Development of a Tropical Cyclone Rain Forecasting Tool Frank D' Marks NOAAAOML, Hurricane Research

1
Development of a Tropical Cyclone Rain
Forecasting ToolFrank D. MarksNOAA/AOML,
Hurricane Research Division, Miami, FL 33149

• QPE Techniques in TCs
• TRMM and gauges
• QPF Techniques in TCs
• R-CLIPER
• 2002 Storms
• Where do we go from here?

Support from NOAA JHT and NASA TRMM and CAMEX
2
QPE techniques in TCs

• DATA
• TMI R estimates for 245 storms (December
  1997-2000), globally, yielding 2121 events, from
  TD to CAT5 (Lonfat et al 2003).
• 1998-2000 TMI events by Intensity
• Storm Intensity Events
• TD/TS 1361 64
• Category 1-2 548 26
• Category 3-5 212 10
• Total 2121
• 560,000 hourly gauge estimates in 46 US
  landfalling hurricanes (1948-2000) (DeMaria 2003)

3

• QPE techniques (continued)
• TMI and gauge R, 10 km annuli, PDF 1 dBR from
  0.3-300 mm h-1. Stratify by intensity and motion.

4
QPF techniques in TCs

• Develop R-CLIPER from gauge and TMI climatology
  for operational and model QPF comparisons.
• Project R-climatology along NHC forecast track.

5
R-CLIPER

• Radial distribution of R for gauge and TMI
  climatology consistent.
• TMI peak R increases with intensity 3 mm h-1
  TD/TS 7.2 mm h-1 CAT 1-2 12.5 mm h-1 CAT 3-5.
• TMI peak R radius decreases with storm intensity.

6
R-CLIPER Cases
7.7
15.9
4.8
color contours denote Rtotal (inches, peak at
landfall listed)
Position and intensity from best track (6 h)
8.6
4.1
7
R-CLIPER Cases
Danny 97

• R-CLIPER underestimates area of Rtotal by factor
  of 2 using CDF.

of gauges
PMM R-CLIPER/Gauge Rtotal
90
90
Andrew 92
Allison 01
50
50
of area
20
Floyd 99
8
QPF in Landfalling TCs

• For 2002 season ran 584 forecasts in 32 storms in
  the ATL, EPAC, and CPAC for both the gauge and
  TRMM R-CLIPER.

9
QPF in Landfalling TCs
Gauge
TRMM
10
Where do we go from here?

• R-CLIPER ran operationally in 2002 at NHC.
• Provides benchmark for evaluation of other QPF
  techniques.
• Develop data products for hurricane specialists.
• Compare R-CLIPER forecasts to models and 6-h
  average rainfall amounts on HPC 1X1 grid.
• Work with HPC to provide track guidance after
  landfall.

11
Comparison with Models
Rogers et al 2002

• RCLIPER Rmax along track, with a probability of
  heavy rain right or left of track.
• RCLIPER Rmax increases when storm slows or turns.
• Model shows symmetry and asymmetry.

RCLIPER
12
QPF in Landfalling TCs

• ISSUES
• Are we providing our users what they need?
  Different users have different needs.
• How do we develop an evaluation/validation system
  for high-resolution QPF product?
• How will it be used operationally
  (probabilistic)?
• NEEDS
• New products developed for forecasters and
  hydrologists.
• New display products to convey uncertainty to
  forecaster and others.

13
Why TC QPF is important

• TCs pose significant QPF problem. i.e.,
  Hurricanes Mitch (1998), Floyd (1999), and TS
  Allison (2001).
• Last 30 years majority of TC deaths caused by
  flooding (Rappaport 2000).
• Flooding is a function of R and duration.
• Enhanced rainfall from orography and fronts.
• QPF limited by complexity of precipitation
  processes and lack of microphysics data.
• Improved TC QPF challenge of USWRP.

14
QPE techniques in TCs

• STRENGTHS
• TMI and PR. Strength global coverage with single
  instrument.
• Hourly gauges and WSR-88D. Strength temporal
  resolution over long time.
• NEED
• Develop R climatology in TCs regionally and
  globally
• Develop methodology to validate NWP forecasts.
• GOAL
• Understand and describe TC rain using
  climatology.
• Improve TC rain QPF.

15
QPE techniques in TCs

• TMI, radar and gauge compared to R probability
  distributions by Miller (1958) and Frank (1977).

16
QPE Techniques in TCs

• Scale dependence
• 10-s PMS sample area 1 m-2
• 1-h gauge sample area 1-16 km2 (wind speed
  dependent)
• 1-h radar sample area 16 km2
• TMI sample are 25 km2
• PDF narrower and skewed toward lower R with
  increasing scale

17
QPF techniques in TCs

• Simplest TC QPF predicts Rmax
• RmaxRDVs-1
• Krafts "rule of thumb
• Rmax130.8 Vs-1
• Rmaxcm, Vsm s-1, R0.49 cm h-1 D103 km.
• Tropical Rainfall Potential (TRaP) uses satellite
  R-estimates
• No adjustments for intensity, topography, etc.

18
R-CLIPER Cases
13.5
7.7
4.2
6.7
10
31