ARs from radiosondes and
occurs primarily in atmospheric rivers and is focused dropsondes providing many examples of landfalling
in the lower troposphere. (Henceforth, the term trans- ARs over the northeast Pacific with IVT magnitudes
port represents the vertically integrated horizontal flux >1,000 kg m–1 s–1 and IWV values >3.5 cm, as shown
of water vapor, and not the vertical fluxes of water in Figs. 2a and 2b. In this example, a landfalling AR in
vapor such as land–air, sea–air, and convection that early January 2017 contained radiosonde-derived IWV
are largely the focus of past meteorological studies.) and IVT magnitudes of 3.5 cm and 1,102 kg m–1 s–1,
This definition represents an AR as an “object” in respectively, at Bodega Bay (BBY), California, where AR
space and time that has the appropriate character- conditions (i.e., IVT ≥ 250 kg m–1 s–1) persisted for ~36 h.
istics. This has led to the development of many AR Furthermore, offshore dropsonde measurements from
detection techniques (ARDTs) that can identify the 21 ARs collected over several field campaigns show that
locations, times, and characteristics of ARs region- observed AR IVT intensities can exceed 1,250 kg m–1 s–1.
ally or globally (Shields et al. 2018). However, it is In addition to observational data, experience with the
also useful (especially in context of many practical, AR Landfall Probability Tool (Cordeira et al. 2017),
on-the-ground applications and communications) to which uses 3-hourly Global Forecast System (GFS) out-
consider an Eulerian perspective, that is, that of an put, shows that ensemble forecasts contain many ARs
observer at a specific location monitoring conditions with IVT ≥ 500, and on occasion IVT ≥ 1,000 kg m–1 s–1.
over time at that site. From this perspective, the ob- It has already been found useful to represent the “AR
server experiences the passage of an AR as a sequence intensity” as the maximum value of IVT at a given
of meteorological conditions associated with the AR. location during the AR event at that location. Based on
Many studies take this approach, particularly those these observations and analyses of ARs that have made
describing the climatological characteristics of ARs landfall on the U.S. West Coast, the following intensity
(e.g., Rutz et al. 2014; Guan and Waliser 2015). The AR thresholds are chosen for the AR scale:
scaling method introduced here takes this Eulerian
perspective. For this reason, the term atmospheric • weak is ≥250–500 kg m–1 s–1,
river event will henceforth refer to the period that AR • moderate is ≥500–750 kg m–1 s–1 (transitional from
conditions occur at a given location. mostly beneficial to hazardous),

272 | FEBRUARY 2019


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
Fig. 2. (a) Satellite-observed IWV and (b) GFS-analyzed IVT over the northeastern Pacific Ocean and western
United States at 1800 UTC 8 Jan 2017. (c) Rainfall category (R-CAT; Ralph and Dettinger 2012) precipitation
totals during 6–8 Jan 2017 and (d) USGS streamflow percentiles at 1500 UTC 9 Jan 2017.

AMERICAN METEOROLOGICAL SOCIETY FEBRUARY 2019 | 273


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
• strong is ≥750–1,000 kg m–1 s–1, that also concluded ARs were more common in the
• extreme is ≥1,000–1,250 kg m–1 s–1, and Pacific Northwest than in California.
• exceptional is ≥1,250 kg m–1 s–1.
The important role of duration of AR conditions at landfall.
These thresholds can be applied to observational Although the maximum intensity of IVT during a
analyses of past events, or to forecasts. landfalling AR largely controls the hourly rain rates,
To better communicate the magnitude of the it has also been shown that the storm-total precipita-
strongest ARs, based on their frequencies of occur- tion (and hence runoff) is strongly controlled by the
rence, Dettinger et al. (2018) used reanalysis data storm-total water vapor transport (Ralph et al. 2013).
from 1948 to 2015 to diagnose IVT return periods In this framework, a “storm” or an “AR event” is de-
for different locations along the U.S. West Coast. An fined by the period during which AR conditions are
example of this is shown in Fig. 3 for the landfall loca- continuously met at a given location. This analysis was
tion of the storm highlighted in Fig. 2. It shows that based on observations of AR conditions using an atmo-
the return periods for 750, 1,000, and 1,250 kg m–1 s–1 spheric river observatory (ARO) and found that 74% of
are roughly 1, 3, and 20 years , respectively, for those the variance in storm-total precipitation is explained
locations (these IVT thresholds are 3-h averages). A by variance in storm-total upslope IVT (where the
more comprehensive analysis of the entire U.S. West upslope direction is based on regional terrain orienta-
Coast is provided in Dettinger et al. (2018), which tion). This study found that an average AR lasted 20 h
reveals that the Oregon coast has the strongest ARs at this location, a result confirmed by analysis from
on average, with IVT reaching 1,000 kg m–1 s–1 about Rutz et al. (2014) using reanalysis methods. Rutz et al.
once per year, while the Washington coast sees this (2014) extended this result from one coastal location
once every 2 years, San Francisco once every 3 years, to the entire coast and showed that AR duration varies
and Los Angeles about once every 10 years. These from about 18 h on average to about 25 h depending
results refine earlier studies of AR landfall frequency on latitude. Lamjiri et al. (2017) concluded also that in
by Neiman et al. (2008), Rutz et al. (2014), and others the western United States, storm duration was a more

Fig . 3. Return period of selected values of IVT within landfalling ARs at 25°–40°N on the coast, based on 3-h
MERRA reanalysis data from 1980 to 2016.

274 | FEBRUARY 2019


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
important parameter in deter- Table 2. (top) An AR intensity scale based on maximum instantaneous
mining storm-total precipita- IVT magnitude and duration of AR conditions (i.e., IVT > 250 kg m –1 s –1),
tion than was the magnitude of and (bottom) a subjective assessment of the potential for beneficial or
peak hourly precipitation rate hazardous impacts.
within the storm. Max IVT Duration of AR conditions (h)
In summary, AR duration (kg m –1 s –1) ≤24 ≥24–48 ≥48
and (IVT) intensity are the key ≤250 Not an AR Not an AR Not an AR
characteristics in determining
≥250–500 Weak AR AR Cat 1 AR Cat 2
streamflow magnitudes and
≥500–750 AR Cat 1 AR Cat 2 AR Cat 3
related hydrologic impacts.
Since AR durations greater ≥750–1,000 AR Cat 2 AR Cat 3 AR Cat 4
than ~24 h are, subjectively, ≥1,000–1,250 AR Cat 3 AR Cat 4 AR Cat 5
often recognized as being as- ≥1,250 AR Cat 4 AR Cat 5 AR Cat 5
sociated with greater impacts, AR category scale Assessment of beneficial vs hazardous impacts
this study classifies AR dura- AR Cat 1 Primarily beneficial
tions as ≤24, ≥24–48, and AR Cat 2 Mostly beneficial, but also hazardous
≥48 h. The ~24-h threshold is
AR Cat 3 Balance of beneficial and hazardous
simply a convenient, though
AR Cat 4 Mostly hazardous, but also beneficial
arbitrary, marker of event-
based impacts as a function AR Cat 5 Primarily hazardous
of duration.
Some dramatic recent examples illustrate ways that Based on this behavior, the AR scale will use 24–
stronger and more persistent IVTs along the U.S. West 48-h duration as the norm. To accommodate the fact
Coast have been associated with more hazardous im- that some ARs will have shorter duration, and some
pacts as compared to weaker and less persistent ARs. longer, as well as the fact that impacts on streamflow
For example, the 8 January 2017 AR depicted in Figs. 2a are so sensitive to duration (e.g., Ralph et al. 2013), the
and 2b produced 72-h rainfall totals on 6–8 January AR will be downgraded by one level if its duration is
2017 that exceeded 200 mm at 22 Cooperative Observer <24 h, and upgraded if ≥48 h.
Program (COOP) observing sites across Northern and
central California, and totals that exceeded 300 mm A SCALE TO CATEGORIZE AR STRENGTHS
at five COOP observing sites (Fig. 2c). These rainfall AND IMPACTS. The previous section established
totals resulted in record streamflow for the date on the roles of IVT and the duration of AR conditions
several rivers and streams across the California Coastal in producing high-impact hydrometeorological
Ranges and northern Sierra Nevada (Fig. 2d). events. This section lays out a strategy for a scaling
of ARs that considers both these
factors, based on time series of
observed or predicted conditions
at individual points. Because
this approach does not include
shape requirements typical of
earlier studies’ object-oriented
AR identification methods, and
it uses only time series of IVT in
an Eulerian framework, it greatly
simplifies implementation of the
scale in gridded datasets like re-
analyses and forecasts.
After much consideration, it
Fig. 4. A scale that categorizes AR was decided to use a 0.5° × 0.5°
events based on the maximum
spatial grid to assess and display
instantaneous IVT associated with
a period of AR conditions (i.e., AR conditions in terms of the
IVT ≥ 250 kg m –1 s –1) and the dura- scale. The resolution is ultimately
tion of those conditions at a point. somewhat arbitrary. The value

AMERICAN METEOROLOGICAL SOCIETY FEBRUARY 2019 | 275


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
Table 3. Examples of AR Cat 1–5 events making landfall along the California coast during water year 2017.
Max IVT Max IWV Duration of AR AR scale
Date and time of max IVT (kg m –1 s –1) (mm) conditions (h) category AR strength
1200 UTC 2 Feb 2017 >400 >25 −24 AR Cat 1 Weak
1200 UTC 19 Nov 2016 >650 >30 −42 AR Cat 2 Moderate
1200 UTC 14 Oct 2016 >800 >35 −36 AR Cat 3 Strong
1800 UTC 8 Jan 2017 >1,100 >35 −36 AR Cat 4 Extreme
1200 UTC 7 Feb 2017 1,183 >30 −84 AR Cat 5 Exceptional

used here is based on the following considerations: identified as a long and narrow region of large IWV
i) the spatial areas of ARs are much larger than this; and IVT extending poleward and eastward toward
ii) gridded data on this resolution are increasingly the U.S. West Coast. Generally, the magnitude of
available from reanalyses, operational global weather IWV and IVT increases as a function of increasing
prediction models, and even climate models, at scales AR Cat, but because IWV magnitude is not directly
comparable to this; and iii) although it is technically considered as part of the AR scale, it does not increase
feasible to apply the AR categorization method to uniformly as a function of increasing AR Cat. For
finer grids, it is likely that the greatest value of the example, the IWV magnitude associated with the
AR scale will be for situational awareness. listed AR Cat 5 event is smaller than that associated
The AR scale categorizes AR events based on with the listed AR Cat 3 and 4 events.
the maximum instantaneous IVT “intensity”
and the duration of the event (i.e., the duration of FREQUENCY OF OCCURRENCE OF AR
IVT ≥ 250 kg m–1 s–1) at a given point (Table 2; Fig. 4). E VE NTS I N TH E WE STE R N U N ITE D
An AR event at a given location is categorized by lo- STATES BASED ON AR CATEGORY. This sec-
cating the row associated with the maximum IVT and tion presents the average annual number of AR Cat 1–5
the column associated with the event duration. For events from January 1980 to April 2017 over the west-
example, a maximum IVT ≥500 and <750 kg m–1 s–1 ern United States, based on IVT data calculated from
would be classified as being of “moderate” intensity, Modern-Era Retrospective Analysis for Research and
and a duration ≥24 and <48 h would rank this as an Applications (MERRA; Fig. 6; Rienecker et al. 2011).
AR category (Cat) 2 event. If the same event instead ARs of each category tend to be most frequent over
had a duration ≥48 h, it would be upgraded to an AR the northeastern Pacific Ocean as shown in earlier
Cat 3 event, and a duration <24 h would downgrade studies (e.g., Rutz et al. 2014; Guan and Waliser 2015).
it to an AR Cat 1 event. This system of classification AR Cat 5 and Cat 4 events over land are generally lim-
works up and down the AR scale with two excep- ited to the coastal regions north of Point Conception
tions. First, the maximum category on the scale is (~34°N) and west of the Cascades and Sierra Nevada.
AR Cat 5, even if event duration is ≥48 h. Second, The map of AR Cat 5 events (Fig. 6a) highlights a
“weak” ARs (i.e., those with maximum IVT ≥250 local maximum in occurrence of the strongest ARs
and <500 kg m–1 s–1, but with a duration <24 h) do near the Oregon coast, reflecting patterns in the IVT
not receive a categorical ranking on the AR scale, return period analysis of Dettinger et al. (2018). Since
as represented by the gray area in Fig. 4 (even weak IWV generally increases equatorward and low-level
events require a minimum duration of 12 h). winds associated with ARs increase poleward (as
Figure 5 and Table 3 provide examples of a range of shown in Ralph et al. 2017b), this likely represents the
AR Cat 1–5 events using both satellite-observed IWV most favorable geographic overlap. It should also be
[from Special Sensor Microwave Imager (SSM/I)] and noted that the scale identifies a few (once every few
IVT (from GFS) from the 2016/17 cool season. The years) AR Cat 4 and 5 events along the very southern
synoptic-scale pattern associated with these events is portion of the domain (i.e., the Baja Peninsula, South-
characterized by a midlatitude cyclone–anticyclone ern California, and southern Arizona). These events
pair over the northeast Pacific and an AR located occur during summer and fall and are likely related to
between these features. The AR in each case is readily tropical cyclones, the North American monsoon, and

◀ Fig . 5. Satellite-observed IWV and GFS-analyzed IVT over the northeastern Pacific and western United
States during examples of (a) AR Cat 1, (b) AR Cat 2, (c) AR Cat 3, (d) AR Cat 4, and (e) AR Cat 5 events from
the 2016/17 cool season. An example of an AR Cat 4 event is shown in Fig. 1.

AMERICAN METEOROLOGICAL SOCIETY FEBRUARY 2019 | 277


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
other non-AR features but
are identified by the scale,
since it has no geometric
requirements for meteoro-
logical features.
T he d i s t r ibut ion of
AR Cat 2 and 3 events
(Figs. 6c,d) more clearly
shows the influence of to-
pography on ARs over the
western United States. As
midlatitude cyclones sup-
portive of ARs move east-
ward across this region,
they interact with complex
and formidable mountain
ranges, which tend to result
in cyclolysis and AR decay
through rainout and other
processes (Rutz et al. 2015).
Therefore, achieving the
combination of IVT mag-
nitude and event duration
associated with AR Cat 2
and 3 events becomes in-
creasingly difficult as one
moves inland. The distri-
bution of AR Cat 1 and
weak AR events (Figs. 6e,f)
is characterized by a simi-
lar spatial pattern to that of
higher AR Cat events, but
their frequency is larger
due to less restrictive crite-
ria. Note that the frequency
of events over the northern
Great Plains is inflated by
warm-season events featur-
ing strong IVT.
In summary, the spa-
tial distributions of weak
AR events and AR Cat 1–5
events over the western
United States closely re-
sembles the pattern of AR
frequency shown by Rutz
et al. (2014), with the num-
ber of events becoming
Fig . 6. Average annual number of weak ARs and AR Cat 1–5 events from increasingly restricted to
Jan 1980 to Apr 2017. Analysis is based on MERRA. Values of at least one per lower-elevation corridors
year or greater are shown in color fill, and the frequency of one per 4 years and coastal regions as a
on average is shown (dashed line).
function of increasing AR
Cat (i.e., increasing IVT

278 | FEBRUARY 2019


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
strongest and longest lasting.
This section analyzes ARs
that arrived over the MERRA
grid cell closest to Bodega
Bay (38°N, 123.125°W) to il-
lustrate the seasonality of AR
Cat events at that location,
and the typical spatial extent
of AR conditions and associ-
ated precipitation.
The seasonality of all AR
Cat events near Bodega Bay
is characterized by a rapid
increase during October, a
maximum during November–
March (peaking in December),
a gradual decline from April to
June, and a minimum during
July–September (Fig. 7). The
strongest, AR Cats 4 and 5,
Fig. 7. Seasonality of AR Cat events near Bodega Bay based on events from events are primarily restricted
MERRA from Jan 1980 to Apr 2017. to the October–March period,
and are most frequent during
and event duration). AR Cat 4 and especially AR January and February. During the July–September
Cat 5 events are primarily limited to coastal locations, minimum, events stronger than AR Cat 2 are rare.
whereas weak AR events occur multiple times per The average characteristics of ARs in each AR Cat
year, on average, nearly everywhere over the western are calculated and summarized in Table 4. It is not
United States. surprising that, from AR Cat 1 to Cat 5, the number
of events decreases (from 268 to 10), the mean dura-
SPATIAL EXTENT AND PRECIPITATION tion increases (from 21 to 72 h), and the maximum
IMPACTS OF AR CAT 1–5 EVENTS MAK- 3-h mean IVT increases (from 480 to 1,118 kg m–1 s–1).
ING LANDFALL AT A NORTHERN CALI- It is more surprising, however, that these maximum
FORNIA COASTAL LOCATION. It is common 3-h mean IVT values for AR Cats 3–5 are slightly less
for strong AR events to have impacts across broad than the IVT thresholds used in determining these
areas of the West, although the heaviest precipitation categories. This indicates that at least some AR Cat
and greatest impacts are limited to where the AR is 3–5 events achieve their categorical rating by being

Table 4. Summary of average AR Cat characteristics for ARs that struck Bodega Bay from
Jan 1980 to Apr 2017.
Characteristic Weak Cat 1 Cat 2 Cat 3 Cat 4 Cat 5
No. of events 392 268 138 78 22 10
No. of AR Cat days 599 484 316 225 74 36
Avg duration (h) 13.3 21.0 32.1 47.6 58.2 71.7
Std dev duration (h) 4.7 9.5 10.7 18.4 20.6 29.4
Avg max IVT (kg m –1 s –1) 373.7 480.0 599.4 701.1 896.6 1,118.0
Std dev max IVT (kg m –1 s –1) 62.4 108.6 104.4 147.0 111.1 127.6
Avg storm-total IVT (107 kg m –1) 1.6 2.8 4.8 7.7 10.7 15.4
Std dev storm-total IVT (10 kg m )
7 –1
1.2 1.1 1.4 2.8 3.8 5.7
Avg 3-day IVT (107 kg m –1)a 4.9 5.7 6.7 7.9 9.5 11.2
Std dev 3-day IVT (107 kg m –1)a 0.4 1.5 1.8 2.4 3.0 3.7
a
3-day IVT is calculated as the 72-h total IVT beginning at 0000 UTC on any AR Cat day.

AMERICAN METEOROLOGICAL SOCIETY FEBRUARY 2019 | 279


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
Table 5. AR Cat 5 events at 38°N, 123.125°W, near Bodega Bay between Jan 1980 and Apr 2017. Flood in-
formation is from either the annual peak flow USGS data table or from streamflow time series.
AR Cat 5 Water Duration Flood stage (32 ft)
event No. year Year Start time/date End time/date (h) at Guerneville?
1 1980 1980 0600 UTC 11 Jan 1200 UTC 14 Jan 78 Yes (37 ft)
2 1984 1983 0600 UTC 9 Nov 0600 UTC 11 Nov 48 —
3 1986 1986 0000 UTC 14 Feb 0000 UTC 20 Feb 144 Yes (49 ft)
4 1991 1991 1200 UTC 2 Mar 0000 UTC 5 Mar 60 Yes (33 ft)
5 1996 1995 1500 UTC 10 Dec 0000 UTC 13 Dec 57 Yes (32 ft)
6 1997 1996 1800 UTC 16 Nov 0000 UTC 20 Nov 78 No
7 1997 1996/97 1800 UTC 28 Dec 0600 UTC 3 Jan 132 Yes (45 ft)
8 2011 2010 0900 UTC 23 Oct 1200 UTC 25 Oct 51 No
9 2015 2015 0900 UTC 5 Feb 1200 UTC 7 Feb 51 No
10 2017 2017 0600 UTC 7 Jan 0600 UTC 9 Jan 48 Yes (40 ft)
1 event Oct (1), Nov (2), Dec (2), Average:
Frequency
per 3.8 yr Jan (2), Feb (2), Mar (1) 75 h

Table 6. All dates with stream stage at Guerneville exceeding 12.2 m (40 ft) stage
(daily) between 1980 and Nov 2017. AR Cat levels are evaluated for each date, and for
several days preceding, at Bodega Bay using reanalysis. CFS: cubic feet per second.
CFS AR Cat AR Cat AR Cat AR Cat Max AR
Year Date CFS rank day 0 day −1 day −2 day −3 Cat
1980 14 Jan 55,300 23 5 — — — 5
1981 20 Dec 64,300 14 4 — — — 4
1983 27 Jan 68,500 10 4 — — — 4
1983 1 Mar 58,700 18 4 4 — — 4
1983 2 Mar 57,600 19 4 4 4 —
1983 3 Mar 56,800 20 — 4 4 4
1986 15 Feb 65,600 12 5 — — — 5
1986 16 Feb 59,800 17 5 5 — —
1986 17 Feb 82,300 5 5 5 5 —
1986 18 Feb 97,700 1 5 5 5 5
1986 19 Feb 80,100 6 5 5 5 5
1986 20 Feb 56,300 21 — 5 5 5
1995 9 Jan 86,200 3 4 4 4 4 4
1995 10 Jan 88,700 2 4 4 4 4
1995 11 Jan 62,000 15 0 4 4 4
1995 14 Jan 55,800 22 3 3 3 — 3
1995 10 Mar 65,100 13 1 4 4 — 4
1997 1 Jan 72,600 8 5 — — — 5
1997 2 Jan 75,000 7 5 — — —
2004 18 Feb 60,100 16 4 — — — 4
2005 31 Dec 69,000 9 3 3 3 3 3
2006 1 Jan 82,400 4 1 3 3 3
2006 2 Jan 65,700 11 — 1 3 3

280 | FEBRUARY 2019


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
“promoted” due to their du-

with the day the event started at Bodega Bay. (c) As in (b), but the maximum 3-day totals across all 10 AR Cat 5 events are shown. Panels (b) and (c) use
Fig. 8. (a) Average of maximum AR Cat conditions during all AR Cat 5 events that occurred at a grid cell near Bodega Bay, CA [10 events at 38°N, 123.125°W
(star)], from MERRA Jan 1980 to Apr 2017. (b) Average of 3-day precipitation accumulation on 10 AR Cat 5 events that hit Bodega Bay; 3-day intervals start
ration exceeding 48 h, and
that this happens more often
than being “demoted” due to
the duration being less than
24 h. The increases in mean
AR duration, maximum 3-h
mean IVT, and storm-total
IVT progress steadily from
AR Cat 1 to Cat 5, suggesting
the scaling is representing
systematic changes in the core
characteristics of AR intensity
and duration.
It is useful to examine the
relationship between AR Cat
events and major f looding
along the Russian River {i.e.,
12.2 m or 40 ft at the Guern­
eville stream gauge [U.S. Geo-
logical Survey (USGS) stream
gauge 11467000]}. Of the 10
AR Cat 5 events near this
location, 6 were associated
with major flooding, 3 struck
either early in the season or
during a major drought when
dry soils attenuated runoff,
and 1 occurred when stream-
flow data were not available
(Table 5). Another approach
is to compare all major flood

daily COOP precipitation observations (i.e., each dot shown).


events on the Russian River
to the presence of AR Cat
events, as shown in Table 6.
A number of these dates were
consecutive, and the events
can be viewed as 11 separate
floods, with some being of 2,
3, or even 5 days in duration.
Because of the typical 1–2-day
time lag between precipitation
and flooding on this river, the
AR Cat levels on the day of
the flood and on the 1–2 days
before are assessed. As with
the Ralph et al. (2006) study,
which first showed the con-
nection between landfalling
ARs and flooding on the Rus-
sian River, all 11 floods in this
analysis were associated with
landfalling ARs. In particular,

AMERICAN METEOROLOGICAL SOCIETY FEBRUARY 2019 | 281


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
282 |
FEBRUARY 2019
Fig. 9. (top) Average storm-total precipitation during all AR Cat 1–5 events near Bodega Bay based on daily COOP precipitation observations (i.e., each
dot shown). (bottom) As in top panels, but for temperature anomaly.

Unauthenticated | Downloaded 05/15/23 03:34 PM UTC


3 out of 10 (33% of) AR Cat 5 events, 6 out of 22 (30% each AR Cat at Bodega Bay, is shown in Fig. 9 (top).
of) AR Cat 4 events, and only 2 out of 78 (3% of) AR Cat Because many weaker ARs last much less than
3 were associated with major flooding at Guerneville. 3 days, this analysis uses a variable time window for
The geographic spread of AR Cat intensities during accumulations, based on the observed times of each
AR Cat 5 landfalls near Bodega Bay is shown in Fig. 8a AR at Bodega Bay. The precipitation associated with
(for each grid point, this is calculated by determining AR Cat 1 and Cat 2 events is modest enough (mostly
the maximum AR Cat achieved during an AR Cat 5 25–75 mm) to be largely beneficial (unless preceded
event at Bodega Bay, and then averaging over the 10 by a strong AR, or producing other hazards), whereas
such events). On average, AR Cat 4 and Cat 5 condi- AR Cat 5 events produce well over 200 mm at many
tions are localized near Bodega Bay and the immediate sites. To further this analysis, the frequency of oc-
West Coast. Meanwhile, at least AR Cat 1 conditions currence of different 3-day precipitation totals, at all
exist along the entire coast from Southern California to COOP sites in California, associated with different
Washington and extend inland, affecting much of the AR Cat events is shown in Fig. 10. It reveals that AR
western United States. A test of whether this method Cat 1–3 events group closely together in terms of
is contaminated by propagation of the AR along the precipitation, but that AR Cat 4 and 5 events have a
West Coast was made by including up to 2 days before much higher probability of producing extreme 3-day
and after the AR Cat 5 conditions at Bodega Bay, and precipitation totals. This distinction suggests a sig-
results were generally insensitive to this. nificant increase in hazardous impacts during AR Cat
The precipitation associated with the 10 AR Cat 4 and 5 events, and it could be the basis for classifying
5 events near Bodega Bay is examined by calculat- these as “major” ARs, an approach analogous to that
ing mean 3-day accumulations at COOP weather used in hurricane and tornado scales.
stations across the western United States. The 3-day The large precipitation totals associated with the
accumulation window begins with the onset of each higher AR Cat levels in Fig. 9 (top) are often more
AR Cat 5 event, with 3 days chosen for simplicity (as hazardous falling as rain, adding immediate runoff
in Ralph and Dettinger 2012), and because the aver- that floods rivers and lowlands, and more beneficial
age duration of AR Cat 5 events is ~3 days (Table 4). falling as snow, increasing water storage via mountain
The largest precipitation values occur over Northern snowpack. However, ARs are often warm storms, lead-
California, where many sites exceed 150 mm, and some ing to increased snow levels and a skewing away from
reach 250–300 mm (8–12 in.; Fig. 8b). Many sites in benefits toward hazards as heavy rainfall occurs at
Oregon receive 100–150 mm, while sites as distant as higher elevations. Indeed, Fig. 9 (bottom) shows that
Idaho and Utah receive 50–100 mm (a significant, but on days with AR landfall near Bodega Bay (together
mostly beneficial, amount given the dryness of many with the subsequent day), temperatures across the Cali-
inland areas). Figure 8c shows the maximum 3-day ac- fornia–Nevada region become increasingly warmer as
cumulation for each COOP
site during any of the 10
AR Cat 5 events and reveals
that many sites in North-
ern California experienced
greater than 300 mm of
precipitation, and some over
500 mm (~20 in.) during
at least one of the AR Cat
5 events. It is notable that
events exceeding 400 mm
(~16 in.) in 3 days are very
rare nationally, with only
four COOP sites per year
reaching this threshold
(Ralph and Dettinger 2012).
The spatial distribu-
tion of precipitation over Fig. 10. Frequency of 3-day precipitation totals among COOP observing sta-
the western United States, tions in California, for AR Cat events that struck Bodega Bay, on Northern
during the occurrence of California’s coast.

AMERICAN METEOROLOGICAL SOCIETY FEBRUARY 2019 | 283


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
Fig. 11. (a) Analyzed maximum IVT and (b) analyzed AR duration used to calculate the analyzed AR Cat, (c) the
forecast AR Cat, and (d) the analyzed AR Cat based on the GFS forecast initialized at 0000 UTC 3 Feb 2017 and
valid during a 5-day forecast period from 1200 UTC 5 Feb 2017 through 1200 UTC 10 Feb 2017, which represents
forecast lead times from 60 to 180 h. Note that (a), (b), and (d) are based on GFS analyses during this period,
whereas (c) is based on the GFS forecast. In (c) and (d), the gray star indicates the location of the Oroville Dam.
In (d), filled (unfilled) small and large dots indicate underforecasts (overforecasts) of 1 and 2 AR Cats in (c).

a function of AR Cat. This increases the chances that Cats during a 5-day forecast period valid from 1200
these storms will yield larger fractions of rain than UTC 5 February 2017 through 1200 UTC 10 February
snow, particularly at higher elevations that usually 2017, which represents forecast lead times from 60 to
receive snow, contributing to a greater risk of floods 180 h. Shown for this period are the analyzed maxi-
and related hazards across this region. mum IVT (Fig. 11a) and the analyzed AR duration
used to calculate the analyzed AR Cat (Fig. 11b), the
AR SCALE FORECAST EXAMPLE. The AR forecast AR Cat (Fig. 11c), and the analyzed AR Cat
scale can be implemented using forecast data, and a (Fig. 11d). Note that Figs. 11a, 11b, and 11d are based
preliminary concept is highlighted here by focusing on GFS analyses during this period, whereas Fig. 11c
on a series of landfalling AR events coinciding with is based on the GFS forecast.
the Oroville Dam crisis during early February 2017 In this example, the 60–180-h forecast indicates a
(Fig. 11). In this example, the GFS forecast initialized broad swath of AR Cat 5 conditions along the axis of
at 0000 UTC 3 February 2017 is used to calculate AR the AR, which verified. The forecast also highlights

284 | FEBRUARY 2019


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
the potential for AR Cat 4
conditions along the coast
of Northern California and
southern Oregon, as well as
AR Cat 3 conditions over
the northern Sierra Nevada
(near the Oroville Dam) and
AR Cat 2 conditions extend-
ing into the Great Basin
(Fig. 11c). In general, verifi-
cation is ~1 AR Cat stronger
than predicted forecast over
much of the western United
States, and ~1–2 AR Cat
stronger from coastal Cali-
fornia northeastward into
eastern Oregon, Idaho, and
southwestern Montana. AR
Cat 5 conditions occurred
over land north of San Fran-
cisco Bay, AR Cat 4+ condi-
tions occurred along a much
greater stretch of the coast
and over the Oroville Dam,
and AR Cat 3 conditions oc-
curred much farther inland
than anticipated (Fig. 11d).
Taking a broader view, the
Fig. 12. Average maximum AR Cat annually.
distribution of higher (low-
er) AR Cat verifications
along the northwestern and southeastern (south- duration to characterize AR strength, providing a
western and northeastern) edges of the AR suggests a framework for differentiating between the impacts of
more zonally oriented AR than was anticipated in the those that are primarily beneficial and those that are
forecast. In addition, the underforecast AR Cat over the primarily hazardous. The scale is also readily applied
northern Rockies indicates that the GFS struggled to to gridded datasets such as atmospheric reanalyses,
capture the inland penetration of the AR, particularly weather forecasts, and climate projections. Given an
downstream from the core of the AR. Hence, we stress increased focus on AR-related science and impacts,
that while the AR scale provides a solid framework for it is likely that this AR Cat scale will be widely used
characterizing the strength and impacts of an AR, it to communicate the benefits and hazards associated
is dependent on an accurately modeled atmospheric with ARs both in the western United States and in
forecast, and will only be as reliable as the forecast other regions where ARs contribute strongly to hy-
model being used. drometeorological impacts.
While the concept shown here is only preliminary, In addition to introducing the AR Cat scale, this
it is the authors’ plan to implement the AR forecast paper provides a few diagnostics regarding their fre-
operationally over the coming months, with a fin- quency of occurrence and their impacts on precipita-
ished product likely resembling that shown here. tion. Figure 12 shows the average annual maximum
AR Cat event across the western United States and
SUMMARY AND FUTURE WORK. This paper illustrates that the scale is broadly applicable across
presents a scale to characterize atmospheric river the region. AR Cat 4 or 5 events occur primarily
strength and impacts in a way that is both useful along the Northern California and Pacific North-
to scientists and conducive to communication with west coasts, whereas the strongest ARs affecting the
nonexperts. The scale uses the intensity of verti- Southern California coast at roughly annual scales are
cally integrated water vapor transport and AR event typically AR Cat 2 or 3. Most of the West (except the

AMERICAN METEOROLOGICAL SOCIETY FEBRUARY 2019 | 285


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
highest interior) experience AR Cat 1 events at least APPENDIX: KEY TERMINOLOGY AND
annually. In addition, the well-established connec- CALCULATION OF IVT. A clear understand-
tion between ARs and extreme precipitation in the ing of the AR scale presented here benefits from the
western United States is highlighted, revealing that definition of a few key terms.
the strongest AR events often affect the entire West
to some degree. It was shown that 30% of landfalling • An atmospheric river (AR) is a long, narrow,
AR Cat 4 and 5 events (combined) were associated transient corridor of strong horizontal water
with significant floods on the Russian River, and that vapor transport that is typically associated with a
all of the floods in the period studied were associated low-level jet stream ahead of the cold front of an
with landfalling ARs. extratropical cyclone (American Meteorological
A few shortcomings of the AR scale are worth Society 2018). An AR can be thought of as both
pointing out. First, some high-impact events (e.g., an Eulerian object and a Lagrangian process—it
debris flows) result primarily from very intense rain- encompasses a three-dimensional region of the
fall rates that can be short-lived (Oakley et al. 2017). atmosphere in which certain dynamic processes
Hence, the duration requirement of the scale may are taking place, with a shape and location that
overexert itself by not recognizing weather systems change as these processes evolve.
that will be short-lived, but capable of producing such • AR conditions denote the instantaneous presence
intense rates. Second, the impacts associated with of IVT ≥ 250 kg m–1 s–1 at a point (i.e., in an Eule-
a particular AR Cat may vary due to other factors rian sense).
such as temperature throughout the column, and • An AR event refers to the full period of time over
particularly the rain/snow line. More broadly, since which AR conditions occur at a fixed geographical
the AR scale is not linked to a particular location, point (i.e., applies in an Eulerian sense). Note that
the impacts associated with the AR Cats will vary the AR scale presented is based on AR conditions
spatially as a function of geographic considerations and AR events at a point and does not account for
such as topography, land surface type, and anteced- the movement or tracking of ARs themselves (i.e.,
ent conditions. as objects).
Future work is envisioned to more fully describe • The spatial and temporal distribution of AR events
the transition between primarily beneficial and as a function of AR scale category relies on verti-
primarily hazardous AR Cat events in terms of a va- cally integrated vapor transport (IVT), which is
riety of hydrometeorological benefits (e.g., increased calculated as
reservoir levels, drought relief, and water supply) p
1 t
g p∫b
and hazards (e.g., heavy rain, flooding, and high IVT = − qVh dp, (A1)
winds). Since the AR scale is easily applied to grid-
ded datasets, these analyses can shed light on how where q is the specific humidity, Vh is the horizon-
the relationship between AR Cat and impacts varies tal wind vector, g is the acceleration due to gravity,
geographically. The unique mesonet throughout pb is 1,000 hPa, and pt is 200 hPa (e.g., Neiman et al.
the western United States (Ralph et al. 2014; White 2008; Moore et al. 2012). The IVT is calculated us-
et al. 2013), combined with storm reports from the ing atmospheric data from the MERRA reanalysis
National Weather Service, is uniquely tailored for at 0.5° × 0.625° horizontal resolution (Rienecker
this work. The scale presented here can also be used, et al. 2011). Vertical levels used to calculate IVT
along with verification studies such as those by Wick are every 25 hPa from 1,000 to 700 hPa and every
et al. (2013), DeFlorio et al. (2018), and Nardi et al. 50 hPa from 700 to 200 hPa.
(2018), to reveal whether AR forecast skill varies as a
function of AR Cat over different locales. An intrigu-
ing new direction is emerging on the correspondence REFERENCES
between strong ARs and strong extratropical cyclo- Albano, C., M. Dettinger, and C. Soulard, 2017: Influ-
genesis (Eiras-Barca et al. 2018; Zhang et al. 2019), ence of atmospheric rivers on vegetation produc-
which could benefit from an objective scaling of ARs. tivity and fire patterns in the southwestern U.S. J.
Finally, the AR scale, along with information from Geophys. Res. Biogeosci., 122, 308–323, https://doi
numerical weather prediction model ensembles, can .org/10.1002/2016JG003608.
be leveraged to produce gridded, probabilistic fore- American Meteorological Society, 2018: Atmospheric riv-
casts of AR intensity, which should prove valuable to er. Glossary of Meteorology, http://glossary.ametsoc
a variety of users. .org/wiki/Atmospheric_river.

286 | FEBRUARY 2019


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
Backes, T. M., M. L. Kaplan, R. Schumer, and J. F. —, —, and D. Lavers, 2015: Setting the stage for a glob-
Mejia, 2015: A climatology of the vertical structure al science of atmospheric rivers. Eos, Trans. Amer. Geo-
of water vapor transport to the Sierra Nevada in phys. Union, 96, https://doi.org/10.1029/2015EO038675.
cool season atmospheric river precipitation events. J. —, —, and J. J. Rutz, 2018: Empirical return periods
Hydrometeor., 16, 1029–1047, https://doi.org/10.1175 of the most intense vapor transports during historical
/JHM-D-14-0077.1. atmospheric river landfalls on the U.S. West Coast. J.
Bao, J.-W., S. A. Michelson, P. J. Neiman, F. M. Ralph, Hydrometeor., 19, 1363–1377, https://doi.org/10.1175
and J. M. Wilczak, 2006: Interpretation of enhanced /JHM-D-17-0247.1.
integrated water vapor bands associated with extra- Eiras-Barca, J., A. M. Ramos, J. G. Pinto, R. M. Trigo,
tropical cyclones: Their formation and connection to M. L. R. Liberato, and G. Miguez-Macho, 2018: The
tropical moisture. Mon. Wea. Rev., 134, 1063–1080, concurrence of atmospheric rivers and explosive
https://doi.org/10.1175/MWR3123.1. cyclogenesis in the North Atlantic and North Pa-
Cheng, B. S., A. L. Chang, A. Deck, and M. C. Ferner, cific basins. Earth Syst. Dyn., 9, 91–102, https://doi
2016: Atmospheric rivers and the mass mortality of .org/10.5194/esd-9-91-2018.
wild oysters: Insight into an extreme future? Proc. Elsner, J., and A. B. Kara, 1999: Hurricanes of the North
Biol. Sci., 283, 20161462, https://doi.org/10.1098 Atlantic: Climate and Society. Oxford University
/rspb.2016.1462. Press, 488 pp.
Cordeira, J. M., F. M. Ralph, and B. J. Moore, 2013: The Florsheim, J., and M. Dettinger, 2015: Promoting atmo-
development and evolution of two atmospheric riv- spheric-river and snowmelt-fueled biogeomorphic
ers in proximity to western North Pacific tropical processes by restoring river-floodplain connectivity
cyclones in October 2010. Mon. Wea. Rev., 141, 4234– in California’s Central Valley. Geomorphic Approach-
4255, https://doi.org/10.1175/MWR-D-13-00019.1. es to Integrated Floodplain Management of Lowland
—, —, A. Martin, N. Gaggini, J. R. Spackman, P. J. Fluvial Systems in North America and Europe, P.
Neiman, J. J. Rutz, and R. Pierce, 2017: Forecasting Hudson and H. Middelkoop, Eds., Springer, 119–141,
atmospheric rivers during CalWater 2015. Bull. Amer. https://doi.org/10.1007/978-1-4939-2380-9_6.
Meteor. Soc., 98, 449–459, https://doi.org/10.1175 Fujita, T., 1981: Tornadoes and downbursts in the con-
/BAMS-D-15-00245.1. text of generalized planetary scales. J. Atmos. Sci.,
Dacre, H. F., P. A. Clark, O. Martinez-Alvarado, M. A. 38, 1511–1534, https://doi.org/10.1175/1520-0469
Stringer, and D. A. Lavers, 2015: How do atmospheric (1981)038<1511:TADITC>2.0.CO;2.
rivers form? Bull. Amer. Meteor. Soc., 96, 1243–1255, Guan, B., and D. E. Waliser, 2015: Detection of atmo-
https://doi.org/10.1175/BAMS-D-14-00031.1. spheric rivers: Evaluation and application of an algo-
DeFlorio, M. J., D. E. Waliser, B. Guan, D. A. Lavers, rithm for global studies. J. Geophys. Res. Atmos., 120,
F. M. Ralph, and F. Vitart, 2018: Global assessment 12 514–12 535, https://doi.org/10.1002/2015JD024257.
of atmospheric river prediction skill. J. Hydrome- —, N. P. Molotch, D. E. Waliser, E. J. Fetzer, and P.
teor., 19, 409–426, https://doi.org/10.1175/JHM J. Neiman, 2010: Extreme snowfall events linked
-D-17-0135.1. to atmospheric rivers and surface air temperature
Dettinger, M. D., 2004: Fifty-two years of “Pineapple- via satellite measurements. Geophys. Res. Lett., 37,
Express” storms across the west coast of North L20401, https://doi.org/10.1029/2010GL044696.
America. PIER Project Rep. CEC-500-2005-004, —, D. E. Waliser, F. M. Ralph, E. J. Fetzer, and P. J.
California Energy Commission, 15 pp., www.energy Neiman, 2016: Hydrometeorological characteristics
.ca.gov/2005publications/CEC-500-2005-004/CEC of rain-on-snow events associated with atmospheric
-500-2005-004.PDF. rivers. Geophys. Res. Lett., 43, 2964–2973, https://doi
—, 2013: Atmospheric rivers as drought busters on .org/10.1002/2016GL067978.
the U.S. West Coast. J. Hydrometeor., 14, 1721–1732, Hatchett, B. J., S. Burak, J. J. Rutz, N. S. Oakley, E. H.
https://doi.org/10.1175/JHM-D-13-02.1. Bair, and M. L. Kaplan, 2017: Avalanche fatalities
—, and D. Cayan, 2014: Drought and the California during atmospheric river events in the western
delta—A matter of extremes. San Francisco Estu- United States. J. Hydrometeor., 18, 1359–1374, https://
ary Watershed Sci., 12 (2), https://doi.org/10.15447 doi.org/10.1175/JHM-D-16-0219.1.
/sfews.2014v12iss2art4. Herbst, D. B., and S. D. Cooper, 2010: Before and af-
—, F. M. Ralph, T. Das, P. J. Neiman, and D. Cayan, ter the deluge—Rain-on-snow flooding effects on
2011: Atmospheric rivers, floods, and the water re- aquatic invertebrate communities of small streams
sources of California. Water, 3, 455–478, https://doi in the Sierra Nevada, California. J. N. Amer. Benthol.
.org/10.3390/w3020445. Soc., 29, 1354–1366, https://doi.org/10.1899/09-185.1.

AMERICAN METEOROLOGICAL SOCIETY FEBRUARY 2019 | 287


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
Khouakhi, A., and G. Villarini, 2016: On the relationship Oakley, N. S., J. T. Lancaster, M. L. Kaplan, and F. M.
between atmospheric rivers and high sea water levels Ralph, 2017: Synoptic conditions associated with
along the U.S. West Coast. Geophys. Res. Lett., 43, cool season post-fire debris flows in the Transverse
8815–8822, https://doi.org/10.1002/2016GL070086. Ranges of southern California. Nat. Hazards, 88,
Kocin, P. J., and L. W. Uccellini, 2004: A snowfall im- 327–354, https://doi.org/10.1007/s11069-017-2867-6.
pact scale derived from Northeast storm snowfall Ralph, F. M., and M. D. Dettinger, 2012: Historical and
distributions. Bull. Amer. Meteor. Soc., 85, 177–194, national perspectives on extreme West Coast precipi-
https://doi.org/10.1175/BAMS-85-2-177. tation associated with atmospheric rivers during De-
Konrad, C. P., and M. D. Dettinger, 2017: Flood runoff cember 2010. Bull. Amer. Meteor. Soc., 93, 783–790,
in relation to water vapor transport by atmospheric https://doi.org/10.1175/BAMS-D-11-00188.1.
rivers over the western United States, 1949–2015. —, P. J. Neiman, and G. A. Wick, 2004: Satellite and
Geophys. Res. Lett., 44, 11 456–11 462, https://doi CALJET aircraft observations of atmospheric rivers
.org/10.1002/2017GL075399. over the eastern North Pacific Ocean during the
Lamjiri, M. A., M. D. Dettinger, F. M. Ralph, and winter of 1997/98. Mon. Wea. Rev., 132, 1721–1745,
B. Guan, 2017: Hourly storm characteristics along https://doi.org/10.1175/1520-0493(2004)132<1721:SA
the U.S. West Coast: Role of atmospheric rivers in CAOO>2.0.CO;2.
extreme precipitation. Geophys. Res. Lett., 44, 11 456– —, —, —, S. I. Gutman, M. D. Dettinger, D. R.
11 462, https://doi.org/10.1002/2017GL075399. Cayan, and A. B. White, 2006: Flooding on Califor-
Lavers, D. A., R. P. Allan, E. F. Wood, G. Villa- nia’s Russian River: Role of atmospheric rivers. Geo-
rini, D. J. Brayshaw, and A. J. Wade, 2011: Winter phys. Res. Lett., 33, L13801, https://doi.org/10.1029
f loods in Britain are connected to atmospheric /2006GL026689.
rivers. Geophys. Res. Lett., 38, L23803, https://doi —, —, G. N. Kiladis, K. Weickman, and D. W.
.org/10.1029/2011GL049783. Reynolds, 2011: A multi-scale observational case
—, D. E. Waliser, F. M. Ralph, and M. D. Dettinger, study of a Pacific atmospheric river exhibiting
2016: Predictability of horizontal water vapor trans- tropical-extratropical connections and a mesoscale
port relative to precipitation: Enhancing situational frontal wave. Mon. Wea. Rev., 139, 1169–1189, https://
awareness for forecasting western U.S. extreme doi.org/10.1175/2010MWR3596.1.
precipitation and flooding. Geophys. Res. Lett., 43, —, T. Coleman, P. J. Neiman, R. J. Zamora, and M. D.
2275–2282, https://doi.org/10.1002/2016GL067765. Dettinger, 2013: Observed impacts of duration and
Moore, B. J., P. J. Neiman, F. M. Ralph, and F. Barthold, seasonality of atmospheric-river landfalls on soil
2012: Physical processes associated with heavy flood- moisture and runoff in coastal Northern California.
ing rainfall in Nashville, Tennessee and vicinity J. Hydrometeor., 14, 443–459, https://doi.org/10.1175
during 1–2 May 2010: The role of an atmospheric /JHM-D-12-076.1.
river and mesoscale convective systems. Mon. Wea. —, and Coauthors, 2014: A vision for future obser-
Rev., 140, 358–378, https://doi.org/10.1175/MWR vations for western U.S. extreme precipitation and
-D-11-00126.1. flooding. J. Contemp. Water Res. Educ., 153, 16–32,
Nardi, K. M., E. A. Barnes, and F. M. Ralph, 2018: As- https://doi.org/10.1111/j.1936-704X.2014.03176.x.
sessment of numerical weather prediction model —, and Coauthors, 2016: CalWater field studies de-
reforecasts of the occurrence, intensity, and location signed to quantify the roles of atmospheric rivers
of atmospheric rivers along the West Coast of North and aerosols in modulating U.S. West Coast pre-
America. Mon. Wea. Rev., 146, 3343–3362, https:// cipitation in a changing climate. Bull. Amer. Meteor.
doi.org/10.1175/MWR-D-18-0060.1. Soc., 97, 1209–1228, https://doi.org/10.1175/BAMS
Neiman, P. J., F. M. Ralph, G. A. Wick, J. D. Lundquist, -D-14-00043.1.
and M. D. Dettinger, 2008: Meteorological char- —, and Coauthors, 2017a: Atmospheric rivers emerge
acteristics and overland precipitation impacts of as a global science and applications focus. Bull. Amer.
atmospheric rivers affecting the West Coast of North Meteor. Soc., 98, https://doi.org/10.1175/BAMS-D
America based on eight years of SSM/I satellite -16-0262.1.
observations. J. Hydrometeor., 9, 22–47, https://doi —, and Coauthors, 2017b: Dropsonde observations
.org/10.1175/2007JHM855.1. of total water vapor transport within North Pacific
—, L. J. Schick, F. M. Ralph, M. Hughes, and G. A. atmospheric rivers. J. Hydrometeor., 18, 2577–2596,
Wick, 2011: Flooding in western Washington: The https://doi.org/10.1175/JHM-D-17-0036.1.
connection to atmospheric rivers. J. Hydrometeor., 12, —, M. D. Dettinger, M. M. Cairns, T. J. Galarneau,
1337–1358, https://doi.org/10.1175/2011JHM1358.1. and J. Eylander, 2018: Defining “atmospheric river”:

288 | FEBRUARY 2019


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
How the Glossary of Meteorology helped resolve a characteristics. J. Hydrometeor., 12, 481–507, https://doi
debate. Bull. Amer. Meteor. Soc., 99, 837–839, https:// .org/10.1175/2010JHM1284.1.
doi.org/10.1175/BAMS-D-17-0157.1. Waliser, D., and B. Guan, 2017: Extreme winds and
Ramos, A. M., R. M. Trigo, M. L. R. Liberato, and precipitation during landfall of atmospheric rivers.
R. Tome, 2015: Daily precipitation extreme events Nat. Geosci., 10, 179–183, https://doi.org/10.1038
in the Iberian Peninsula and its association with /ngeo2894.
atmospheric rivers. J. Hydrometeor., 16, 579–597, Warner, M. D., C. F. Mass, and E. P. Salathé Jr., 2012:
https://doi.org/10.1175/JHM-D-14-0103.1. Wintertime extreme precipitation events along the
Rienecker, M., and Coauthors, 2011: MERRA: NASA’s Pacific Northwest coast: Climatology and synoptic
Modern-Era Retrospective Analysis for Research and evolution. Mon. Wea. Rev., 140, 2021–2043, https://
Applications. J. Climate, 24, 3624–3648, https://doi doi.org/10.1175/MWR-D-11-00197.1.
.org/10.1175/JCLI-D-11-00015.1. White, A. B., and Coauthors, 2013: A twenty-first-century
Rutz, J. J., W. J. Steenburgh, and F. M. Ralph, 2014: california observing network for monitoring extreme
Climatological characteristics of atmospheric rivers weather events. J. Atmos. Oceanic Technol., 30, 1585–
and their inland penetration over the western United 1603, https://doi.org/10.1175/JTECH-D-12-00217.1.
States. Mon. Wea. Rev., 142, 905–921, https://doi Wick, G. A., P. J. Neiman, F. M. Ralph, and T. M.
.org/10.1175/MWR-D-13-00168.1. Hamill, 2013: Evaluation of forecasts of the water
—, —, and —, 2015: The inland penetration of at- vapor signature of atmospheric rivers in operational
mospheric rivers over western North America: A La- numerical weather prediction models. Wea. Fore-
grangian analysis. Mon. Wea. Rev., 143, 1924–1944, casting, 28, 1337–1352, https://doi.org/10.1175/WAF
https://doi.org/10.1175/MWR-D-14-00288.1. -D-13-00025.1.
Shields, C. A., and Coauthors, 2018: Atmospheric Wuebbles, D. J., and Coauthors, 2017: Our globally
River Tracking Method Intercomparison Project changing climate. Climate Science Special Report:
(ARTMIP): Project goals and experimental design. Fourth National Climate Assessment, Vol. 1, D. J.
Geosci. Model Dev., 11, 2455–2474, https://doi Wuebbles et al., Eds., U.S. Global Change Research
.org/10.5194/gmd-11-2455-2018. Program, 35–72, https://doi.org/10.7930/J08S4N35.
Sodemann, H., and A. Stohl, 2013: Moisture origin Young, A. M., K. T. Skelly, and J. M. Cordeira, 2017:
and meridional transport in atmospheric rivers and High-impact hydrologic events and atmospheric
their association with multiple cyclones. Mon. Wea. rivers in California: An investigation using the
Rev., 141, 2850–2868, https://doi.org/10.1175/MWR NCEI Storm Events Database. Geophys. Res. Lett., 44,
-D-12-00256.1. 3393–3401, https://doi.org/10.1002/2017GL073077.
Stohl, A., C. Forster, and H. Sodemann, 2008: Remote Zhang, Z., F. M. Ralph, and M. Zheng, 2019: The rela-
sources of water vapor forming precipitation on tionship between extratropical cyclone strength and
the Norwegian west coast at 60° N—A tale of hur- atmospheric river intensity and position. Geophys. Res.
ricanes and an atmospheric river. J. Geophys. Res., Lett., https://doi.org/10.1029/2018GL079071, in press.
113, D05102, https://doi.org/10.1029/2007JD009006. Zhu, Y., and R. E. Newell, 1998: A proposed algorithm
Viale, M., and M. N. Nuñez, 2011: Climatology of win- for moisture fluxes from atmospheric rivers. Mon.
ter orographic precipitation over the subtropical Wea. Rev., 126, 725–735, https://doi.org/10.1175/1520
central Andes and associated synoptic and regional -0493(1998)126<0725:APAFMF>2.0.CO;2.

AMERICAN METEOROLOGICAL SOCIETY FEBRUARY 2019 | 289


Unauthenticated | Downloaded 05/15/23 03:34 PM UTC
Attention AMS Student Members

Stay connected to AMS after graduation


for half the regular membership rate

Let AMS help you build your expertise, your


network, your career. There’s never been a more
important time to be a member.

http://www.ametsoc.org/earlycareer
Unauthenticated | Downloaded 05/15/23 03:34 PM UTC

You might also like