The anomalous circulation patterns during an unusually prolonged stormy-weather period in Hawaii from 19 February to 2 April 2006 are analyzed and are compared with those of two previously known prolonged heavy-rainfall periods (March 1951 and February 1979). The circulation patterns for these three periods are characterized by 1) a negative Pacific–North American (PNA) pattern in the midlatitudes with a blocking high southwest of the Aleutian Islands, 2) retraction and splitting of the zonal jet into a polar jet north of 50°N and a persistent subtropical jet to the south over the central Pacific Ocean, 3) an anomalous low west of the Hawaiian Islands embedded in the subtropical jet, and 4) a weaker-than-normal Hadley circulation in the mid-Pacific. The moisture advected from low latitudes by the southerly wind component east of the persistent anomalous low, combined with upward motion, provides the large-scale setting for the unusually prolonged unsettled weather across the Hawaiian Islands. For all three cases, the prolonged stormy weather started after the onset of large-scale blocking and a negative PNA pattern over the North Pacific and the occurrence of a persistent anomalous low embedded in the subtropical jet west of the Hawaiian Islands. Furthermore, the persistent low was located at the optimal position to bring moisture from the central equatorial Pacific to Hawaii. The stormy weather ceased after the midlatitude blocking pattern weakened and the anomalous low in the subtropics decayed and/or shifted westward. There are no apparent common precursors in the 2-week period prior to the prolonged stormy weather among these three cases, however.
Unusually prolonged stormy weather from 19 February to 2 April 2006 had a serious impact on the infrastructure, economy, and livelihoods of those living in Hawaii. During this period, the National Weather Service Honolulu Forecast Office issued over 500 nonroutine products including 111 flash-flood warnings, 88 special marine warnings, 11 severe-thunderstorm warnings, 5 winter-weather advisories, 3 severe-thunderstorm watches, 2 winter-storm watches, 2 high-wind warnings, 1 winter-storm warning, and 1 tornado warning. In addition, the state of Hawaii experienced 1 damaging tornado, 3 verified severe thunderstorms, and 26 days with flash-flood warnings during this prolonged heavy-rainfall period. On average, 10 flash floods occur per year in Hawaii. These individually located heavy-rainfall events occur under the influence of one of four types of synoptic-scale disturbances: kona storm, cold front, upper-tropospheric trough, and tropical system (Blumenstock and Price 1967; Kodama and Barnes 1997; Wang et al. 1998); Tu and Chen (2011) studied a localized heavy-rainfall event over the south shore of Oahu associated with a kona storm near the end of this prolonged period of heavy rainfall.
The strength and zonal extent of the Asian jet can modulate the occurrence of kona storms. A zonally retracted jet provides a favorable situation for the equatorward propagation of upper-level cyclones leaving the exit region of the Asian jet (Otkin and Martin 2004a; Jaffe et al. 2011). Kona storms are more frequent during the period of a weaker, zonally retracted Asian jet stream because of the weakening of the midlatitude waveguide (Otkin and Martin 2004b; Businger and Caruso 2006; Chu et al. 1993).
In March of 2006, trade winds were only present for five days and winds were mostly from the southeast through southwest because of a persistent low pressure system west of the Hawaiian Islands. This unusual, prolonged period of heavy rainfall is characterized by a 6-week Rex blocking pattern (Rex 1950) in the North Pacific Ocean (Nash et al. 2006), which is atypical in this region. In general, a blocking pattern usually lasts only about 2 weeks (Rex 1950). The occurrence of blocking in the Bering Strait region may be sensitive to the phase of the El Niño–Southern Oscillation (ENSO). From the analysis of 90-day periods covering the months of December–February for the 44 winters 1950/51 through 1993/94 inclusive, Renwick and Wallace (1996) found that 69% more blocking days are observed over the North Pacific during La Niña winters than during El Niño winters. During 1979–2000, the average number of days of atmospheric blocking over the Alaskan region during La Niña, neutral, and El Niño winters is 27, 31.2, and 12 days, respectively (Carrera et al. 2004). More-intense blocking events may occur during a La Niña event (Barriopedro et al. 2006; Huang et al. 2002). The occurrence of blocking in the Bering Strait region is also sensitive to the Pacific–North American (PNA) pattern (Wallace and Gutzler 1981). The positive and negative polarities of the PNA pattern are related to dry and wet weather over the Hawaiian Islands, respectively (Chu et al. 1993). On the interannual time scale, Hawaii tends to be dry during most of the El Niño years (Chu 1995).
Another factor that might result in above-normal rainfall in Hawaii is related to the Madden–Julian oscillation (MJO; Madden and Julian 1994). Wang and Rui (1990) noted that for moderate events outgoing longwave radiation (OLR) anomalies originate in the equatorial African region, intensify while passing through the Indian Ocean, weaken near the “Maritime Continent,” reintensify in the western Pacific, and dissipate near the date line. For strong events, OLR anomalies do not dissipate near the date line but move toward North America and may affect the Hawaiian Islands (Wang and Rui 1990). The MJO may modulate the weather/climate in the subtropics and midlatitudes. Although the prominent convective activities of the MJO are limited to the equatorial tropics from the Indian Ocean to the tropical western Pacific, several studies have demonstrated the influence of the MJO on precipitation in the western United States (e.g., Mo and Higgins 1998; Bond and Vecchi 2003) and eastern Asia (Jeong et al. 2008).
During March of 1951, the Hawaiian Islands received some of the heaviest rainfall on record as a result of marked cyclonic activity associated with a persistent low-latitude trough to the west of Hawaii (Winston 1951). Another period of record torrential rainstorms over the Island of Hawaii occurred during January–February of 1979 (Cram and Tatum 1979). In comparing the 2006 case with these two historical, long-lasting anomalous winter-weather periods over the Hawaiian Islands, do these prolonged periods of heavy rainfall occur under similar anomalous large-scale circulation patterns? Are these unusually prolonged periods of heavy rainfall related to any anomalous circulation features from the equatorial tropics (e.g., La Niña/Walker circulations or MJO) and/or extratropics (PNA)? Are there any anomalous moisture distributions related to circulation-pattern changes that are favorable for the occurrence of the prolonged heavy-rainfall periods over the Hawaiian Islands? Are there any common precursors among these three cases?
2. Data and method
The monthly rainfall index (Chu and Chen 2005) for March of 2006 is well above the climatological average (1.77), and it is about normal in February of 2006 (~0). The period of unsettled weather during 2006 (19 February–2 April) is determined from the beginning and end dates of precipitation over the state of Hawaii (Fig. 1a) from the time series of the area-averaged 3-h Tropical Rainfall Measuring Mission (TRMM) 3B42 (0.25° × 0.25° × 3 h) (Huffman et al. 2007) rainfall data (Fig. 1b). All of the flash-flood warnings issued by the National Weather Service Honolulu Forecast Office were within this period. Before this period, it is very dry in February of 2006. The Special Sensor Microwave Imager (SSM/I) precipitable water (PW) data (0.25° × 0.25° × 1 day; http://www.remss.com/ssmi/), Quick Scatterometer (QuikSCAT) ocean surface winds (0.25° × 0.25° × 1 day; http://www.ssmi.com/qscat/), Reynolds sea surface temperature (SST) data (0.25° × 0.25° × 1 day; Reynolds et al. 2007), and National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis data (2.5° × 2.5° × 1 day; Kalnay et al. 1996) are analyzed over the Pacific Ocean to study the anomalous circulation patterns during this period.
During March of 1951, many stations in Hawaii reported rainfall totals ranging from 200% to 700% of normal for the month as a result of the marked cyclonic activity associated with a deep low-latitude trough to the west of Hawaii (Winston 1951). An examination of the monthly rainfall index over Hawaii (Chu and Chen 2005) during the months of January, February, and March since 1950 shows that February of 1979 and March of 1951 are the only two months with a monthly rainfall index well above 2. For the March 1951 case (18 February–30 March), the beginning and ending dates were determined from the period during which the area-averaged precipitable water from the NCEP–NCAR reanalysis was elevated with at least two days of sustained area-averaged southerly winds over the Hawaiian Islands. Under these conditions, the normal northeast trade winds are interrupted. The monthly rainfall index for February of 1951 is 0.66, indicating above-normal monthly rainfall. The beginning and ending dates for the February 1979 (4–22 February) case were also determined on the basis of the above criteria. In addition, rainfall analysis from the Climate Prediction Center Merged Analysis of Precipitation (CMAP) data (2.5° × 2.5° × 5 day) (Xie and Arkin 1997) over the state of Hawaii was also examined. For the 1979 case, there is a major heavy-rainfall period, for example, 10–17 January, prior to 4–22 February with a secondary storm period during 21–24 January (Cram and Tatum 1979). The monthly rainfall index for January of 1979 is only 0.2, however, because most of the heavy-rainfall occurrences for these disturbed periods are on the windward side of the island of Hawaii. It is relatively dry before 9 January. Furthermore, there is a break of more than 10 days between the secondary storm period during 21–24 January and the beginning of the next storm period starting 4 February. Thus, the persistent unsettled weather period for the 1979 case does not include these two storms in January.
Means and anomalies of horizontal wind, geopotential height, divergence, vorticity, and vertical velocity at the 850-, 500-, and 250-hPa levels are constructed for the unsettled weather periods to identify the large-scale circulation patterns and compare them with the long-term mean. Climatological means from the NCEP–NCAR reanalysis data from 1971 to 2000 are used to represent the long-term mean. Vertical cross sections of mean and anomalies of zonal wind, meridional wind, geopotential height, vorticity, temperature, moisture, and equivalent potential temperature are also constructed using data from 159° to 157°W, between Kauai and Maui (Fig. 1a), for the unsettled weather periods. Analyses of divergence and vertical motion are crucial for the identification of atmospheric circulation cells (Huang et al. 2004). Atmospheric cells, such as the Hadley and Walker cells, are examined from vertical cross sections. The mean and transient meridional moisture fluxes over the Hawaiian Islands during this wet-weather period are computed to study moisture transport from the equatorial tropics to higher latitudes. The mean transient moisture flux is given by
where is the mean moisture flux, is the mean specific humidity, and is the mean meridional wind component.
3. Circulation patterns during the unusually prolonged period of heavy rainfall in 2006
a. Blocking and anomalous circulation patterns
Anomalies of geopotential height at the 500-hPa level (Fig. 2a) for the wet-weather period show a North Pacific blocking pattern with geopotential height anomalies of >100 m between 160°E and 140°W and between 35° and 55°N. A blocking pattern is identified when the 500-hPa-level geopotential height anomalies exceed 100 m for at least 8 consecutive days (Carrera et al. 2004). A negative PNA pattern (Wallace and Gutzler 1981) is apparent with positive height anomalies near 45°N, 175°W, southwest of the Aleutian Islands, as well as over the southeastern United States. Negative height anomalies are present near 20°N, 178°W, west of the Hawaiian Islands, and near 40°N, 130°W over the northwestern coast of North America (Fig. 2a).
The blocking pattern often represents the breakdown of the normal tropospheric zonal flow at midlatitudes with the splitting of the zonal jet (Huang et al. 2004). The disturbances are forced to move in anomalous storm tracks to the north and south of the anticyclonic blocking (Nakamura and Wallace 1990; Carrera et al. 2004; Huang et al. 2004). The mean horizontal wind for the period from 19 February to 2 April 2006 (Fig. 2b) shows a weakening of the upper-level jet and splitting of the jet into a polar jet and a subtropical jet near 170°E. Westward retraction of the zonal jet toward eastern Asia is clearly evident. This westward retraction and splitting of the polar jet are often associated with a negative PNA pattern (Otkin and Martin 2004a).
During 19 February–2 April 2006 there were frequent transient midlatitude synoptic disturbances moving eastward along the polar jet to the north. Several previous studies suggest that the maintenance of blocking events involves nonlinear interactions between large-scale, low-frequency waves and high-frequency baroclinic waves (Mullen 1987; Dole 1989; Nakamura and Wallace 1990). Mullen (1987) suggests that the main mechanism appears to be barotropic given that the anticyclone is maintained by poleward advection of anticyclonic vorticity by high-frequency disturbances. Jaffe et al. (2011) composited 19 jet retraction periods over the North Pacific during 28 (1979–2007) boreal winters (November–March) using NCEP–NCAR reanalysis data and found that the composite structures of upper- and lower-level anomalies are equivalent barotropic.
The subtropical jet lies over Hawaii with a persistent trough to the west of the Hawaiian Islands and anomalous southwesterlies over the islands (Fig. 2c). There are transient disturbances embedded in the subtropical jet. During this period, four kona storms, three upper-level short waves, and two cutoff midlevel lows occurred. At the surface, the anomalous southeasterly flow over the Hawaiian region, associated with an inverted trough around 170°W, is apparent (Fig. 3a).
The weakening of the intertropical convergence zone (ITCZ) in the central equatorial Pacific, southwesterly winds east of the semipermanent low in the subtropics, and southerly winds upstream of the persistent blocking in midlatitudes resulted in southerly flow from the tropics to midlatitudes over the central Pacific (from near the date line to 150°W). Above-normal values of precipitable water over the Hawaiian Islands, the Maritime Continent, and the South Pacific convergence zone (SPCZ) and below-normal values over the central and eastern Pacific are present from mid-February through March 2006 (Fig. 3b). Over the Hawaiian Islands (between 159° and 156°W) (Fig. 4a), the PW is elevated, with a moist tongue extending from the lower latitudes to the subtropics. Between the date line and 150°W, the presence of the southerly moisture flux, which extends from the equatorial areas to the Hawaiian Islands at the 850-hPa level (Fig. 4a), suggests that the abnormal low-level southerly wind component plays an important role in bringing the moisture sources from the tropics over the Hawaiian Islands (Fig. 4a).
The MJO is relatively weak from January to April of 2006 (see online at http://www.cpc.ncep.noaa.gov/products/precip/CWlink/MJO/ARCHIVE/). Enhanced convection over Indonesia, northern Australia, and the surrounding ocean and along the SPCZ is apparent from early January to early April (not shown), consistent with higher-than-normal PW in these areas (Fig. 3b). There is no evidence of convection anomalies moving eastward or northeastward beyond the Maritime Continent that are associated with the MJO.
b. La Niña conditions
The oceanic Niño index (ONI) is defined as a 3-month running-mean value of SST departures from the average in the Niño-3.4 region. From the ONI, the National Oceanic and Atmospheric Administration (NOAA) defines El Niño as a positive ONI of greater than or equal to 0.5°C and defines La Niña as a negative ONI of less than or equal to −0.5°C. An El Niño episode is classified when these conditions are satisfied for a period of at least 5 consecutive months. Short periods during which these criteria last fewer than 5 consecutive months are defined as having “El Niño and La Niña conditions” (Kousky and Higgins 2007).
For this unsettled wet period over Hawaii, typical La Niña features, such as cold SST anomalies (−1.5°C) in the central and eastern equatorial Pacific and warm SST anomalies (>0.5°C) over the Maritime Continent, are present (Fig. 3c). The ONI of November–January (NDJ), December–February (DJF), and January–March (JFM) in 2005/06 are −0.7, −0.8, and −0.7, respectively. During this period, ONI exceeds the −0.5°C threshold for La Niña conditions for 3 consecutive overlapping months. Positive values of the Southern Oscillation index (SOI) are often associated with La Niña conditions. The SOI of January, February, March, and April in 2006 are 12.7, 0.1, 13.8, and 15.2, respectively. Hence, La Niña conditions are present during this period.
In all other La Niña years, or during strong La Niña events, extreme heavy rainfall may not always occur over Hawaii, even though rainfall over Hawaii and La Niña are positively correlated (Chu and Chen 2005). In fact, the March 2006 record-breaking event is not the strongest La Niña event on record. Furthermore, La Niña conditions exist prior to and after this prolonged stormy event. This extremely prolonged heavy-rainfall period is related to anomalous circulations in high latitudes as well as in the equatorial region, which gave rise to copious moisture with an anomalous rising motion in the subtropics. Note that around 20°N the moist tongue from low latitudes has an east–west extent of about 20° with the maximum axis over the Hawaiian Islands (Fig. 3b). If this moist tongue were displaced 10° eastward or westward, then the Hawaiian Islands would be relatively dry. It is apparent that the persistent low embedded in the subtropical jet (Fig. 2a) is located at the optimal position to bring moisture from the central equatorial Pacific to the Hawaiian Islands. In section 3c, the anomalous weather patterns will be further investigated using vertical cross sections.
c. Anomalous circulations as revealed from vertical cross sections
For a better understanding of anomalous circulation patterns, vertical cross sections are constructed between 159° and 157°W, which is close to the axis of the moist tongue (Fig. 3b) for the period 19 February to 2 April. The weakening and splitting of the upper-level jet into a polar jet to the north (~47°N) and a subtropical jet to the south (~18°N) are evident with maximum intensities near the 250- and 200-hPa levels, respectively (Fig. 5a). Climatological data show that the position of the midlatitude jet is between 30° and 40°N at this time of the year (Fig. 5b). The large vertical zonal wind shear associated with the subtropical jet between 15° and 30°N is evident, especially above the 700-hPa level. A relatively large southerly wind component from 10° to 30°N in the upper levels is associated with the trough embedded in the subtropical jet and weak northerly winds from 40° to 60°N ahead of the ridge axis in the polar jet associated with a persistent blocking pattern (Fig. 5c). The normal mean meridional wind pattern in the North Pacific (Fig. 5d) is reversed. With the southerly wind component at low levels in the subtropics, moisture is advected from the tropics to midlatitudes (Fig. 4a). An anomalous sinking motion near the equator and anomalous rising motion in the subtropics (20°–30°N) (Fig. 5e) indicates the weakening and reversing of the local Hadley circulation between 159° and 157°W. Positive vorticity anomalies with maximum magnitude between the 400- and 1000-hPa levels near the Hawaiian Islands (20°–30°N) are associated with the persistent trough in the subtropics, whereas negative vorticity anomalies between 40° and 50°N are associated with anticyclonic blocking in midlatitudes (Fig. 5f).
The vertical cross section of geopotential height anomalies (Fig. 6a) shows higher-than-normal heights (>100 m) from the upper to lower troposphere between 40° and 50°N corresponding to the midlatitude blocking pattern, with maximum anomalies (>200 m) around the 300-hPa level where temperature anomalies change from positive below to negative above (Fig. 6b) (Bals-Elsholz et al. 2001; Crum and Stevens 1988). A warm-core structure over the blocking region (35°–60°N) (Fig. 6b) between the 850- and the 300-hPa levels is consistent with the anomalous sinking motion present there (Fig. 5e). The maximum temperature anomalies (>4 K) occur near the 400-hPa level. In the subtropics, negative geopotential height anomalies are present with a maximum of −30 m aloft (Fig. 6a). Small negative temperature anomalies (~−1 K), associated with the subtropical trough, are evident except below the 750-hPa level (Fig. 6b). The moisture distribution exhibits negative specific humidity anomalies (~1.5 g kg−1) (Fig. 6c) over the equatorial atmosphere and in the low levels of the midlatitudes. Positive values (>2.5 g kg−1) are apparent between 15° and 30°N. Positive equivalent potential temperature anomalies (Fig. 6d) are evident between 20° and 30°N, with maximum values near the 850-hPa level where moisture anomalies are the largest (Fig. 6c). Negative equivalent potential temperature anomalies are apparent throughout the atmosphere near the equator. In the midlatitudes, negative equivalent potential temperature anomalies are apparent at low levels up to the 850-hPa level and positive anomalies are evident in the middle and upper levels, with maximum values near the 400-hPa level where the temperature anomalies are the highest (Fig. 6b).
In summary, during this period, with persistent blocking and a negative PNA pattern in midlatitudes, the normal tropospheric zonal flow at midlatitudes breaks down, with westward retraction of the midlatitude jet toward eastern Asia and splitting of the jet into a polar jet north of 50°N and a persistent subtropical jet over the Hawaiian Islands. At the same time, a persistent trough is embedded in the subtropical jet west of the Hawaiian Islands near the date line. The southerly wind component associated with the persistent low-latitude trough embedded in the subtropical jet brings warm, moist air with high equivalent potential temperature values from the equatorial tropics to the Hawaiian Islands. The rising motion associated with the persistent subtropical trough occurs under a much-weaker-than-normal Hadley cell in the central Pacific under La Niña conditions. Copious moisture together with persistent rising motion provides favorable settings for prolonged stormy weather with frequent heavy rainfall over the Hawaiian Islands.
4. Evolution of large-scale circulation patterns for the prolonged period of heavy rainfall in 2006
The temporal evolution of the circulation patterns for this extremely heavy rainfall period is useful to study the precursors, if any, of the development of the anomalous circulations and flow patterns after this event. To assist in this endeavor, we examined conditions 2 weeks before the onset and 2 weeks immediately following the termination of the unsettled period.
Geopotential height anomalies (contour interval every 30 m) at the 500-hPa level for 2 weeks before the wet weather period (4–18 February), during the wet period (19 February–2 April), and 2 weeks after the wet period (3–17 April) were computed. During the 2-week period before the event, a persistent anomalous low with negative 500-hPa height anomalies is present to the southeast of the Hawaiian Islands (Fig. 7a). The persistent low moves westward and lies to the west of the Hawaiian Islands during the wet period (Fig. 2c), and then it moves farther northwestward after the period (Fig. 7b). The North Pacific blocking pattern to the south of the Aleutian Islands exists during the period (Fig. 2c). Positive height anomalies appear to the northwest of the Aleutian Islands before the period (Fig. 7a). The blocking pattern weakens afterward (Fig. 7b). Notable changes in circulation patterns occur concurrently in low latitudes, as the negative height anomalies west of the Hawaiian Islands move westward. During the 2-week period before the event a strong polar jet is evident in midlatitudes (Fig. 8a). A weakening of the polar jet, similar to that found in Jaffe et al. (2011) and Otkin and Martin (2004a), is apparent during the period and in the 2-week period afterward (Figs. 2b and 8b).
In the subtropics, during the 2-week period before the onset of the prolonged period of heavy rainfall, an anomalous low is present to the southeast of the Hawaiian Islands, bringing anomalous northerly winds to them (Fig. 8c). The persistent low moves westward (Fig. 9) and lies to the west of the Hawaiian Islands during the period, bringing southerly winds over the islands (Fig. 2c), and then it moves farther northwestward after the period (Fig. 8d). With anomalous northerly winds before the period, PW values are below normal over the Hawaiian Islands (Fig. 10a) but become above normal during the period (Fig. 3b) because of the presence of the abnormal southerly winds east of the persistent low. The PW anomalies are below normal over the Hawaiian Islands after the period as the persistent low moves northwestward (Fig. 10b).
During the 2-week period before the onset, the weakening of the ITCZ over the central and eastern Pacific is already evident, with negative PW anomalies over the central and eastern equatorial Pacific (Fig. 10a). These conditions continue during the prolonged period of heavy rainfall and the 2-week period afterward (Figs. 3c and 10b).
5. A comparison with the previous prolonged heavy-rainfall periods
a. The 1951 prolonged heavy-rainfall period (18 February–30 March)
The NCEP–NCAR reanalysis is less reliable for February–March 1951 than for February–March 2006 because data sources over the Pacific Ocean were fewer in 1951 before the existence of satellite data. The geopotential height anomalies at the 500-hPa level (Fig. 11a) for the wet-weather period in 1951 show conditions that are similar to those of the wet-weather period in 2006, such as midlatitude blocking and a negative PNA pattern (Fig. 2a). The PNA index in March of 1951 is −0.81; the blocking pattern during the wet-weather period in 1951 is not as significant as that of 2006, however, with smaller height anomalies (100 m vs >180 m). The negative height anomalies in the subtropics are also apparent northwest of the Hawaiian Islands with a minimum at 25°N, 170°W (Fig. 11a). Similar to 2006, the weakening and westward retraction of the upper-level jet and splitting into two branches are also apparent in 1951 (Fig. 11b) with the subtropical jet over the Hawaiian region and the polar jet around 50°N. An anomalous cyclonic circulation and upper-level trough embedded in the subtropical jet northwest of the Hawaiian Islands are also present (Fig. 11c).
The vertical cross section of the mean zonal wind between 159° and 157°W during the 1951 wet-weather period (Fig. 12a) shows that the maximum intensities of the polar jet and the subtropical jet are around 50°N near the 300-hPa level and around 20°N at the 200-hPa level, respectively. The subtropical jet is much stronger than the polar jet, with large vertical wind shear over the Hawaiian Islands. Similar to 2006, the anomalous southerly flow is evident over the Hawaiian Islands in 1951 (Fig. 12b) and extends southward to 10°–15°N. Similar to what was observed in 2006, a strong rising motion between 20° and 30°N in the subtropics, with maximum values between the 400- and 500-hPa levels, and sinking motion between 50° and 60°N are apparent in 1951 (Fig. 12c). Pressure, vertical velocity, and divergence (not shown) together represent the weakening of the local Hadley circulation over the central equatorial tropics in 1951 (Fig. 12c); unlike in 2006, however, the reversal of the local Hadley circulation does not occur.
Between 40° and 60°N, positive geopotential height anomalies (>100 m) along 159°–157°W for 1951 (Fig. 13a), corresponding to the midlatitude blocking pattern, are evident. The maximum values of positive height anomalies are around the 300-hPa level. The intensity of the blocking is weaker in 1951 than in 2006 (Fig. 6a). Similar to the pattern of March of 2006, negative height anomalies (<−100 m) in the upper levels, associated with a persistent upper-level low near the Hawaiian Islands, are apparent. Note that the upper-level low is deeper in 1951 than in 2006.
The vertical cross section of mean temperature anomalies between 159° and 157°W for 1951 (Fig. 13b) shows similar conditions to those of 2006—namely, negative anomalies above the 250-hPa level and positive anomalies below the 250-hPa level between 40° and 60°N, consistent with the presence of a blocking anticyclone in this region (Bals-Elsholz et al. 2001). Between 40° and 50°N, the positive temperature anomalies are the highest (>4°–5°C) near the 500-hPa level. Similar to what was observed in 2006, maximum values of geopotential height anomalies occur near the 300-hPa level (Fig. 13a) where temperature anomalies change from positive below to negative above.
Similar to the pattern of 2006, positive anomalies of specific humidity are evident between 15° and 30°N with negative anomalies in the upper levels near the equator (Fig. 13c). Positive equivalent potential temperature anomalies (Fig. 13d) are evident between 20° and 30°N with the maximum values near the 850-hPa level, where positive moisture anomalies are present (Fig. 13c) but are smaller than in 2006.
The period from the winter of 1950 to the spring of 1951 is considered to be a La Niña episode. The ONI of September–November (SON), October–December (OND), NDJ, DJF, JFM, and February–April (FMA) are −0.8, −0.9, −1.0, −1.0, −0.8, and −0.6, respectively. Typical La Niña features such as above-normal values of precipitable water over the Hawaiian Islands, the Maritime Continent, and the SPCZ are evident in 1951 (Fig. 11d).
b. The February 1979 prolonged heavy-rainfall period (4–22 February)
A series of slow-moving troughs and shear lines passed over the Hawaiian Islands during January and February of 1979, producing excessive rainfall over the Hawaiian island chain (Cram and Tatum 1979). The island of Hawaii was the area that was hardest hit. A North Pacific blocking pattern with positive height anomalies near 43°N, 178°E and negative height anomalies west of Hawaii is evident during the wet period of 4–22 February (Fig. 14a). The PNA index for February of 1979 is −1.82. The weakening and westward retraction of the upper-level jet and the splitting of the upper-level jet into two branches are also evident (Fig. 14b). Similar to the 2006 event (Fig. 2b), a subtropical jet is over the Hawaiian region. In the upper levels, an east–west-oriented trough occurs north of the Hawaiian Islands and results in anomalous southwesterly winds over the Hawaiian Islands (Fig. 14c). With southerly winds at low levels, relatively high values of PW over the Hawaiian Islands are present (Fig. 14d).
The vertical cross section of the mean zonal wind between 159° and 157°W (Fig. 15a) shows that the maximum intensity of the polar jet (>25 m s−1) and the subtropical jet (>35 m s−1) are around 43°N near the 250-hPa level and around 20°N at the 200-hPa level, respectively. The southerly flow is evident between 10° and 30°N except near the surface (Fig. 15b). The rising motion is present between the equator and 10°N and between 20° and 30°N in the middle and upper troposphere, with the sinking motion occurring between 35° and 45°N and between 60° and 70°N (Fig. 15c). Weakening of the local Hadley circulation between 159° and 157°W is evident, with an anomalous rising motion between 20° and 30°N. Thus, the circulation patterns for this event are similar to those in 2006.
The midlatitude blocking pattern is evident, with above-normal geopotential height anomalies (>100 m) between 40° and 50°N in the 200–700-hPa layer and with maximum anomalies around the 300-hPa level (Fig. 16a). The blocking pattern is similar to the 1951 event (Fig. 13a) but weaker than the 2006 event (Fig. 6a). Negative height anomalies are evident in the 100–300-hPa layer near the Hawaiian Islands between 15° and 25°N during this period. Similar to the 2006 situation, negative temperature anomalies are present between 35° and 45°N above the 250-hPa level with positive anomalies below, consistent with the presence of the blocking pattern in this region (Bals-Elsholz et al. 2001) (Fig. 16b). In the subtropics, positive anomalies of specific humidity (Fig. 16c) and equivalent potential temperature (Fig. 16d) are evident, with maximum values near the 850-hPa level.
The ONI for SON, OND, NDJ, DJF, JFM, and FMA are 0.4, −0.2, −0.1, −0.1, and 0.0, respectively, during the 1978/79 winter, which is considered to be an ENSO-neutral year. Note that for both the 2006 and 1951 events a negative ONI index is present not only during the wet period but also before and after.
The common ingredients for all three late-winter–early-spring heavy-rainfall episodes presented in this study are persistent North Pacific blocking with a negative PNA pattern, a weakening and westward retraction of the polar jet over the western Pacific, splitting of the midlatitude jet into a polar jet and a subtropical jet, and a weaker-than-normal (or reversal of) local Hadley circulation over the central Pacific. Also, a persistent low embedded in the subtropical jet west of Hawaii with a southerly wind component over the Hawaiian Islands provides favorable background settings—for example, a moist tongue and rising motion—for the development of frequent heavy precipitation over the island chain. Transient disturbances propagating eastward along the subtropical jet will be enhanced as they move toward the Hawaiian island chain (Tu and Chen 2011). A schematic diagram showing the anomalous circulation patterns associated with this type of event is given in Fig. 17.
c. Evolution of the 1951 and 1979 cases
Even though no blocking pattern is evident over the North Pacific 2 weeks prior to the 1951 prolonged rainfall period, positive height anomalies are present to the south of the Aleutian Islands (Fig. 18a). Similar to what was observed in 2006, the North Pacific blocking pattern to the south of the Aleutian Islands and the persistent low to the west of the Hawaiian Islands exist only during the heavy-rainfall period (Fig. 11a). The blocking pattern dissipates after the period (Fig. 18b), concurrent with the disappearance of the persistent low in the subtropics. For the 2-week period before the onset, a strong polar jet is evident in midlatitudes (Fig. 19a) with stronger-than-normal westerly winds (Fig. 19c). The weakening and splitting of the polar jet is apparent during the period (Fig. 11b) with a persistent low west of Hawaii, bringing south/southwesterly winds over the Hawaiian Islands (Fig. 11c). For the 2-week period after the ending of the 1951 case, the polar jet slightly strengthens as the blocking pattern dissipates (Fig. 19b). In the meantime, the south/southwesterly winds over the Hawaiian Islands disappear (Fig. 19d).
In contrast to the 2006 and 1951 events, a blocking pattern to the southeast of the Aleutian Islands (around 40°N, 160°W) is apparent during the 2-week period prior to the onset of the 1979 case (Fig. 20a). A pronounced low center was located over the western United States, with the trough axis extending southwestward to the island of Hawaii. During the 1979 persistent stormy-weather period (4–22 February), the blocking pattern moves westward and lies to the southwest of the Aleutian Islands (around 40°N, 180°), with a persistent low to the west of the Hawaiian Islands (Fig. 14a). The blocking pattern dissipates after the ending of the 1979 case (Fig. 20b). During the 2-week period prior to the onset, a weaker-than-normal polar jet and splitting of the polar jet near 170°W are evident (Fig. 21a). An abnormal anticyclone, associated with midlatitude blocking, and a southwest–northeast-oriented persistent, stationary trough to the northeast of the Hawaiian Islands are both apparent (Fig. 21c). For the 1979 case, the weakening and splitting of the polar jet continue, with the splitting position moving westward to 180° (Fig. 14b) as the blocking pattern moves westward. The persistent low to the west of the Hawaiian Islands brings south/southwesterly winds over the islands (Fig. 14c). After the ending of the 1979 case, the polar jet strengthens (Fig. 21b) as the blocking pattern dissipates concurrent with the replacement of the southwesterly winds over the Hawaiian Islands by stronger-than-normal northeasterly winds (Fig. 21d).
For all three cases, there are no common precursors during the 2-week period prior to the onset. These results are in agreement with the study of the life cycles of persistent blocking over the North Pacific by Dole (1986, 1989). Dole showed that the blocking pattern over the North Pacific often occurred rapidly (e.g., within 1 week), with little indication of significant anomalies over the key region until immediately prior to the onset. All three events end when midlatitude blocking and the negative PNA pattern dissipate (or weaken) without an abnormal low system in the subtropics west of Hawaii.
6. Summary and discussion
The SSM/I, QuikSCAT, and NCEP–NCAR reanalysis data are analyzed to study the anomalous circulation patterns associated with an unusually prolonged stormy-weather event in Hawaii from 19 February to 2 April 2006. The circulation patterns during this period are compared with those of two previously known prolonged periods of heavy rainfall (March 1951 and February 1979) that occurred at approximately the same time of the year. The circulations for all three prolonged periods of heavy rainfall are characterized by 1) a negative PNA pattern in the midlatitudes with a blocking high southwest of the Aleutian Islands, 2) retraction and splitting of the zonal jet into a polar jet north of 50°N along with a persistent subtropical jet to the south over the central Pacific, 3) an abnormal, persistent low west of the Hawaiian Islands embedded in the subtropical jet, and 4) a weaker-than-normal Hadley circulation in the mid-Pacific. For both the 1951 and 2006 cases, La Niña features such as a strengthening of the Walker circulation with higher-than-normal PW values over the Maritime Continent and the SPCZ, lower-than-normal PW values over the central and eastern Pacific, and a relatively weak ITCZ in the eastern Pacific are present. Nevertheless, for this type of prolonged period of heavy rainfall, La Niña features are not the only necessary and sufficient conditions to account for the prolonged stormy weather over the Hawaiian Islands. For both the 2006 and 1951 cases La Niña features exist prior to the onset of the wet period, whereas for the 1979 case it is a neutral year.
The presence of midlatitude blocking and a negative PNA pattern, with westward retracting and splitting of the zonal jet into a polar jet and a subtropical jet in the central Pacific, combined with a persistent low in the subtropics west of the Hawaiian Islands, and a weaker-than-normal ITCZ in the tropics are the key features for the occurrence of this type of event. Over the Hawaiian Islands, the flow is characterized by anomalous rising motion and positive vorticity associated with a semipermanent, abnormal subtropical low near the date line under weakening (or reversal) of the local Hadley cell in the central equatorial Pacific. The abnormal, persistent southerly winds east of this semipermanent low in the subtropics bring in warm, moist air with high equivalent potential temperature from the equatorial tropics to the subtropics. High amounts of low-level and midlevel moisture together with the rising motion provide a favorable environmental setting for the occurrence of excessive rainfall over the Hawaiian Islands. For all three cases (1951, 1979, and 2006), there are no common precursors. The blocking pattern over the North Pacific often occurs rapidly with little indication of significant anomalies until immediately prior to the onset. The ending of the wet period occurs as the blocking pattern dissipates (or weakens) together with the disappearance (or westward shifting) of the persistent low in the subtropics west of the Hawaiian Islands.
Author I. M. S. P. Jayawardena’s graduate studies at the University of Hawaii were funded by an East–West Center Fellowship and also by the Association of American University Women Fellowship (AAUW). This work was funded by COMET/UCAR under Grant S07-66828. The authors thank reviewers for their helpful comments, Ms. Y. Chen for providing us the Hawaii rainfall index, May Izumi and David Hitzl for editing the text, and the Joint Institute of Marine and Atmospheric Research (JIMAR)/NOAA for funding the publication costs under Cooperative Agreement NA17RJ1230/NA09OAR4320075. Figure 9 is provided by the Physical Sciences Division, NOAA/ESRL, Boulder, Colorado (obtained online at http://www.esrl.noaa.gov/psd/).
Current affiliation: Department of Meteorology—Sri Lanka, Colombo, Sri Lanka.
Current affiliation: National Weather Service, Burlington, Vermont.