1. Introduction
The National Weather Service (NWS) Forecast Office in Honolulu recorded an unusual wet period in Hawaii from 19 February to 2 April 2006. Climatologically, there are only about two or three flash flood events during this time of the year (Nash et al. 2010). However, during the 2006 wet period, 111 flash flood warnings were issued. On 31 March 2006, a Kona low northwest of Hawaii brought in heavy rainstorms over southeastern Oahu causing Makiki Stream and Manoa Stream to overflow. Floods occurred in several areas in the Manoa Valley and at the Kahala Mall (Fig. 1) on the southeastern shore of Oahu around 1100 Hawaiian standard time (HST, 2100 UTC). The Kahala area on the lee side of the Ko’olau Mountains is normally under easterly–northeasterly trades and does not receive much rainfall. The flooding on the ground floor of the Kahala Mall forced the mall to shut down for several days for the first time (Nash et al. 2010). One of the unique aspects of this case is that the heavy rainfall occurred on the lee side of low-level easterly winds in contrast to the windward heavy rainfall cases that are associated with convective cells anchored by the mountains or caused by persistent orographic lifting of the moist upslope-wind component (Schroeder 1977; Kodama and Barnes 1997; Zhang et al. 2005). As will be shown later, for this particular case, heavy rainfall occurred because of the orographic enhancement of preexisting storms that drifted inland.
The horizontal distribution of accumulated rainfall from 1100 to 1300 HST 31 Mar 2006 (from 2100 to 2300 UTC 31 Mar) over Oahu, HI. The surface rain gauge observations include those from NWS Hydronet (closed triangles) and other (NCDC, ASOS, RAWS, USGS; open triangles) stations. Rainfall isopleths are plotted for 10, 20, 40, 60, and 80 mm. Terrain height (m), from USGS, is shaded. Niu Valley (located at N) and Palolo FS (located at P) are two rain gauge stations close to the Kahala Mall [located at the filled square (▪)].
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The horizontal distribution of accumulated rainfall from 1100 to 1300 HST 31 Mar 2006 (from 2100 to 2300 UTC 31 Mar) over Oahu, HI. The surface rain gauge observations include those from NWS Hydronet (closed triangles) and other (NCDC, ASOS, RAWS, USGS; open triangles) stations. Rainfall isopleths are plotted for 10, 20, 40, 60, and 80 mm. Terrain height (m), from USGS, is shaded. Niu Valley (located at N) and Palolo FS (located at P) are two rain gauge stations close to the Kahala Mall [located at the filled square (▪)].
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The horizontal distribution of accumulated rainfall from 1100 to 1300 HST 31 Mar 2006 (from 2100 to 2300 UTC 31 Mar) over Oahu, HI. The surface rain gauge observations include those from NWS Hydronet (closed triangles) and other (NCDC, ASOS, RAWS, USGS; open triangles) stations. Rainfall isopleths are plotted for 10, 20, 40, 60, and 80 mm. Terrain height (m), from USGS, is shaded. Niu Valley (located at N) and Palolo FS (located at P) are two rain gauge stations close to the Kahala Mall [located at the filled square (▪)].
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
In this study, favorable large-scale conditions for the development of the Kahala Mall flood will be analyzed. The effects of orography on the precipitation process and rainfall distribution will also be examined. The high-resolution numerical model is one of the important forecasting tools at this time. We will examine the performance of the Weather Research and Forecasting Model (WRF) and its strengths and limitations in regard to this leeward-side heavy rainfall (cold rain) case using model-derived accumulated rainfall and radar reflectivities. The model-derived horizontal distributions of stability indices are also computed to provide useful forecasting guidance for this case. Impacts of latent heat release for the development of the Kahala Mall flood will be analyzed through model sensitivity tests.
Chu et al. (1993) studied large-scale flow patterns during January–February 1982 when the Hawaiian Islands experienced prolonged wet conditions. They found that a negative Pacific–North American (PNA) teleconnection pattern associated with a westward retraction of the Asian jet was accompanied by an anomalous strong cyclonic circulation over Hawaii. During a blocking pattern in the central Pacific, the Asian jet splits into subtropical and polar jets (Rex 1950). Rossby waves–transient disturbances propagate north and south of the blocking pattern (Marques and Rao 1999; Otkin and Martin 2004; Zhu et al. 2007) along the jet streams (Hoskins and Ambrizzi 1993). Jayawardena (2007) analyzed the climatological large-scale conditions for the development of the unusually persistent Hawaiian wet period in 2006. From her analysis, the negative PNA pattern was accompanied by an atmospheric blocking pattern in the central Pacific with a relatively weak Asian jet split into subtropical and polar jets. In low latitudes, a weak semipermanent trough in the subtropics, northwest of Hawaii, was accompanied by a weaker than normal ITCZ in the equatorial central Pacific with anomalous rising motion over the Hawaiian Islands and an anomalous moist tongue rooted in the tropics extending northward to the extratropics.
The Kona storm (subtropical cyclone) is one of the four synoptic disturbances–transient disturbances that frequently bring in heavy rainfall occurrences over the Hawaiian Islands. The other three are upper-level trough, midlatitude cold front, and tropical disturbance (Blumenstock and Price 1967; Schroeder 1977; Kodama and Barnes 1997). Kona storms are cold-core lows situated aloft with dominant circulations at the 250-hPa level that gradually diminish downward. Simpson (1952) used limited surface and rawinsonde sounding data to show that the maximum winds and rainfall associated with Kona storms occur in the eastern or southeastern quadrant. He also found that the encircling southerly flow in the eastern quadrant of a Kona storm brings moist tropical air toward the cyclone center.
Morrison and Businger (2001) studied the evolution of a Kona storm affecting the eastern end of the Hawaiian Island chain from 24 to 28 February 1997. They suggested that upper-level positive vorticity advection (PVA) associated with the Kona low contributed to the development of deep convective cells in the eastern quadrant of the Kona low. Martin and Marsili (2002), Martin and Otkin (2004), and Caruso and Businger (2006) showed that upper-level baroclinic forcing associated with subtropical cyclones enhances the upward motion, convective activity, and surface cyclogenesis over the positive potential vorticity advection region.
Water vapor is a crucial element for the development of heavy rainfall events. For a cold front system in the northeast Pacific, Ralph et al. (2004) and Bao et al. (2006) found a high Special Sensor Microwave Imager (SSM/I) vertically integrated water vapor (IWV) band (>2 cm) rooted in the tropics; the high content of the atmospheric moisture was transported from the tropics to the west coast of North America within a narrow band. For a quasi-stationary upper-level cutoff low (COL) over the eastern Pacific to the west of southern California, Knippertz and Martin (2007) showed low-level poleward water vapor fluxes (WVFs) from the tropics to the southwestern United States. The abundant moisture from the tropics contributed to heavy rainfalls over southern California, central Arizona, southern Nevada, and northwestern New Mexico (Knippertz and Martin 2007). Note that, WVF is defined as qv where q is the specific humidify and v is the horizontal wind vector at a given pressure level. Jiang et al. (2008a,b) used the water vapor budget for a volume of air to evaluate the rainfall in tropical cyclones (TCs). They showed that horizontal moisture convergence and ocean moisture flux are the dominant sources for precipitation as expected from the water budget equation. They found that the total precipitable water (TPW) is closely related to the sum of the horizontal moisture convergence and ocean moisture flux. They suggested that over the open ocean, where conventional observations are sparse, retrieved TPW from satellite observations can be used as one of the main parameters for predicting TC rainfall. In section 3, we investigate the relationship, if any, between the temporal variations of high TPW extending from the tropics to the extratropics and the heavy rainfall over the Hawaiian Islands during the 2006 wet period. In section 4, the impacts of latent heat release on horizontal moisture convergence and the development of heavy rainfall for this particular case will be presented.
Stability indices, including the K index (KI; George 1960), the total totals index (TTI; Miller 1972), the lifted index (LI; Galway 1956), and convective available potential energy (CAPE), are commonly used for estimating thunderstorm probability over the continental United States. The K index reflects the moisture content at the lower and middle troposphere in addition to the lapse rate between the 850- and 500-hPa levels. When it is moist at the 850- and 700-hPa levels, KI values are relatively high. The TTI is the sum of two other indices: the vertical totals index (temperature at 850 hPa minus temperature at 500 hPa) and the cross totals index (dewpoint at 850 hPa minus temperature at 500 hPa). When the TTI reaches 48, a few scattered moderate (or even a few heavy–isolated severe) thunderstorms are likely to occur. The LI is a measure of stability representing the buoyancy of a surface air parcel at the 500-hPa level. Negative LI indicates the possibility for thunderstorm occurrences.
Kodama and Barnes (1997) used station soundings and the surface temperature from buoy stations to calculate the stability indices over the island of Hawaii (e.g., LI, KI, and CAPE). They examined the correlation between these stability indices and 24-h rainfall accumulation averaged from six stations over the southeastern flank of Mauna Loa volcano (SE ML) from 1978 through 1992. They showed that the LI and CAPE are less correlated with rainfall than the KI. The KI serves as better guidance in forecasting heavy rainfall since the midlevel moisture is positively correlated with the development of SE ML (windward side) heavy rainfall events.
However, there are only two rawinsonde sites with twice-daily soundings over the state of Hawaii [PHLI at Lihue, Kauai; and PHTO at Hilo (Big Island), Hawaii] (Fig. 2). The atmospheric soundings are available about 9 h prior to heavy rainfall occurrence and there are no sounding sites on Oahu and the surrounding ocean. Although there was high KI and TTI and low CAPE and negative LI from the Lihue sounding at 1200 UTC 31 March, the KI derived from this sounding cannot provide precise locations where rainstorms may have occurred. Since the WRF model can provide high temporal evolution patterns of stability indices with a fine spatial resolution over the open ocean, we would like to investigate if horizontal distributions of instability indices derived from the model output are useful in pinpointing the location for the development of heavy rainfall in advance. The model-predicted radar reflectivities and accumulated rainfall patterns will be compared with observed patterns to evaluate the model performance. We will assess the model-predicted horizontal distributions of stability indices and their relationship with the simulated heavy rainfall occurrences associated with this Kona low event, especially the K index [KI = (T850 − T500) + Td850 − (T700 + Td700)].
The area used for area-averaged TPW (kg m−2) and 850-hPa meridional wind (m s−1) over HI (18.5°–22.5°N, 160.5°–154.5°W). The box represents the area used for area-averaged WSR-88D (from the Molokai site located at ×) derived rainfall over Oahu (21°–22°N, 158.5°–157.5°W). The two routine rawinsonde stations are located at Lihue, Kauai (PHLI, located at L) and Hilo on the Big Island (PHTO, located at H). The dashed line is the NE–SW cross section line along the storm propagation direction for Fig. 21.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The area used for area-averaged TPW (kg m−2) and 850-hPa meridional wind (m s−1) over HI (18.5°–22.5°N, 160.5°–154.5°W). The box represents the area used for area-averaged WSR-88D (from the Molokai site located at ×) derived rainfall over Oahu (21°–22°N, 158.5°–157.5°W). The two routine rawinsonde stations are located at Lihue, Kauai (PHLI, located at L) and Hilo on the Big Island (PHTO, located at H). The dashed line is the NE–SW cross section line along the storm propagation direction for Fig. 21.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The area used for area-averaged TPW (kg m−2) and 850-hPa meridional wind (m s−1) over HI (18.5°–22.5°N, 160.5°–154.5°W). The box represents the area used for area-averaged WSR-88D (from the Molokai site located at ×) derived rainfall over Oahu (21°–22°N, 158.5°–157.5°W). The two routine rawinsonde stations are located at Lihue, Kauai (PHLI, located at L) and Hilo on the Big Island (PHTO, located at H). The dashed line is the NE–SW cross section line along the storm propagation direction for Fig. 21.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The large-scale conditions favorable for the development of the Kahala Mall flood will be investigated in section 4. Influences of the eastward-moving midlatitude trough on the deepening of the anomalous low northwest of Hawaii during the Kahala Mall flood case will also be studied. The low-level tropical moisture and the upper-level baroclinic forcing associated with the large-scale circulation patterns will be examined. The high-resolution WRF model will be used as a reasonable tool to study the impacts of latent heat (LH) release on flow patterns, moisture fields, and atmospheric instability for this Kona low case.
Local terrain affects rainfall patterns over the Hawaiian Island chain. Orographic effects may focus rainfall in certain areas. Previous studies of orographic effects on heavy precipitation over Hawaii have focused on the orographic lifting of the low-level warm, moist flow. Schroeder (1977) studied a heavy rainfall event over Oahu associated with an upper-level trough. He showed that the low-level wind direction is important in determining horizontal distributions of rainfall patterns. With the Ko’olau Mountains serving as an anchor for the cumulonimbi, heavy rainfall persisted for 5 h for this particular case. Kodama and Barnes (1997) found that orographic lifting of the warm, moist upslope-wind component is important for the occurrence of heavy rainfall on the SE ML on the island of Hawaii for the Kona storm and cold front cases. Zhang et al. (2005) studied a wintertime heavy rainfall event over the island of Hawaii during 1–2 November 2000. They used the coupled Mesoscale Spectral Model with an advanced Land Surface Model (MSM-LSM) to study the effects of orographic lifting on rainfall amounts over the island of Hawaii and found that the rainfall amount is significantly less with a reduced mountain height. Chu et al. (2009) showed that, climatologically, the horizontal distribution of rainfall over Oahu was affected by terrain for heavy–very heavy rainfall events with maximum rainfall at the top of the southeastern portion of the Ko’olau Mountain Range.
Cram and Tatum (1979) studied heavy rainfall events on the island of Hawaii during January–February 1979. They reported a heavy rainfall event on 20 February caused by widespread rainstorms over the island of Hawaii. From the Hilo sounding data, the freezing level was at 4084 m. Based on visual observations from the Mauna Loa Observatory (3398 m), all cloud tops were below the freezing level, which implies that the rainfall was mainly from the warm rain process. In section 5 of this paper, the Weather Surveillance Radar-1988 Doppler (WSR-88D) data from the Molokai site (Fig. 2) will be used to depict the horizontal distribution of echo tops and orographic effects on the development of localized heavy rainfall in Kahala. We would like to determine whether orographic effects are important for focusing rainfall over land, after the storms drift inland, for this leeside heavy rainfall case.
2. Data and methodology
a. NCEP FNL global tropospheric analyses data and NCEP–NCAR reanalysis data
The National Centers for Environmental Prediction (NCEP) Final (FNL) global tropospheric analyses 1° × 1° gridded data are used to analyze the large-scale weather conditions for the Kahala Mall flood case. The climatological mean and anomalous atmospheric conditions are constructed from the 2.5° × 2.5°NCEP–NCAR reanalysis data. The long-term climatological mean data are computed over a 29-yr period (1968–96).
b. Satellite, Doppler radar, and rain gauge observations
The daily SSM/I (Hollinger 1989) TPW data (Alishouse et al. 1990) from the Defense Meteorological Satellite Program (DMSP) series of satellites (F-13, F-14, and F-15) are used. The SSM/I TPW data derived from Wentz’s (1997) algorithm are ideal for describing the horizontal distribution of moisture over the ocean. The 3-h precipitation (Fulton et al. 1998) derived from the WSR-88D data from the Molokai (PHMO) site (Fig. 2) is used to compute 12-h rainfall accumulation during the 2006 wet period. The level-III radar data including the horizontal distribution of composite reflectivities, echo tops, and vertically integrated liquid water content (Klazura and Imy 1993; Greene and Clark 1972) associated with the rainstorms will be used.
The rain gauge stations, including 26 stations from the NWS Hydronet network, 33 other stations operated by the National Climate Data Center (NCDC), the Automated Surface Observing Systems (ASOS), the Remote Automated Weather Stations (RAWS), and the U.S. Geological Survey (USGS) (Fig. 1), are used to investigate the rainfall distribution for the Kahala Mall flood case.
c. Numerical modeling
The WRF, which is a numerical weather prediction (NWP) and atmospheric simulation system, has been widely used for research and operational purposes (information online at http://www.wrf-model.org). In this study, we examine the model performance and limitations of the Advanced Research version of the WRF (ARW) for this heavy rainfall case. In addition, a sensitivity test concerning the impact of latent heat (LH) release on the development of the Kahala Mall flood is made by turning off the latent heating.
The WRF-ARW uses the sigma (terrain following) hydrostatic-pressure vertical coordinate (Skamarock et al. 2008). The Rapid Radiative Transfer Model (RRTM; Mlawer et al. 1997), Dudhia (1989) radiation schemes, the Noah land surface model (LSM) (Chen and Dudhia 2001) and the Yonsei University (YSU) planetary boundary layer scheme (Hong et al. 2006) are used. The Betts–Miller–Janjić (BMJ) cumulus parameterization scheme (Janjić 1994, 2000) and the Ferrier microphysics scheme (Rogers et al. 2001; Ferrier et al. 2002) are employed. The Ferrier grid-scale cloud and precipitation scheme predicts variations of six species of water substances (cloud water, cloud ice/small ice crystals, rain, snow, graupel, and sleet).
There are 28 sigma levels1 from the surface to the 50-hPa level. The three domains with resolutions of 18, 6, and 2 km, respectively, employed in this study are shown in Fig. 3. The cumulus parameterization is turned off in the 2-km-resolution domain (domain 3). The Global Forecast System (GFS) model output, with 1° horizontal resolution, provides the initial and boundary conditions for the model simulation.
The 3 domains employed in this study, with resolutions of 18, 6, and 2 km, respectively.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The 3 domains employed in this study, with resolutions of 18, 6, and 2 km, respectively.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The 3 domains employed in this study, with resolutions of 18, 6, and 2 km, respectively.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
3. Relationship between temporal variations of the moist tongue and heavy rainfall during the entire wet period
Throughout the entire 2006 wet period, the areal-averaged TPW over Hawaii (Fig. 2) was almost always higher than the climatological value during this time of the year (~28.85 kg m−2), especially during the five periods with a southerly wind component over the island chain. These five periods were 1) 19–22 February, 2) 28 February–3 March, 3) 8–11 March, 4) 13–19 March, and 5) 21 March–2 April. During these periods, the areal-averaged TPW is greater than 35 kg m−2 (Fig. 4). These 5 periods corresponded to the five heavy rainfall episodes when flash flood watches (meaning flooding possible within the next 36 h) issued by the NWS Forecast Office were in effect (Nash et al. 2010).
Time series of area-averaged TPW (kg m−2) and 850-hPa meridional wind (m s−1) over HI throughout the wet period. The NCEP FNL TPW is represented by the plus sign (+), the daily SSM/I TPW by the thick solid line, and the NCEP–NCAR reanalysis daily long-term mean TPW by the thin solid line. The NCEP FNL 850-hPa meridional wind is represented by the thick dashed line. The histograms show the 12-h accumulated rainfall (mm) over Oahu (Fig. 2) derived from the Molokai radar site during 19 Feb–2 Apr 2006. The five gray horizontal bars near the panel top represent the five periods with moist (when area-averaged TPW was >35 kg m−2) southerly winds prevailing over the Hawaiian Islands. The five periods are 1) 19–22 Feb, 2) 28 Feb–3 Mar, 3) 8–11 Mar, 4) 13–19 Mar, and 5) 21 Mar–2 Apr. (These five periods with moist southerly winds correspond to the five heavy rainfall episodes.)
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
Time series of area-averaged TPW (kg m−2) and 850-hPa meridional wind (m s−1) over HI throughout the wet period. The NCEP FNL TPW is represented by the plus sign (+), the daily SSM/I TPW by the thick solid line, and the NCEP–NCAR reanalysis daily long-term mean TPW by the thin solid line. The NCEP FNL 850-hPa meridional wind is represented by the thick dashed line. The histograms show the 12-h accumulated rainfall (mm) over Oahu (Fig. 2) derived from the Molokai radar site during 19 Feb–2 Apr 2006. The five gray horizontal bars near the panel top represent the five periods with moist (when area-averaged TPW was >35 kg m−2) southerly winds prevailing over the Hawaiian Islands. The five periods are 1) 19–22 Feb, 2) 28 Feb–3 Mar, 3) 8–11 Mar, 4) 13–19 Mar, and 5) 21 Mar–2 Apr. (These five periods with moist southerly winds correspond to the five heavy rainfall episodes.)
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
Time series of area-averaged TPW (kg m−2) and 850-hPa meridional wind (m s−1) over HI throughout the wet period. The NCEP FNL TPW is represented by the plus sign (+), the daily SSM/I TPW by the thick solid line, and the NCEP–NCAR reanalysis daily long-term mean TPW by the thin solid line. The NCEP FNL 850-hPa meridional wind is represented by the thick dashed line. The histograms show the 12-h accumulated rainfall (mm) over Oahu (Fig. 2) derived from the Molokai radar site during 19 Feb–2 Apr 2006. The five gray horizontal bars near the panel top represent the five periods with moist (when area-averaged TPW was >35 kg m−2) southerly winds prevailing over the Hawaiian Islands. The five periods are 1) 19–22 Feb, 2) 28 Feb–3 Mar, 3) 8–11 Mar, 4) 13–19 Mar, and 5) 21 Mar–2 Apr. (These five periods with moist southerly winds correspond to the five heavy rainfall episodes.)
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
During 19 February–19 March, as the areal-averaged 850-hPa southerly wind component gradually developed and reached a peak, the areal-averaged TPW over Hawaii also reached a peak with higher rainfall over Oahu (Fig. 4). When the southerly wind weakened or was replaced by a northerly wind component, the areal-averaged TPW over Hawaii also decreased with correspondingly less rainfall. The latitude–time cross sections clearly show that during five heavy rainfall episodes the low-level southerly flow extended from low latitudes to Hawaii with tropical moist air covering the state (Fig. 5). Figure 6 shows anomalous moist conditions throughout the entire wet period, especially during the five heavy rainfall episodes that had a significant positive southerly wind anomaly over Hawaii. The longitude–time cross section also shows that the development of anomalous strong southerly flow over Hawaii was related to the deepening of the subtropical semipermanent low pressure system west of Hawaii (~180° at the beginning of the wet period) as eastward-moving upper-level transient disturbances arrived (Fig. 7b). Both the upper- and lower-level southerly winds east of the low pressure system strengthened (Figs. 7a and 7c) with relatively moist conditions over Hawaii during the five heavy rainfall episodes (Figs. 7c and 7d). During the fifth heavy rainfall episode, the semipermanent low pressure system shifted eastward toward the Hawaiian region (~165°W) (Fig. 7b). As a result, low-level southerly winds prevailed continuously with high TPW and frequent heavy rainfall events over Oahu during this period (Figs. 4, 7c, and 7d).
Latitude–time cross sections of (a) SSM/I, (b) NCEP FNL TWP (contoured every 10 kg m−2, shaded for values >30 kg m−2), and (c) NCEP FNL 850-hPa meridional wind (contoured every 5 m s−1, shaded for values >0 m s−1) along 158°W from 19 Feb to 2 Apr 2006. The thin light gray horizontal bands mark the Hawaiian region (18.5°–22.5°N). The five black horizontal bars represent the five heavy rainfall episodes.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
Latitude–time cross sections of (a) SSM/I, (b) NCEP FNL TWP (contoured every 10 kg m−2, shaded for values >30 kg m−2), and (c) NCEP FNL 850-hPa meridional wind (contoured every 5 m s−1, shaded for values >0 m s−1) along 158°W from 19 Feb to 2 Apr 2006. The thin light gray horizontal bands mark the Hawaiian region (18.5°–22.5°N). The five black horizontal bars represent the five heavy rainfall episodes.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
Latitude–time cross sections of (a) SSM/I, (b) NCEP FNL TWP (contoured every 10 kg m−2, shaded for values >30 kg m−2), and (c) NCEP FNL 850-hPa meridional wind (contoured every 5 m s−1, shaded for values >0 m s−1) along 158°W from 19 Feb to 2 Apr 2006. The thin light gray horizontal bands mark the Hawaiian region (18.5°–22.5°N). The five black horizontal bars represent the five heavy rainfall episodes.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
As in Fig. 5, but for the anomalous NCEP–NCAR reanalysis along 157.5°W.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
As in Fig. 5, but for the anomalous NCEP–NCAR reanalysis along 157.5°W.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
As in Fig. 5, but for the anomalous NCEP–NCAR reanalysis along 157.5°W.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
As in Fig. 6, but for longitude–time cross section (a) 250-hPa meridional winds (contoured every 20 m s−1, shaded for values >20 m s−1), (b) 250-hPa geopotential height (102 gpm, contoured every 0.5 102 gpm, shaded for values <0 gpm), (c) 850-hPa meridional winds (contoured every 5 m s−1, shaded for values >0 m s−1), and (d) TPW (contoured every 5 kg m−2, shaded for values >0 kg m−2) along 20°N. The thin light gray vertical bands mark the Hawaiian region (160.5°–154.5°W).
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
As in Fig. 6, but for longitude–time cross section (a) 250-hPa meridional winds (contoured every 20 m s−1, shaded for values >20 m s−1), (b) 250-hPa geopotential height (102 gpm, contoured every 0.5 102 gpm, shaded for values <0 gpm), (c) 850-hPa meridional winds (contoured every 5 m s−1, shaded for values >0 m s−1), and (d) TPW (contoured every 5 kg m−2, shaded for values >0 kg m−2) along 20°N. The thin light gray vertical bands mark the Hawaiian region (160.5°–154.5°W).
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
As in Fig. 6, but for longitude–time cross section (a) 250-hPa meridional winds (contoured every 20 m s−1, shaded for values >20 m s−1), (b) 250-hPa geopotential height (102 gpm, contoured every 0.5 102 gpm, shaded for values <0 gpm), (c) 850-hPa meridional winds (contoured every 5 m s−1, shaded for values >0 m s−1), and (d) TPW (contoured every 5 kg m−2, shaded for values >0 kg m−2) along 20°N. The thin light gray vertical bands mark the Hawaiian region (160.5°–154.5°W).
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
4. Favorable large-scale conditions for the development of the Kahala Mall flood
The time series of hourly rainfall at two NWS hydronet rain gauge stations, Niu Valley and Palolo F S, close to the Kahala Mall (Fig. 1) show that the heaviest rainfall occurred from 1100 to 1300 HST (2100 to 2300 UTC) on 31 March (Fig. 8). The 2-h accumulated rainfall totals were 73.2 mm at Niu Valley and 61 mm at Palolo F S. Figure 9 shows that the simulated Kona low exhibits cyclonic flow at upper levels with diminishing strength at lower levels. At 2200 UTC, easterly winds prevailed from the surface up to the 950-hPa level, just below the top of the Ko’olau Mountain Range (~700 m), over southeastern Oahu (Figs. 9b and 9d). At the 850-hPa level, southerly winds prevailed over Oahu, veering to southwesterly winds at the 300-hPa level (Figs. 9a and 9c), which was on the east flank of the Kona low, implying warm advection over Oahu. The low-level convergence over Oahu (Figs. 9b–d) was associated with significant geopotential height drops due to latent heating (Fig. 17b) associated with the heavy rainstorm (Fig. 19a).
Time series of hourly rainfall (mm) at the Niu Valley rain gauge (HI-06) and the Palolo FS rain gauge (HI-25) on 31 Mar 2006.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
Time series of hourly rainfall (mm) at the Niu Valley rain gauge (HI-06) and the Palolo FS rain gauge (HI-25) on 31 Mar 2006.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
Time series of hourly rainfall (mm) at the Niu Valley rain gauge (HI-06) and the Palolo FS rain gauge (HI-25) on 31 Mar 2006.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The WRF simulated horizontal winds (each full barb = 10 kt, or 5.14 m s−1) at the (a) 300-, (b) 950-, and (c) 850-hPa levels and (d) the 10-m level at 2200 UTC 31 Mar from domain 1.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The WRF simulated horizontal winds (each full barb = 10 kt, or 5.14 m s−1) at the (a) 300-, (b) 950-, and (c) 850-hPa levels and (d) the 10-m level at 2200 UTC 31 Mar from domain 1.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The WRF simulated horizontal winds (each full barb = 10 kt, or 5.14 m s−1) at the (a) 300-, (b) 950-, and (c) 850-hPa levels and (d) the 10-m level at 2200 UTC 31 Mar from domain 1.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
From the 2-h rainfall accumulation, it is apparent that significant rainfall occurred on the leeside of the Ko’olau Mountain Range (relative to the easterly prevailing winds below the mountaintop) along the southeastern portion of Oahu with maximum rainfall at the summit (Fig. 1). The heavy rainfall was generated by a convective line followed by an intense rainstorm moving from southwest of Oahu onshore over the southeastern coast. The heaviest rainfall occurred around 1157 HST (2157 UTC; see Fig. 10). The intense rainstorm covered almost the entirety of Oahu with high reflectivities (~55 dBZ), echo tops > 30 000 ft (~9.1 km), and vertically integrated liquid water content VIL > 25 kg m−2 over southeastern Oahu. More details about the orographic effects on the rainfall distribution for this case will be presented in section 5.
(a) The composite reflectivities (dBZ), (b) echo tops (kft; 1 kft = 304.8m), and (c) VIL (kg m−2) from the Molokai radar site at 1157 HST (2157 UTC) 31 Mar 2006.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
(a) The composite reflectivities (dBZ), (b) echo tops (kft; 1 kft = 304.8m), and (c) VIL (kg m−2) from the Molokai radar site at 1157 HST (2157 UTC) 31 Mar 2006.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
(a) The composite reflectivities (dBZ), (b) echo tops (kft; 1 kft = 304.8m), and (c) VIL (kg m−2) from the Molokai radar site at 1157 HST (2157 UTC) 31 Mar 2006.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
a. Moist tongue from the tropics
Figure 11 shows the geopotential height at the 250- and 850-hPa levels during 26–28 March 2006, prior to the Kahala Mall flooding. At the 250-hPa level, the geopotential height field was characterized by an atmospheric blocking pattern in the central North Pacific with a ridge over the Aleutian Islands and an anomalous low west of Hawaii in the subtropics. Eastward-moving troughs occurred simultaneously in the midlatitudes and the subtropics (Fig. 11, left). At the 850-hPa level, a stationary subtropical cyclone was located west of Hawaii (Fig. 11, right).
The NCEP FNL (left) 250- (102 gpm) and (right) 850-hPa geopotential heights (gpm) at 1200 UTC on (a) 26, (b) 27, and (c) 28 Mar 2006.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The NCEP FNL (left) 250- (102 gpm) and (right) 850-hPa geopotential heights (gpm) at 1200 UTC on (a) 26, (b) 27, and (c) 28 Mar 2006.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The NCEP FNL (left) 250- (102 gpm) and (right) 850-hPa geopotential heights (gpm) at 1200 UTC on (a) 26, (b) 27, and (c) 28 Mar 2006.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
On 31 March, at the 850-hPa level, southerly winds prevailed over Hawaii with a cyclonic flow to the west and an anticyclonic flow to the east (Fig. 9c). The horizontal distribution of SSM/I daily mean total precipitable water on 31 March shows a moist tongue (TPW > 35 kg m−2) extending from the deep tropics across Hawaii (Fig. 12a). The moisture content for the NE–SW TPW band (30–45 kg m−2) reaching Hawaii was much higher than the climatological areal-averaged TPW over Hawaii (~28.85 kg m−2) during this time of the year. The tropical moisture was brought into the Hawaii region by the low-level (850-hPa level) southerly flow (Fig. 9c). The SSM/I TPW image also captures the counterclockwise spiral of moisture due to the cyclonic flow associated with the Kona low (Fig. 12a). The horizontal distribution of NCEP FNL TPW shows a similar pattern as the SSM/I TPW (Figs. 12a and 12b). A high equivalent potential temperature (θe) axis, extending from the tropical central Pacific to Hawaii at the 850-hPa level (Fig. 12c) and corresponding to the moist tongue (Figs. 12a and 12b), is also evident. It is apparent that the moist tongue rooted in the deep tropics contributed to the convective instability over the Hawaiian region.
The daily mean (a) SSM/I and (b) NECP FNL TPW (kg m−2), and (c) NCEP FNL θe (K) at 850 hPa on 31 Mar.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The daily mean (a) SSM/I and (b) NECP FNL TPW (kg m−2), and (c) NCEP FNL θe (K) at 850 hPa on 31 Mar.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The daily mean (a) SSM/I and (b) NECP FNL TPW (kg m−2), and (c) NCEP FNL θe (K) at 850 hPa on 31 Mar.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
b. Midlatitude baroclinic forcing on upper-level cyclogenesis and convective activity associated with the Kona storm
At 1200 UTC 28 March, the upper-tropospheric QG frontogenesis associated with both the midlatitude and the subtropical troughs (Fig. 13c) was diagnosed as the northerly flow behind the troughs (Fig. 13a) brought colder air from the north (Fig. 13b). The high IPV (IPV on the 335-K isentropic surface >2 PVU) air associated with tropopause folding (Fig. 13d) was clearly evident west of the trough axis. At 0000 UTC 30 March, the midlatitude transient trough moved to around 150°W (Fig. 14c). The midlatitude transient trough deepened with significant northerly winds behind the trough axis bringing cold air from the midlatitudes to the subtropics (Figs. 14a–c). The midlatitude trough then merged with the subtropical trough west of Hawaii. A NE–SW-oriented axis of pronounced upper-level QG frontogenesis extended from the northeast Pacific to the central Pacific with a maximum located around 25°N, 167°W (Fig. 14c). The significant cold advection by northerly winds behind both the midlatitude and the subtropical troughs and warm advection by southerly winds ahead of the troughs induced enhanced QG frontogenesis and tropopause folding in the subtropics (Fig. 14).
(a) The 250-hPa winds (m s−1) contoured every 20 m s−1 (shaded for values >40 m s−1), (b) the 250-hPa potential temperature (K, contoured, shaded for values <330 K), (c) the 250-hPa QG frontogenesis (10−9 K m−1 s−1, shaded) and geopotential height (102 gpm, contoured), and (d) IPV (PVU) on the 335-K isentropic surface at 1200 UTC 28 Mar 2006 from the NCEP 1° × 1° FNL global tropospheric analyses.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
(a) The 250-hPa winds (m s−1) contoured every 20 m s−1 (shaded for values >40 m s−1), (b) the 250-hPa potential temperature (K, contoured, shaded for values <330 K), (c) the 250-hPa QG frontogenesis (10−9 K m−1 s−1, shaded) and geopotential height (102 gpm, contoured), and (d) IPV (PVU) on the 335-K isentropic surface at 1200 UTC 28 Mar 2006 from the NCEP 1° × 1° FNL global tropospheric analyses.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
(a) The 250-hPa winds (m s−1) contoured every 20 m s−1 (shaded for values >40 m s−1), (b) the 250-hPa potential temperature (K, contoured, shaded for values <330 K), (c) the 250-hPa QG frontogenesis (10−9 K m−1 s−1, shaded) and geopotential height (102 gpm, contoured), and (d) IPV (PVU) on the 335-K isentropic surface at 1200 UTC 28 Mar 2006 from the NCEP 1° × 1° FNL global tropospheric analyses.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
As in Fig. 13, but for 0000 UTC 30 Mar 2006 and with the addition of ageostrophic winds (m s−1, vectors) in (c).
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
As in Fig. 13, but for 0000 UTC 30 Mar 2006 and with the addition of ageostrophic winds (m s−1, vectors) in (c).
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
As in Fig. 13, but for 0000 UTC 30 Mar 2006 and with the addition of ageostrophic winds (m s−1, vectors) in (c).
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
According to Carlson (1998), the tropopause folding associated with the high-PV air intruding from the stratosphere into the upper, middle, and even lower troposphere is linked to the descending motions in upper levels. In this case, the tropopause folding took place in the western (cold) flank of both the midlatitude and the subtropical troughs where the upper-level convergence of ageostrophic winds (descending motion) was apparent (Fig. 14). The divergence of ageostrophic winds (ascending motion) took place over the eastern (warm) flank of the subtropical trough. Ahead of the trough axis, the southeasterly ageostrophic winds were perpendicular to geopotential height contours (Fig. 14c). The Coriolis force acting on the southeasterly ageostrophic winds contributed to the strengthening of the southwesterly subtropical jet from 0000 UTC 30 March to 0000 UTC 31 March (Figs. 14a and 15a).
As in Fig. 13, but for at 0000 UTC 31 Mar 2006 and with the addition of 500-hPa pressure vertical velocity (Pa s−1, shaded) in (d).
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
As in Fig. 13, but for at 0000 UTC 31 Mar 2006 and with the addition of 500-hPa pressure vertical velocity (Pa s−1, shaded) in (d).
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
As in Fig. 13, but for at 0000 UTC 31 Mar 2006 and with the addition of 500-hPa pressure vertical velocity (Pa s−1, shaded) in (d).
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The high-PV air intruding from the stratosphere into the upper troposphere contributed to the spinup of the cyclonic flow associated with the subtropical trough in the upper troposphere (Figs. 14a and 15a). When stratospheric high-PV air intrudes into the troposphere with a less stable environment, the absolute vorticity (ςθ + f) must increase to satisfy the conservation of isentropic potential vorticity (Carlson 1998). For this case, the baroclinic forcing associated with the southwestward extension of the midlatitude trough that merged with the subtropical trough was a crucial mechanism of the deepening subtropical trough and the cyclogenesis of the Kona low (Fig. 14).
On 31 March, the midlatitude trough continued to move eastward. The deepened subtropical trough was cut off from the midlatitude westerly current (Figs. 15a and 15c). Meanwhile, a relatively warm ridge behind the midlatitude trough occupied the region north and northeast of the Kona low (Figs. 15a–c). The advection of the high-PV air at upper levels by the southwesterly wind induced the upward motion downstream over Hawaii (Figs. 15a and 15d). In addition, the southeast portion of the subtropical jet streak exit region was just over Oahu (Fig. 15a). All these factors contributed to the occurrence of a NE–SW band of upward motion across Oahu at 0000 UTC 31 March (Fig. 15d).
Figure 16 shows the longitude–height cross sections of EPV, potential temperature, absolute vorticity, zonal winds, and pressure vertical velocity along 21°N at 0000 UTC 31 March. It clearly shows that the descending motion brought down the high-PV stratospheric air into the upper-level frontal zone (Figs. 16a and 16d) and spun up the Kona low (with high vorticity values; see Fig. 16b). The high-PV air aloft, advected by the westerly wind component, enhanced the ascending motion downstream over Hawaii (around 158°W), in the eastern flank of the Kona storm (centered near 163°W) (Figs. 16c and 16d).
Longitude–height cross sections of (a) EPV (PVU, gray dashed lines) and potential temperature (K, black contours), (b) absolute vorticity (10−5 s−1, contours, shaded for values >2.5 × 10−4 s−1), (c) EPV (PVU, gray dashed lines) and zonal winds (m s−1, black contours), and (d) EPV (PVU, gray dashed lines) and pressure vertical velocity (Pa s−1, black contours) along 21°N at 0000 UTC 31 Mar 2006 from the NCEP 1° × 1° FNL global tropospheric analyses (Hawaii region is 160.5°–154.5°W).
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
Longitude–height cross sections of (a) EPV (PVU, gray dashed lines) and potential temperature (K, black contours), (b) absolute vorticity (10−5 s−1, contours, shaded for values >2.5 × 10−4 s−1), (c) EPV (PVU, gray dashed lines) and zonal winds (m s−1, black contours), and (d) EPV (PVU, gray dashed lines) and pressure vertical velocity (Pa s−1, black contours) along 21°N at 0000 UTC 31 Mar 2006 from the NCEP 1° × 1° FNL global tropospheric analyses (Hawaii region is 160.5°–154.5°W).
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
Longitude–height cross sections of (a) EPV (PVU, gray dashed lines) and potential temperature (K, black contours), (b) absolute vorticity (10−5 s−1, contours, shaded for values >2.5 × 10−4 s−1), (c) EPV (PVU, gray dashed lines) and zonal winds (m s−1, black contours), and (d) EPV (PVU, gray dashed lines) and pressure vertical velocity (Pa s−1, black contours) along 21°N at 0000 UTC 31 Mar 2006 from the NCEP 1° × 1° FNL global tropospheric analyses (Hawaii region is 160.5°–154.5°W).
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
c. Effects of latent heat release/thermally induced circulation
The WRF-ARW is used to investigate the impacts of LH on the environment. The model is initialized at 0000 UTC 31 March, about 1 day before the flooding occurs. The LH release on the warm, moist, cloudy (eastern) flank of the Kona low (Morrison and Businger 2001) warms the midtroposphere by up to 1°–3°C (Fig. 17a) and leads to the geopotential height drop at the 850-hPa level at 2200 UTC 31 March (Fig. 17b). The 850-hPa geopotential height drop northwest of Hawaii, with the maximum magnitude centered at 24°N, 154°W, modifies the Kona storm cyclonic flow pattern in low levels (Fig. 18). The southerly wind component of the cyclonic flow associated with the Kona storm shifts farther eastward toward Oahu in the control run (CTRL; see Figs. 18a and 18c) as compared to the run without LH release (WOLH; see Figs. 18b and 18d). The convergence between moist southwesterly and southeasterly flows is evident over the area south of Oahu at low levels in the CTRL run (Fig. 18c). The significant geopotential height drop at the 850-hPa level associated with latent heat release (Fig. 17b) enhances the low-level moisture convergence axis across Oahu (Figs. 18e and 18f), which is favorable for the development of heavy rainfall.
The WRF model simulated (a) 700-hPa temperature difference (°C, shaded, contoured for values <0°C), (b) 850-hPa geopotential [m2 s−2; geopotential height × g (9.8 m s−2)] differences between the LH and WOLH cases (LH minus WOLH), and lapse rate term in the K index (T850 − T500) in the (c) LH and (d) WOLH cases at 2200 UTC 31 Mar 2006 from domain 1.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The WRF model simulated (a) 700-hPa temperature difference (°C, shaded, contoured for values <0°C), (b) 850-hPa geopotential [m2 s−2; geopotential height × g (9.8 m s−2)] differences between the LH and WOLH cases (LH minus WOLH), and lapse rate term in the K index (T850 − T500) in the (c) LH and (d) WOLH cases at 2200 UTC 31 Mar 2006 from domain 1.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The WRF model simulated (a) 700-hPa temperature difference (°C, shaded, contoured for values <0°C), (b) 850-hPa geopotential [m2 s−2; geopotential height × g (9.8 m s−2)] differences between the LH and WOLH cases (LH minus WOLH), and lapse rate term in the K index (T850 − T500) in the (c) LH and (d) WOLH cases at 2200 UTC 31 Mar 2006 from domain 1.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The WRF-simulated 700-hPa dewpoint (°C, shaded) and winds (m s−1) (a) with and (b) without LH release. The 850-hPa dewpoint (°C, shaded) and winds (m s−1) (c) with and (d) without LH release. The 850-hPa moisture convergence (10−4 g kg−1 s−1) (e) with and (f) without LH release. All at 2200 UTC 31 Mar 2006 from domain 1.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The WRF-simulated 700-hPa dewpoint (°C, shaded) and winds (m s−1) (a) with and (b) without LH release. The 850-hPa dewpoint (°C, shaded) and winds (m s−1) (c) with and (d) without LH release. The 850-hPa moisture convergence (10−4 g kg−1 s−1) (e) with and (f) without LH release. All at 2200 UTC 31 Mar 2006 from domain 1.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The WRF-simulated 700-hPa dewpoint (°C, shaded) and winds (m s−1) (a) with and (b) without LH release. The 850-hPa dewpoint (°C, shaded) and winds (m s−1) (c) with and (d) without LH release. The 850-hPa moisture convergence (10−4 g kg−1 s−1) (e) with and (f) without LH release. All at 2200 UTC 31 Mar 2006 from domain 1.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
d. Heavy rainfall forecast and stability indices
The 22-h forecasted horizontal distribution of maximum radar reflectivities shows an intense rainstorm over southeastern Oahu (Fig. 19a), consistent with the observations (Fig. 10a). The WRF-simulated maximum radar reflectivities over southeastern Oahu (~45 dBZ) are slightly smaller than observed (~55 dBZ). The simulated pattern of radar reflectivities over the ocean compare favorably with the observed radar echoes. The simulated northeastward propagation of rainstorms is in agreement with radar observations (not shown). However, the WRF has limited capabilities in predicting the occurrence of each individual storm and its evolution. The model-simulated horizontal distribution of 2-h accumulated rainfall over Oahu (Fig. 20) shows a similar pattern as the rain gauge observations (Fig. 1) with the maximum rainfall over the lee side of the southern end of the Ko’olau Mountain Range. However, the simulated rainfall maximum is underestimated by a factor of 2 compared to the rain gauge observations. The rainfall production in the model may be sensitive to the precipitation processes employed in the model.
The WRF-simulated (a) maximum radar reflectivities (dBZ) from domain 2 (6-km resolution) and (b) TTI, (c) CAPE (J kg−1), (d) LI, (e) KI, and (f) KI in the WOLH case from domain 1 (18-km resolution) at 2200 UTC 31 Mar 2006.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The WRF-simulated (a) maximum radar reflectivities (dBZ) from domain 2 (6-km resolution) and (b) TTI, (c) CAPE (J kg−1), (d) LI, (e) KI, and (f) KI in the WOLH case from domain 1 (18-km resolution) at 2200 UTC 31 Mar 2006.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The WRF-simulated (a) maximum radar reflectivities (dBZ) from domain 2 (6-km resolution) and (b) TTI, (c) CAPE (J kg−1), (d) LI, (e) KI, and (f) KI in the WOLH case from domain 1 (18-km resolution) at 2200 UTC 31 Mar 2006.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The WRF-simulated horizontal distribution of accumulated rainfall from 1100 to 1300 HST 31 Mar 2006 (from 2100 to 2300 UTC 31 Mar) over Oahu from domain 3.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The WRF-simulated horizontal distribution of accumulated rainfall from 1100 to 1300 HST 31 Mar 2006 (from 2100 to 2300 UTC 31 Mar) over Oahu from domain 3.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The WRF-simulated horizontal distribution of accumulated rainfall from 1100 to 1300 HST 31 Mar 2006 (from 2100 to 2300 UTC 31 Mar) over Oahu from domain 3.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The simulated horizontal distributions of TTI, CAPE, LI, and KI (Figs. 19b–e) show that the area with positive CAPE and negative LI corresponds to the area with a low-level southerly wind component (Figs. 9c and 18c). Both CAPE and LI provide information for the wide area where the lifted air would become positively buoyant without any information concerning moisture content above the lifted condensation level. The conditionally unstable air from low latitudes is advected northward to the Hawaii region by the low-level southerly winds. For the Kahala Mall flood case, the model-derived CAPE is relatively low (<1000 J kg−1), without a well-defined maximum axis across Oahu. Only KI and TTI reflect the moisture content. The horizontal distribution of KI (Fig. 19e) has a maximum axis along the low-level moisture convergence (Fig. 18e) across Oahu. The KI contains information for both the lapse rate and the moisture at the 850- and the 700-hPa levels with the maximum values along the axis of the warm, moist tongue. The simulated radar echoes (Fig. 19a) mainly occur along the maximum axis of KI (Fig. 19e). Thus, for this case, the simulated horizontal distribution of KI provides valuable information to help pinpoint the likely locations for possible heavy rainfall occurrences. It is apparent that for the development of heavy rainfall, a deep warm, moist layer (high KI) is required but high CAPE may not be needed.
It is interesting to note that KI is larger in the WOLH run compared to the CTRL run but lacks a well-defined KI maximum axis (Figs. 19e and 19f). The lapse rate term in the K index (temperature at the 850-hPa level minus temperature at the 500-hPa level) is reduced with latent heat release (Figs. 17c and 17d) due to the warming in the midtroposphere (not shown). The high moisture content at the 700-hPa level (Fig. 18a) is simulated above the 850-hPa moisture convergence axis (Fig. 18e). Thus, the horizontal distribution of simulated KI (Fig. 19e) in the CTRL run has a well-defined maximum over the 850-hPa moisture convergence zone (Fig. 18e). The model can capture the low-level moisture axis extending from the tropics to Oahu on the eastern flank of the Kona low where rainstorms are most likely to develop.
5. Orographic effects on rainfall distributions over Oahu
The Lihue sounding (Fig. 2) at 1200 UTC (0200 HST) 31 March shows that the freezing level was at 650 hPa (~4 km). At 2157 UTC (1157 HST), the echo tops of deep convection within the intense storm were greater than 30 000 ft (~9.1 km) over southeastern Oahu. In contrast to the case presented by Cram and Tatum (1979), precipitation with echo tops greater than 20 000 ft (~6.1 km) covered the entirety of Oahu (Figs. 10a and 10b) implying that the cold rain process was present for the Kahala Mall flood case.
To see the orographic effects of the convective line and the intense rainstorm on precipitation, we take a NE–SW cross section (Fig. 2) along the storm’s direction of movement. The time series of the cross sections of composite radar reflectivities (dBZ), echo tops (kft), and VIL (kg m−2) from 2001 to 2359 UTC are shown in Fig. 21. Around 2100 UTC, the convective line moved onshore along southeastern Oahu from the southwest. Within a half hour, the intense rainstorm also arrived at the Ko’olau Mountain Range (Fig. 21a). The echo tops of the convective line and the intense rainstorm increased from 25 000–30 000 ft (7.6–9.1 km) up to 35 000–40 000 ft (10.7–12.2 km) (Fig. 21b). Correspondingly, the VIL within the convective line and the rainstorm increased from 10–20 up to 30 (kg m−2) as they merged on the southern lee side of the Ko’olau Mountain Range (Fig. 21c). The precipitation was enhanced with higher liquid water content and taller echo tops on the south side and over the summit of the Ko’olau Mountain Range (Figs. 1, 10, and 21). It is apparent that as the storm was forced to move over the mountain by the southerly wind component above the ridgetop (Fig. 9c), the storm itself was lifted by the mountain ridge with enhanced precipitation. The orographic lifting of the intense rainstorm that merged with the convective line enhanced both the depth (echo tops) and the liquid water content (VIL) of the storm. The remnant of deep convection passing over the summit continued to move northeastward away from Oahu (Figs. 21a and 21b). The VIL (Fig. 21c) associated with the deep convection decreased back to 10–20 (kg m−2) as the rainstorm, moving offshore, decayed.
The time series cross section of WSR-88D (a) composite reflectivities (dBZ), (b) echo tops (kft), and (c) VIL (kg m−2) along the NE–SW line in Fig. 2 from 2001 to 2359 UTC 31 Mar 2006 with the Ko’olau Mountain height (m) superimposed.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The time series cross section of WSR-88D (a) composite reflectivities (dBZ), (b) echo tops (kft), and (c) VIL (kg m−2) along the NE–SW line in Fig. 2 from 2001 to 2359 UTC 31 Mar 2006 with the Ko’olau Mountain height (m) superimposed.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
The time series cross section of WSR-88D (a) composite reflectivities (dBZ), (b) echo tops (kft), and (c) VIL (kg m−2) along the NE–SW line in Fig. 2 from 2001 to 2359 UTC 31 Mar 2006 with the Ko’olau Mountain height (m) superimposed.
Citation: Weather and Forecasting 26, 3; 10.1175/2010WAF2222449.1
6. Summary
In this study, the relationship between the moisture advected from the tropics to Hawaii and the development of heavy rainfall over the Hawaiian Islands during the 2006 wet period is investigated. Strong anomalous southerly winds prevailed ahead of a deepening subtropical trough bringing in higher than usual moisture from the tropics to Hawaii during the five heavy rainfall episodes when flash flood watches were in effect according to the NWS: 1) 19–22 February, 2) 1–3 March, 3) 8–11 March, 4) 13–19 March, and 5) 21 March–2 April.
The favorable conditions for the development of the Kahala Mall flood on 31 March 2006 are studied. First, at the 850-hPa level, the southerly winds over the state, characterized by a stationary subtropical cyclone west of Hawaii, brought in abundant tropical moisture from the south beginning on 26 March. A high equivalent potential temperature (θe) axis extending from low latitudes to Hawaii indicated the existence of convective instability over the islands.
Second, upper-level baroclinic forcing contributed to the convective activity over Hawaii. When the eastward-moving midlatitude trough, extending southward, merged with the semipermanent subtropical trough at 0000 UTC 30 March, the cold air behind the midlatitude trough was advected to the subtropics. The significant cold advection by the northerly winds behind both the midlatitude and the subtropical troughs and the warm advection by the southerly winds ahead of the troughs induced upper-level QG frontogenesis and tropopause folding in the subtropics. The descending motion on the western (cold) flank of the subtropical trough brought down high-PV air from the stratosphere to a less stable upper troposphere. Consequently, the absolute vorticity increased to maintain the conservation of the isentropic potential vorticity (IPV). This led to the spinup of the Kona low from 30 to 31 March. The PV advection by the strengthened subtropical jet aloft (west of Hawaii for this case) contributed to the upward motion and cyclonic activity downstream over Hawaii.
The Weather Research and Forecasting Model (WRF) is used to study the impacts of latent heat (LH) release on the environmental flow for the Kahala Mall flood case. LH release warms the lower- and midlevel troposphere by up to 1°–3°C on the moist, cloudy (eastern) flank of the Kona low. The 850-hPa geopotential height drop northwest of Hawaii, due to the latent heating, contributes to the eastward shift of the Kona low as well as the low-level moisture tongue ahead of the low. At 2200 UTC 31 March, the large-scale flow patterns are characterized by a cyclonic flow to the west of Oahu, an anticyclonic flow to the east and a pronounced southerly wind component across Oahu in the LH run. The warming and the significant geopotential height drop are accompanied by more pronounced moisture convergence between southwesterly and southeasterly flows, at low levels, as compared to the WOLH run. The abundant tropical moisture brought by the southerly flow, east of the Kona low, at low levels is transported upward to the midtroposphere.
Horizontal distributions of the instability indices were also computed with high temporal resolution from the WRF output to assess the effectiveness of these indices in pinpointing the most likely areas for the development of storms and heavy rainfall. At 2200 UTC 31 March, the wide region with positive CAPE and negative LI coincides with the area with a southerly wind component extending from the tropics to Hawaii. The horizontal distribution of KI reflects the low-level moisture convergence axis across Oahu where KI has maximum values. Intense rainstorms over Oahu occur within this maximum KI area. However, CAPE is relative low (<1000 J kg−1) and lacks a well-defined maximum over Oahu. It is apparent that for the development of heavy rainfall, the presence of a deep warm, moist layer is required but high CAPE may not be needed. Both CAPE and LI provide information for the wide area over which the lifted air would become positively buoyant without any information regarding moisture content above the lifted condensation level. On the other hand, the KI contains information for both the lapse rate and the moisture at the 850- and the 700-hPa levels with the maximum values along the axis of the warm, moist tongue. Thus, among all the stability indices, the model-derived horizontal distribution of KI provides the best forecast guidance for this heavy rainfall event.
The WRF performance in predicting the heavy rainstorm is investigated from model-derived horizontal distributions of radar reflectivities and rainfall accumulation. The 2-h accumulated rainfall over Oahu from 2100 to 2300 UTC shows similar patterns to those of the rain gauge observations but the rainfall maximum over southeastern Oahu is underestimated by a factor of 2. The simulated radar reflectivities over the ocean also compare favorably with the observed radar echoes. Both the observed and simulated radar reflectivities move northeastward with the midlevel winds. However, the WRF has its limitation in predicting the exact location for the development of each individual storm and its evolution. There are also uncertainties in rainfall estimates from the model as the predicted rainfall may be sensitive to the precipitation physics used by the model.
Finally, the orographic lifting of the convective systems advancing over the mountain range, which is also crucial for the development of heavy rainfall, is investigated. Easterly winds prevailed upstream of Oahu from the surface to around the top of the southeastern portion of the Ko’olau Mountain Range on 31 March 2006 as an intense rainstorm, followed by a convective line, moved onshore from the southwest over the southeastern part of Oahu. Both the depth and the vertically integrated liquid water content (VIL) of the storm were enhanced by the Ko’olau Mountain Range as the storm merged with the previous convective line on the lee side of the range. Heavy rainfall occurred along the lee side of the Ko’olau Mountain Range, with maximum rainfall at the summit, from 1100 to 1300 HST (2100 to 2300 UTC). Some deep convective cells passed over the range and brought significant rainfall to the windward side of the range as well. The convective cells continued to move northeastward away from Oahu with decreasing radar echoes, echo tops, and VIL.
Acknowledgments
This work was supported by NOAA under Cooperative Agreement NA17RJ1230 and COMET/UCAR under Partners Project Grants S09-75789 and S09-75790. Suggestions from Drs. Thomas A. Schroeder and Fei-Fei Jin are highly appreciated. We also would like to give our acknowledgments to Haiyan Jiang and another anonymous reviewer for their helpful comments. The rain gauge data and WSR-88D radar data from the PHMO (Molokai) site are provided by the NOAA/National Weather Service (NWS) Forecast Office at Honolulu, Hawaii, and the NCDC’s Hierarchical Data Storage System (HDSS). We thank Hiep V. Nguyen, I. M. Shiromani P. Jayawardena, and David Hitzl for their assistance in the WRF modeling, SSM/I data, and text editing, respectively.
REFERENCES
Alishouse, J., Snyder S. , Vongsathorn J. , and Ferraro R. , 1990: Determinations of oceanic total precipitable water from the SSM/I. IEEE Trans. Geosci. Remote Sens., 28, 811–816.
Bao, J.-W., Michelson S. A. , Neiman P. J. , Ralph F. M. , and Wilczak J. M. , 2006: Interpretation of enhanced integrated water vapor bands associated with extratropical cyclones: Their formation and connection to tropical moisture. Mon. Wea. Rev., 134, 1063–1080.
Bluestein, H. B., 1986: Fronts and jet steaks: A theoretical perspective. Mesoscale Meteorology and Forecasting, P. S. Ray, Ed., Amer. Meteor. Soc., 173–215.
Blumenstock, D. L., and Price S. , 1967: Climates of the United States—Hawaii. Climatography of the United States 60-51, Dept. of Commerce/ESSA, 27 pp. [Available from NOAA Central Library, 2nd Fl., SSMC3, 1315 East-West Highway, Silver Spring, MD 20910.]
Carlson, T. N., 1998: Upper-tropospheric fronts and jet streaks. Mid-Latitude Weather Systems, T. N. Carlson, Ed., Amer. Meteor. Soc., 435–436.
Caruso, S. J., and Businger S. , 2006: Subtropical cyclogenesis over the central North Pacific. Wea. Forecasting, 21, 193–205.
Chen, F., and Dudhia J. , 2001: Coupling an advanced land surface–hydrology model with the Penn State–NCAR MM5 modeling system. Part I: Model implementation and sensitivity. Mon. Wea. Rev., 129, 569–585.
Chu, P.-S., Nash A. J. , and Porter F.-Y. , 1993: Diagnostic studies of two contrasting rainfall episodes in Hawaii: Dry 1981 and wet 1982. J. Climate, 6, 1457–1462.
Chu, P.-S., Zhao X. , Ruan Y. , and Grubbs M. , 2009: Extreme rainfall events in the Hawaiian Islands. J. Appl. Meteor. Climatol., 48, 502–516.
Cram, R. S., and Tatum H. R. , 1979: Record torrential rainstorms on the island of Hawaii, January–February 1979. Mon. Wea. Rev., 107, 1653–1662.
Dudhia, J., 1989: Numerical study of convection observed during the Winter Monsoon Experiment using a mesoscale two-dimensional model. J. Atmos. Sci., 46, 3077–3107.
Ertel, H., 1942: Ein neuer hydrodynamischer wirbelsatz. Meteor. Z., 59, 277–281.
Ferrier, B. S., Jin Y. , Lin Y. , Black T. , Rogers E. , and DiMego G. , 2002: Implementation of a new grid-scale cloud and precipitation scheme in the NCEP Eta Model. Preprints, 19th Conf. on Weather Analysis and Forecasting/15th Conf. on Numerical Weather Prediction, San Antonio, TX, Amer. Meteor. Soc., 10.1. [Available online at http://ams.confex.com/ams/SLS_WAF_NWP/techprogram/paper_47241.htm.]
Fulton, R. A., Breidenbach J. P. , Seo D.-J. , Miller D. A. , and O’Bannon T. , 1998: The WSR-88D rainfall algorithm. Wea. Forecasting, 13, 377–395.
Galway, J. G., 1956: The lifted index as a predictor of latent instability. Bull. Amer. Meteor. Soc., 37, 528–529.
George, J. J., 1960: Weather Forecasting for Aeronautics. Academic Press, 673 pp.
Greene, D. R., and Clark R. A. , 1972: Vertically integrated liquid water—A new analysis tool. Mon. Wea. Rev., 100, 548–552.
Hollinger, J. P., 1989: DMSP Special Sensor Microwave/Imager calibration/validation. Naval Research Laboratory Final Rep., Vol. 1, 106 pp.
Hong, S.-Y., Noh Y. , and Dudhia J. , 2006: A new vertical diffusion package with an explicit treatment of entrainment processes. Mon. Wea. Rev., 134, 2318–2341.
Hoskins, B. J., and Ambrizzi T. , 1993: Rossby wave propagation on a realistic longitudinally varying flow. J. Atmos. Sci., 50, 1661–1671.
Janjić, Z. I., 1994: The step-mountain eta coordinate model: Further developments of the convection, viscous sublayer, and turbulence closure schemes. Mon. Wea. Rev., 122, 927–945.
Janjić, Z. I., 2000: Comments on “Development and evaluation of a convection scheme for use in climate models,”. J. Atmos. Sci., 57, 3686
Jayawardena, S., 2007: Analyses of an unusual heavy rainfall period over the Hawaiian Islands during 19 February–2 April, 2006. M.S. thesis, Dept. of Meteorology, University of Hawaii at Manoa, 186 pp.
Jiang, H., Halverson J. B. , Simpson J. , and Zipser E. J. , 2008a: On the differences in storm rainfall from Hurricanes Isidore and Lili. Part II: Water budget. Wea. Forecasting, 23, 44–61.
Jiang, H., Halverson J. B. , and Zipser E. J. , 2008b: Influence of environmental moisture on TRMM-derived tropical cyclone precipitation over land and ocean. Geophys. Res. Lett., 35, L17806, doi:10.1029/2008GL034658.
Klazura, G. E., and Imy D. A. , 1993: A description of the initial set of analysis products available from the NEXRAD WSR-88D system. Bull. Amer. Meteor. Soc., 74, 1293–1312.
Knippertz, P., and Martin J. E. , 2007: A Pacific moisture conveyor belt and its relationship to a significant precipitation event in the semiarid southwestern United States. Wea. Forecasting, 22, 125–144.
Kodama, K., and Barnes G. M. , 1997: Heavy rain events over the south-facing slopes of Hawaii: Attendant conditions. Wea. Forecasting, 12, 347–367.
Lamarque, J.-F., and Hess P. G. , 1994: Cross-tropopause mass exchange and potential vorticity budget in a simulated tropopause folding. J. Atmos. Sci., 51, 2246–2269.
Marques, R. F. C., and Rao V. B. , 1999: A diagnosis of a long-lasting blocking event over the southeast Pacific Ocean. Mon. Wea. Rev., 127, 1761–1776.
Martin, J. E., and Marsili N. , 2002: Surface cyclolysis in the North Pacific Ocean. Part II: Piecewise potential vorticity diagnosis of a rapid cyclolysis event. Mon. Wea. Rev., 130, 1264–1281.
Martin, J. E., and Otkin J. A. , 2004: The rapid growth and decay of an extratropical cyclone over the central Pacific Ocean. Wea. Forecasting, 19, 358–376.
Miller, R. C., 1972: Notes on analysis and severe storm forecasting procedures of the Air Force Global Weather Central. Air Weather Service Tech. Rep. 200 (Rev), USAF, 190 pp.
Mlawer, E. J., Taubman S. J. , Brown P. D. , Iacono M. J. , and Clough S. A. , 1997: Radiative transfer for inhomogeneous atmosphere: RRTM, a validated correlated-k model for the longwave. J. Geophys. Res., 102 (D14), 16 663–16 682.
Morrison, I., and Businger S. , 2001: Synoptic structure and evolution of a Kona low. Wea. Forecasting, 16, 81–98.
Nash, A., Rydell N. , and Kodama K. , cited 2010: Unprecedented extended wet period across Hawaii—February 19 to April 2, 2006. [Available online at http://www.prh.noaa.gov/hnl/pages/events/weeksrain/weeksrainsummary.php.]
Otkin, J. A., and Martin J. E. , 2004: The large-scale modulation of subtropical cyclogenesis in the central and eastern Pacific Ocean. Mon. Wea. Rev., 132, 1813–1828.
Ralph, F. M., Neiman P. J. , and Wick G. A. , 2004: Satellite and CALJET aircraft observations of atmospheric rivers over the eastern North Pacific Ocean during the winter of 1997/98. Mon. Wea. Rev., 132, 1721–1745.
Rex, D. F., 1950: Blocking action in the middle troposphere and its effects upon regional climate. II: The climatology of blocking action. Tellus, 2, 275–301.
Rogers, E., Black T. , Ferrier B. , Lin Y. , Parrish D. , and DiMego G. , 2001: Changes to the NCEP Meso Eta Analysis and Forecast System: Increase in resolution, new cloud microphysics, modified precipitation assimilation, modified 3DVAR analysis. NWS Tech. Procedures Bull. 488, 15 pp. [Available online at http://www.emc.ncep.noaa.gov/mmb/mmbpll/eta12tpb/; also available from National Weather Service, Office of Meteorology, 1325 East-West Highway, Silver Spring, MD 20910.]
Rossby, C. G., 1940: Planetary flow patterns in the atmosphere. Quart. J. Roy. Meteor. Soc., 66 (Suppl.), 68–87.
Schroeder, T. A., 1977: Meteorological analysis of an Oahu flood. Mon. Wea. Rev., 105, 458–468.
Simpson, R. H., 1952: Evolution of the Kona storm, a subtropical cyclone. J. Meteor., 9, 24–35.
Skamarock, W. C., and Coauthors, 2008: A description of the Advanced Research WRF Version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp.
Wentz, F. J., 1997: A well-calibrated ocean algorithm for Special Sensor Microwave/Imager. J. Geophys. Res., 102, 8703–8718.
Zhang, Y., Chen Y.-L. , Hong S.-Y. , Juang H.-M. H. , and Kodama K. , 2005: Validation of the coupled NCEP Mesoscale Spectral Model and an advanced land surface model over the Hawaiian Islands. Part I: Summer trade wind conditions and a heavy rainfall event. Wea. Forecasting, 20, 847–872.
Zhu, X., Sun J. , Liu Z. , Liu Q. , and Martin J. E. , 2007: A synoptic analysis of the interannual variability of winter cyclone activity in the Aleutian low region. J. Climate, 20, 1523–1538.
The full sigma levels are 1.000, 0.981, 0.962, 0.923, 0.903, 0.882, 0.861, 0.839, 0.816, 0.793, 0.770, 0.745, 0.720, 0.694, 0.667, 0.609, 0.577, 0.544, 0.509, 0.471, 0.385, 0.333, 0.259, 0.222, 0.185, 0.148, 0.074, and 0.000.
