a. Heavy Wintertime
Precipitation in the Pacific Northwest
The Pacific Northwest region of the United States receives its most
substantial rains during the cool season. The largest totals are typically observed during
November–February along the coast, and also along the windward slopes of the mountain
ranges that extend from Canada southward through western Washington and Oregon to central
California. During November 1998–February 1999, much of the coastal region recorded
more than 1600 mm of precipitation, with the largest totals exceeding 2000 mm over the
Olympic Peninsula of Washington and northwestern Oregon (Fig. 24a).
Throughout the coastal region, precipitation totals were 400–1200 mm above the
long-term average (Fig. 24b), with many locations recording
150%–250% of normal total seasonal precipitation and some areas recording more than
300% of normal precipitation.
Much of the windward slopes of the Cascade Mountains in Washington and
Oregon, along with portions of the Sierra Nevada range in northern California, also
received more than 1200 mm of precipitation during the period. Precipitation totals in
each of these regions were 200–800 mm above the long-term average, with many
locations recording 150%–250% of normal seasonal totals. At higher elevations much of
this precipitation fell as snow, resulting in an above-average snowpack in all of the
mountain ranges in the Pacific Northwest. An extreme snowfall accumulation of 28.96 m
(1140 inches) was recorded on Mt. Baker in Washington State, which set a new single-season
record for the most snowfall ever measured in the United States. The previous snowfall
record was 28.5 m (1122 inches), recorded at Mt Rainier/ Paradise Station during the
1971/72 winter season.
The excessive precipitation in the Pacific Northwest is highlighted for
two individual stations: Seattle, Washington (Fig. 25a) and
Portland, Oregon (Fig. 26a). Both stations recorded surplus
precipitation beginning in early November 1998, with totals by the end of February 1999
reaching 875 mm (165% of normal). Both stations experienced prolonged periods of
substantial precipitation during these 4-months, interspersed often with only a few days
of dryness between major precipitation events (Figs. 25b, 26b). The most prolonged periods of nearly continuous precipitation
at both stations were recorded between mid-November and mid-December, and from mid-January
through the end of February.
This above-average precipitation resulted from a series of major winter
storms which were associated with an amplified storm track that extended from the central
North Pacific to the Pacific Northwest and southwestern Canada (Fig.
27). This enhanced storm track was located within a region of enhanced westerly winds
along the poleward (cyclonic-shear) side of the mean wintertime jet stream (compare Fig. 27 with Fig. 21), and also contributed
to an enhanced flow of relatively mild marine air into western North America throughout
the period. This anomalous westerly flow was linked to a large-scale pattern characterized
by below-average heights over the high latitudes of the North Pacific, and above-average
heights over the central North Pacific in association with ongoing La Niña conditions
(see section 3e(1), Fig. 20b).
b. The 1999 North Atlantic and Eastern
North Pacific Hurricane Season
(i) The North Atlantic hurricane season
The North Atlantic hurricane season runs from June through November and
exhibits a peak in activity between mid-August and mid-October, primarily in response to
systems developing from African easterly wave disturbances. In an average season,
9–10 tropical storms are generally observed over the North Atlantic, with 5–6
becoming hurricanes and 2 reaching intense hurricane status [measured by a category 3, 4,
or 5 on the Saffir-Simpson scale (Simpson 1974)]. The 1999 season featured 12 named
storms, with 8 of these systems becoming hurricanes and 5 reaching intense hurricane
status. Most systems developed over the tropical Atlantic, Caribbean Sea and the southern
Gulf of Mexico, which is typical of the main development region observed during other
active years (Shapiro and Goldenberg 1998). Only one system during the 1999 season
developed prior to 18 August, and four systems developed after 12 October. Three of these
late-season storms became hurricanes, with the last (Hurricane Lenny) reaching category-4
status. Lenny developed in mid-November and moved eastward across the central Caribbean
Sea. This unusual track enabled it to become the first hurricane to strike the Lesser
Antilles Islands from the west.
Five tropical systems made landfall in the United States during the
1999 season. The first of these was intense Hurricane Bret, which hit a sparsely populated
region of southeastern Texas in late August. The second and third landfalling systems were
Tropical Storm Dennis and Hurricane Floyd, which produced extremely large rainfall totals
over southeastern Virginia and eastern North Carolina and led to record flooding in some
areas [see section 4a(4ii)]. The fourth was Tropical Storm
Harvey, which formed in the Gulf of Mexico and impacted southern Florida and the Florida
Keys. The fifth was Hurricane Irene, which also moved across southern Florida and produced
considerable freshwater flooding in that region.
Two measures of overall seasonal activity are used extensively by Gray
and colleagues (personal communication). The first is Net Tropical Cyclone Activity (NTC),
which is based on a linear combination of the number of named storms, number of
hurricanes, number of intense hurricanes, number of named storm days, number of hurricane
days, and number of intense hurricane days. However, these six parameters represent highly
discrete distributions, and are also strongly statistically correlated. The second measure
of overall seasonal activity is referred to as the Hurricane Destruction Potential (HDP),
which is calculated by summing the squares of the estimated 6-hourly maximum sustained
wind speed (Vmax2) for all periods in which the system is a hurricane. This
index represents a single, continuous distribution that implicitly accounts for numbers of
hurricanes, yet also gives more weight to strong systems and long-lasting systems. A
slight modification of the HDP index involves accumulating Vmax2 for all
6-hourly periods in which the system is either a tropical storm or hurricane, thereby also
accounting for the number and duration of storms while at a Tropical Storm status. This
modified HDP index is referred to as Accumulated Cyclone Energy (ACE) index (Fig. 28), and is both a physically and statistically reasonable
measure of overall activity during a given hurricane season.
Overall, the 1999 hurricane season was extremely active by all
measures, and according to the ACE index was ranked seventh in terms of overall activity
since 1950 (Fig. 28). This index also indicates that four of the
seven most active years since 1950 have occurred since 1995. This frequency of occurrence
of extremely active years since 1995 is in sharp contrast to the 1970–94 period,
which was marked by generally reduced overall tropical storm and hurricane activity.
Tropical storm and hurricane activity over the North Atlantic is
strongly influenced by the vertical wind shear between the upper (200-hPa) and lower (850
hPa) levels of the atmosphere over the western Atlantic and Caribbean Sea. Active years
feature low vertical shear (less than approximately 8 m s-1 in these regions,
while inactive years feature high vertical shear.
The August–October 1999 period featured very low wind shear across
the western tropical Atlantic, Caribbean Sea, and southern Gulf of Mexico (Fig. 29a), with mean shear values in some regions dropping below 4 m
s-1. Over much of the tropical Atlantic, this shear was anomalously low (Fig. 29b), which is typical of other active hurricane seasons
(Landsea et al. 1998, Goldenberg and Shapiro 1997). Over the Caribbean Sea and western
Atlantic, these low shear values resulted from a combination of anomalous upper-level
easterly winds averaging 2–4 m s-1 (Fig. 30a) and
anomalous low-level westerly winds averaging 1–3 m s-1 (Fig. 30b). At upper levels, these easterly wind anomalies covered
the entire Atlantic in both hemispheres between 20°N and 20°S, and also extended across
the entire tropical and subtropical eastern Pacific. At lower levels, the pattern of
westerly wind anomalies was also quite extensive, spanning the area from the eastern North
Pacific eastward to the African Sahel region (Fig. 30b). These
upper-level and lower-level wind anomalies were linked to a global-scale circulation
pattern featuring positive upper-level streamfunction anomalies across the entire middle
latitudes of the Northern Hemisphere (see section 3e(2), Fig. 23c)
and negative streamfunction anomalies across the entire subtropical and middle latitudes
of the Southern Hemisphere.
Tropical cyclogenesis in active years is often associated with African
easterly waves, whose evolution is strongly influenced by the structure and location of
the African Easterly Jet (AEJ). The AEJ is centered near 600–700 hPa, and provides
the background flow within which the African easterly waves move and evolve (Reed et al.
1977). There are notable differences in both the structure and location of the AEJ between
active and inactive years (Bell and Halpert 1998, Halpert and Bell 1997), and these
differences represent another fundamental component of the coherent atmospheric
variability associated with active and inactive hurricane seasons (Bell and Chelliah
1999).
The AEJ normally extends from western Africa to the central subtropical
North Atlantic (Fig. 31a), with the jet core located near 15°N.
The jet reaches peak strength between the 600–700 hPa levels and provides the
"steering flow" for the easterly waves. It is also an important energy source
for the waves, which propagate through the cyclonic shear zone (Fig.
31b) along the equatorward flank of the jet (Reed et al. 1977). This cyclonic shear
zone is normally well-defined over the eastern tropical Atlantic and western Africa
between 8°–15°N and overlaps the area of low vertical wind shear (indicated by
shading in Figs. 31a, b). This overlap is generally most
extensive during the climatological September peak in hurricane activity.
There is considerable year-to-year variability in the latitudinal
position of the AEJ, in the westward extent of the AEJ, and in the strength and westward
extent of the cyclonic vorticity along the southern flank of the jet (Bell and Chelliah
1999). During the active August–October 1999 season, the AEJ was very well-defined
and located between 17.5°–20°N (Fig. 31c), which is
approximately 2.5°–5°N of its climatological mean position. This position of the
AEJ during 1999 is typical of other active hurricane seasons, and is also considerably
farther north than its mean latitude (10°–12°N) often observed during inactive
years (Bell and Halpert 1998). The AEJ also featured strong cyclonic-shear along its
entire equatorward flank between 10°–15°N during 1999 (Fig.
31d), along with a westward extension of this high vorticity zone to the central
tropical Atlantic. These high values of cyclonic vorticity resulted from large westerly
wind anomalies near 10°N throughout the lower troposphere over the central and eastern
Atlantic (Fig. 30b). These high vorticity values and their
pronounced westward extent are typical of other active hurricane seasons. They are also
distinct from the climatological mean conditions (Fig. 31b), and
from inactive hurricane years which feature low values of cyclonic vorticity north of
10°N and a confinement of high cyclonic vorticity to the eastern tropical Atlantic (e.g.,
Bell and Halpert 1998, their Fig. 34).
Also during 1999, the region of high cyclonic vorticity strongly
overlapped the area of low vertical wind shear (indicated by the shading) over the central
Atlantic (Fig. 31d). While the area of overlap between the 1999
season and the climatological normal (Fig. 31b) is similar, the
magnitude of cyclonic relative vorticity is much stronger in the overlap region during the
1999 season. This favorable location and horizontal structure of the African easterly jet,
combined with its proximity to the extended region of low vertical wind shear, contributed
to the recurring tropical cyclogenesis and intense hurricane development from easterly
waves observed during the season. The increased cyclonic vorticity and westerly wind
anomalies at low levels were also evident across the African Sahel region, and contributed
to enhanced July–September 1999 rainfall in that region [see section 4b(2)].
This variability in the AEJ is also strongly coupled to the prominent
global-scale mode of upper-level streamfunction anomalies (Fig. 23c),
which is ultimately linked to the patterns of tropical convection. As discussed in section 3e(2), this mode was established early in the
year (Fig. 23b), and neither the La Niña nor its attendant
convection patterns dissipated during the next several months. The expected persistence of
the global mode during the summer months led to a strong expectation that the 1999
hurricane season would indeed feature anomalous upper-level easterly winds across the
tropical Atlantic, low vertical wind shear across the central and western tropical
Atlantic, and a structure and location of the AEJ which was similar to the observed
features. As a result, it was evident that there was a strong likelihood of above-normal
tropical storm and hurricane activity over the North Atlantic even well prior to the onset
of the season.
(ii) Eastern North Pacific hurricane season
The eastern North Pacific hurricane season runs from May through
November. On average, there are 16 named storms during the season, 9 of which typically
become hurricanes and 4 of which become intense hurricanes. Overall, the 1999 season was
one of the most inactive on record with only 9 tropical storms, 6 of which became
hurricanes and 2 of which became intense hurricanes. This relative inactivity was linked
to a large region of high vertical wind shear that covered much of the main development
region of the eastern North Pacific for most of the season (Fig. 29a).
There is often a negative correlation between Pacific basin and
Atlantic basin hurricane activity, with relatively inactive Pacific hurricane seasons
accompanying active Atlantic hurricane seasons, and vice versa. This relationship results
from the large-scale pattern of anomalous vertical wind shear which, during active
Atlantic hurricane seasons, is below normal across the North Atlantic and above normal
over the eastern tropical North Pacific. This dipole pattern of shear anomalies was
prominent during the 1999 hurricane season (Fig. 29b), and was
related to a combination of upper-level easterly wind anomalies (Figs.
30a) and lower-level westerly wind anomalies (Fig. 30b)
across the North Atlantic and eastern North Pacific.
c. The July–September 1999 Southwestern
U. S. Monsoon
During July–September precipitation over Mexico and the
southwestern U. S. is controlled by the continental-scale monsoon circulation. Interannual
variations in this summer monsoon can be influenced by various ocean-based and land-based
factors (e.g., SST, soil moisture) that provide sources of "memory" of
antecedent climate anomalies such as extremes in the ENSO cycle. In a recent study Higgins
et al. (1998) showed that wet summer monsoons in the southwestern U. S. often follow
winters characterized by a combination of dry conditions in the southwestern U. S. and wet
conditions in the Pacific Northwest, as is typical of La Niña episodes. Consistent with
this finding, Gutzler and Preston (1997) showed that interannual fluctuations of summer
rainfall in New Mexico are linked to antecedent spring snowpack over the southern U. S.
Rocky Mountains. They argue that La Niña events during the preceding winter and spring
often contribute to a relatively poor snowpack in the southern Rockies, and vice-versa for
El Niño. Consequently, relatively less heating is required in low-snowpack years to
establish the necessary land-sea thermal contrast that helps drive the summer monsoon.
Under these circumstances wetter-than-normal conditions are often observed over the
southwestern United States during the monsoon season.
Overall, the 1999 southwestern U. S. monsoon season featured
substantial rainfall throughout the southwestern United States (Fig.
32a), with seasonal totals generally above-average throughout the region (Fig. 32b). The largest anomalies during the season were observed in
Arizona and New Mexico, where the combined rainfall total in these two states was the
second largest in the historical record dating back to 1960 (Fig. 33).
Much of central and northern Arizona, and portions of western New Mexico, recorded totals
that were more than 150 mm above-normal during the season, with many locations recording
more than twice the normal seasonal total.
The onset of the summer rains in the southwestern U.S. occurred during
the first week of July, which is very close to the climatologically favored time for this
region. Much of the monsoonal rainfall occurred during July and August, although heavy
precipitation also continued over much of Arizona during September. Farther south, Mexico
and Central America also experienced increased monsoon rainfall, particularly during June
and September with a midsummer dry period during August.
The wetter-than-normal conditions over the southwestern United States
and Mexico were linked to an enhanced upper-level monsoon anticyclone that was extended
northward of its climatological mean position (Fig. 34). This
feature was embedded in a much larger-scale anomalous large-scale anticyclonic circulation
that covered much of the United States, Mexico, and the adjacent ocean waters (Fig. 23c), and allowed enhanced monsoonal flow to penetrate the
southwestern United States from both the Gulf of Mexico and the Gulf of California. This
anomalous anticyclonic circulation was likely influenced by the strong 1998–99 La
Niña episode, and by the accompanying relatively poor snowpack in the southern Rockies
during the preceding two seasons (see section 2a, Figs. 4a, b). In
contrast, in the region downstream of this enhanced anticyclonic circulation,
significantly below-normal rainfall and severe drought conditions were observed across
large portions of the central and eastern United States during July and August [see section 4a(4)].
d. Drought in the United States
Drought impacted large portions of the United States during 1999, with
46 states experiencing at least some moisture deficits and 29 states affected by
severe– to– extreme drought. Overall, the most significant drought conditions
occurred in three general regions: (a) across large portions of the northeastern quadrant
of the country during April–August , with long-term dryness dating back to July 1998
in some areas; (b) in the Ohio Valley, Tennessee Valley, interior Southeast, middle and
lower Mississippi Valley, and Texas from July through November; and (c) in the Hawaiian
Islands throughout the year, with long-term dryness affecting some parts of the Islands
since October 1997. Elsewhere, less severe drought conditions also impacted interior
sections of the Pacific Northwest and Intermountain West from May through November.
The various droughts across the United States during 1999 were
associated with many agricultural and ecological impacts. According to the United States
Department of Agriculture (USDA) drought conditions during 1999 resulted in a 3.2%
decrease from 1998 in total national corn production, in a 4% decrease in soybean
production, and in a 9.6% decrease in the combined spring and winter wheat crops to levels
not seen since 1996. In addition, approximately 5.7 million acres were consumed by
wildfires across the United States during the year [National Interagency Coordination
Center (NICC)], which is 65% above the 1991–99 average and is the largest consumed
acreage recorded since 1996.
(i) The Northeast, Mid-Atlantic, central Appalachians, and middle
Ohio Valley
Severe drought occurred from April through mid-August 1999 in the
Northeast, the mid-Atlantic, central Appalachians, and the middle Ohio Valley regions (Fig. 35). Precipitation totals for this 4½-month period were less
than 300 mm across most of the region (Fig. 36a), and less than
200 mm in eastern Pennsylvania, northeastern Virginia and central Maryland. Totals were
generally more than 100 mm below the long-term average throughout this region (Fig. 36b), and 200–300 mm below-average in south-central Ohio,
parts of Virginia, central Maryland, and southeastern Pennsylvania. In many locations from
central Maryland northeastward to southeastern Maine, rainfall during the period was less
than half the long-term mean. In Delaware and Rhode Island, statewide precipitation totals
for April–July 1999 were the lowest in the last 105 years of record (Fig. 37a). Statewide totals were the second-lowest on record for
this period in Connecticut, Maryland, New Jersey, and West Virginia, and one of the seven
lowest on record in most of the remaining northeastern states.
The 1999 drought followed a prolonged period of below-average rainfall
that began in July 1998, and was interrupted by near-average to above-average
precipitation only during January–March 1999. For the July 1998–July 1999 period
as a whole, rainfall was 60–65% of normal, with the largest rainfall deficits
(exceeding 300 mm) observed in northern Virginia, northwestern Maryland and scattered
portions of southwestern Virginia, West Virginia, and southern Ohio (Fig.
38). Other parts of the mid-Atlantic region and central Appalachian Mountain region
recorded deficits exceeding 200 mm during the period. Statewide, July 1998–July 1999
was the driest such period in 104 years of record in Virginia, the second-driest in
Maryland, the third-driest in both New Jersey and West Virginia, and the fourth-driest in
Delaware (Fig. 37b).
This drought was associated with a variety of impacts across the
region. The USDA reported that corn crop production in Pennsylvania and Maryland dropped
47% and 23% from 1998, respectively, and the Virginia tobacco crop production fell 9%.
Also, hay production in West Virginia and Pennsylvania declined 15% to 30% from 1998.
Stream flows were generally below the 10th percentile level in the mid-Atlantic and
Northeast, according to the U. S. Geological Survey (USGS), and water supplies dropped to
alarmingly low levels at times in some areas, requiring local and state governments to
implement mandatory water restrictions. In the Chesapeake Bay area, both groundwater and
total inflow to the Bay dropped to near-record low levels during August.
The drought conditions were linked to an enhanced upper-level ridge
that persisted over the center of the country throughout the period (Fig.
34). This feature was embedded in a much larger-scale anomalous large-scale
anticyclonic circulation that covered much of the United States, Mexico, and the adjacent
ocean waters, and also contributed to enhanced monsoonal rains in the southwestern United
States and northern Mexico [see section 4a(3)].
(ii) Drought-ending rains in northeastern
quadrant of United States
Severe drought conditions in the northeastern United States ended
during mid-August–mid-October in response to exceptionally large rainfall totals
throughout the region (Fig. 39a). During this period rainfall
totals exceeded 750 mm in eastern North Carolina and northeastern South Carolina, and 300
mm along the entire eastern Seaboard from the Carolinas’ to central Maine. In eastern
North Carolina these totals were more than 500 mm above the long-term average (Fig. 39b). Elsewhere, totals were more than 200 mm above-average
over most of the drought-stricken region from eastern South Carolina to southern New York.
Two tropical cyclones, Tropical Storm Dennis in late August and
Hurricane Floyd in early September, contributed significantly to this excessive rainfall.
Rainfall totals from Tropical Storm Dennis were largest in eastern North Carolina and
southeastern Virginia where they exceeded 100 mm (Fig. 40a), and
in central North Carolina and central Virginia where they averaged 50 mm–100 mm. Two
weeks later, rainfall from Hurricane Floyd impacted the entire East Coast from South
Carolina to Maine (Fig. 40b), with totals exceeding 200 mm from
South Carolina to New York. The largest rainfall totals associated with Floyd occurred in
northeastern South Carolina and eastern North Carolina, where 5-day totals exceeded 400
mm. Locally, 24-hour amounts during this period approached 500 mm near the approximate
landfall site of Wilmington, North Carolina.
(iii) The Ohio Valley, Tennessee Valley, middle and lower
Mississippi Valley, and Texas
Rainfall was well below-normal from July–November 1999 throughout
the Ohio Valley, the middle and lower Mississippi Valley, and Texas (Fig.
41). Precipitation totals during this 5-month period were more than 100 mm below
normal throughout the region, with the largest deficits reaching 200–400 mm in
northeastern Texas, southern Arkansas, northern Louisiana, northern and western
Mississippi, the western Tennessee Valley, and parts of Missouri. On a statewide basis,
Arkansas experienced its driest July–November period in 105 years of record (Fig. 42), while Indiana and Missouri endured their second-driest
such period, and Illinois and Louisiana experienced their third driest. In addition,
totals were the fourth lowest since 1895 in Kentucky and Texas, and the fifth lowest in
Tennessee. For the region as a whole (for the area depicted in brown in the inset in Fig. 43), July–November 1999 ranks as the driest such period in
the past 105 years, with area-averaged precipitation deficits reaching 180 mm. This
deficit exceeds the previous record negative anomaly of 150 mm set in 1924, and contrasts
with the generally wetter-than-normal conditions that have been observed in this region
since 1970. By the end of 1999, long-term dryness remained a concern throughout the
region, with acute dryness still affecting central and northern Texas and the lower
Mississippi Valley.
This drought was associated with substantial agricultural and
hydrologic impacts. According to the USDA, Missouri and Ohio each reported 13–16%
decrease in corn and soybean production compared to 1998 levels. The wheat crop in
Oklahoma experienced a 24% decline in output compared to 1998 levels, and hay production
was 16–31% lower than 1998 levels in several states. Streamflows were also greatly
reduced during the last half of 1999, with many river levels below the 10th percentile in
areas east of the Mississippi River and south of the Great Lakes (USGS).
(iv) Hawaii
Large portions of the Hawaiian Islands experienced a second consecutive
year of below-normal rainfall during 1999, although totals were generally larger than
those observed in 1998. This dryness contrasts with above-average rains that the Islands
have typically experienced during past La Niña episodes (Ropelewski and Halpert 1989).
The most significant dryness during 1999 (25–50% of normal) occurred in southeastern
Oahu, central and southern Maui, most of Molokai and Lanai, and western sections of Hawaii
Island. Elsewhere, near- or above-normal rainfall totals were confined to much of Kauai,
west-central Oahu, the higher elevations of eastern Oahu, northwestern Maui, and eastern
Hawaii Island.
Rainfall totals from late 1997 through 1999 were exceptionally low at many locations,
including southern and western Oahu, central and southwestern Maui, Molokai (Fig. 44a), Lanai, and southern and western parts of the Big Island,
all of which received less than half of normal rainfall during 1998–99. In Honolulu,
total rainfall during 1998 and 1999 was 412 mm (Fig. 44b), which
is the lowest amount at this site for any two consecutive calendar years since reliable
records began in 1947. This total is also more than 100 mm below the previous record low
value of 524 mm set during 1952–53. Accumulated precipitation shortfalls during
1998–99 were also large across windward sections of Hawaii Island, where normals are
typically extremely large. For example, Laupahoeh, in northeastern Hawaii Island, recorded
4450 mm of rain during 1998 and 1999, which is more than 3000 mm below the long-term
average.
