FLASH FLOOD PREDICTION IN
THE TROPICAL PACIFIC:

a University of Hawaii and National Weather Service Research Collaboration

Steven Businger, Yi-Leng Chen, Thomas R. Birchard, Jr.,Kevin R. Kodama, John Porter, Jian-Jian Wang(UH);
Paul A. Jendrowski, Roger V. Pierce, James C. Weyman, Robert W. Farrell (NWS)


Introduction

The synoptic conditions conducive to heavy precipitation in the tropical Pacific are broadly understood (Figure 1). In the vicinity of the state of Hawaii the marine boundary layer generally has adequate precipitable water for heavy rains at any time. Heavy rainfall events usually occur in the presence of upper-level forcing and an increase in mid-level moisture. The presence of moisture in the midtroposphere is generally an indication of the existence of large-scale ascending motion associated with synoptic-scale or subsynoptic scale circulations.

The factors that contribute to the flash flood hazard on tropical Pacific Islands include (1) orographic influences - flooding in Hawaii often occurs when convective cells, triggered or enhanced by orographic effects, become anchored to the mountains, and (2) small watersheds - peak streamflow typically occurs less than one hour after peak rainfall, thus respose time is short. Since the response time for hazardous flood events is very short, research concentrates on the mesoscale/nowcasting aspects of the problem.

The goals of the collaborative research effort are to:

  1. optimize the utility of the new observing systems as data resources for flood forecasting,
  2. develop a high resolution mesoscale numerical model capability, with regional data enhancement,
  3. conduct case studies and climotological analyses of flood events and develop better conceptual models on the tropical weather systems that produce flooding,
  4. use NWS and GPS data streams to develop diagnostic forecasting tools and algorithms that focus on the mesoscale/nowcasting aspect of flood forecasting, and
  5. promote the exchange of information and understanding concerning flash flood forecasting in the tropics through journal publications, seminars, workshops, and informal exchange.

1. Towards optimizing the utility of the new observing systems

The Pacific Region faces special forecasting challenges that are consequence of geography and a lack of observational data. Improved observations os the mesoscale distribution of atmospheric moisture is a key to better forecasting heavy precipitation. Sources for these moisture data include (1a) GOES-9 satellite, (1b) WSR-88D radar,, (1c) geodetic GPS recievers, and (1d) telemetered rain gauges (Figure 2 and Figure 6).

(1a) Deriving integrated water vapor from the GOES-9 satellite.
The GOES-9 satellite has IR channels at 11 and 12 um that provide the opportunity to derive the total column water vapor wmounts. Figure 3 shows a day with southwesterly synoptic-scale flow at low levels at the time of a bow echo development near Kauai. Integrated water vapor measurements from soundings and GPS will be used to valibrate/calibrate PW measurements made by the new GOES-9 algorithms.

(1b) Improving application of the WSR-88D in the tropics.
Since installation the WSR-88D radars have documented a plethora of mesoscale severe weather signatures. WSR-88D reflectivity and winds data (Figures 4a and b) led to issuance of the the first severe weather warnings in the history of the Honolulu WSFO. Shortly afterwards winds of 80 knots were recorded at Nawiliwili, on the south shore of Kauai.

Rainfall estimation from radar reflectivity is dependant on the particulars of the drop size distribution, which in turn varies with air mass and climate. Therefore, it is not surprising that the performance of the WSR-88D 's rainfall algorithms has been lacking in the tropical Pacific (Figure 5). To construct appropriate rainfall algorithms for the contrasting weather conditions in the tropical Pacific (deep convection vs. shallow warm rain), 15-minute data telemetered from Limited Automatic Remote Collection (LARC) rain gauges are compared with computed 1km values from radar algorithms. The resulting improved radar_derived rainfal estimates will be used to produce storm totals (Figure 6) and to validate precipitation predictions made by our regional mesoscale numerical model and as input for radar-based forecasting tools such as AMBER discussed below.

(1c) Precipitable water Data from Global Positioning System (GPS) receivers based in Hawaii.
L-Band raido signals transmitted by GPS satellites are delayed (refracted) by atmospheric water vapor as they propagate to ground-based GPS receivers. This 'wet delat' is nearly proportional to the quantity of water vapor integrated along the signal path. The all-weather ability of GPS to accurately measure integrated water vapor has been demonstrated. A network of continuously operating GPS recievers, including approximately 25 receivers outfitted with surface barometers, is being constructed across the state of Hawaii for a combination of geodetic, navigational, and meteorological purposes (Figure 2). Precipitable water data from these sites will be used in weather analysis and modeling applications.

(1d) Telemetered rain gauges and basin data.
There are 100 LARC telemetered rain gauges in Hawaii that serve as a verification and multi-use data source. High resolution geographic informations systems data, such as slope angle, aspect, soil and vegetative cover, are being collected to provide more intelligent flood inundation information to the forecasters and end users.


2. Development of the regional spectral model (RSM) as a regional forecast tool

Presently, operational forecasters in Hawaii use the global spectral model (AVN) from the National Center for Environmental Prediction (NCEP) as the primary model guidance for 12- to 72-hour forecasts. The current implementation of the AVN, with a 1-degree grid resolution, provides inadequate forecast guidance in Hawaii; the mountainous terrain of the Hawaiian Islands is not resolved bythe AVN. The inadequacy of the AVN guidance is particularly apparent for mesoscale tropical systems that form near or over the islands.

To address the shortcomings of the AVN, a hydrostatic version of the regional spectral model (RSM), developed at NCEP, is currently being run at UH with a synoptic domain (25 km resolution) and a nested domain (10 km resolution) covering the main islands of the Hawaiian chain. Sample output from the 25 and 10 km funs of the RSM are shown in Figures 7a and 7b, respectively, for the case in which a bow echo formed off Kauai (Figure 4).

RSM data enhancement
Due to the need for mesoscale observational data over the central Pacific Ocean, wind profiles and precipitable water data from GOES-9 satellite and WSR-88D radar and precipitable water from the Earth-based GPS receiver network will be used to enhance the initial state of the RSM.


3. Case studies and climatologies

Research opportunities based on archived data sets have been identified and are being pursued by teams of UH faculty, graduate students and NWS staff. These inslude cast studies and climatologies of heavy precipitation events, severe weather events, andf wind storms. This poster depicts results from two cases under investigation; the bow echo case of 3 November 1995 (Figure 3, Figure 4 and Figure 7) and the flood event of 25 January 1996 (Figure 5, Figure 6 and Figure 8).


4. Development of forecasting tools and algorithms

A set of software applications is being developed that is a complementary to and compatible with the general purpose interactive system and the NWS. These tools give forecasters critical assistance in digesting the voluminous mulitple data streams necessary for the rapid decision-making in flood forecasting. An example of such a tool is the Areal Mean Basin Estimated Rainfall (AMBER) program (Figure 8). AMBER uses high resolution radar reflectivity to compute average basin rainfall estimates for hydrologic basins that range in scale from large scale areas used for river forecasting to very small basins defined specifically for flash flood forecasting covering small streams and urban areas. High resolution stream basins have been defined for use with AMBER for the islands of Hawaii with WSR-88D coverage.


5. Dissemination

The information above was originally published in a poster format. Copies of this poster, additional information, and references are available on request from Profesor Steven Businger. He can be reached at businger@soest.hawaii.edu or at:

Dr. Steven Businger
Department of Meteorology
University of Hawaii
2525 Correa Road
Honolulu, HI 96822

Go Back to Steve's Homepage


Original poster graphics and layout by Brooks G. Bays, Jr. (illust@soest.hawaii.edu)
Html formatting by Ray Tanabe

Last Updated 14 October 1997