1

Challenges in the Development of Ocean Coupled, Operational Hurricane Forecast Model

at National Weather Service (NWS)

​

Hyun-Sook Kim

​

​

​

Environmental Modeling Center (EMC)

National Centers for Environmental Prediction (NCEP)

National Weather Service (NWS)

NOAA

College Park, MD 20740

​

Friday April 10, 2015

2

Outline

1. Challenges
2. Tropical Cyclone
3. The Hurricane Weather Research and Forecast system (HWRF)
4. Comparisons of North Atlantic Hurricane Forecasts between HYCOM and POM coupling
5. Comparisons of Western North Pacific Typhoon Forecasts between HYCOM coupling and non-coupling
6. Introduction of Ocean Model Impact Study
7. Summary

3

Challenges

• Current operational Hurricane Weather Research and Forecast System (HWRF) coupled to POM (since 2007)
• HYCOM (experimental) to be in the operational in 2016
• Generate 5-day numerical guidance products on time, 4 times per a day, for NHC/JTWC/CPHC forecasters
• In 2015, expanding to Indian Ocean and Southern Hemisphere
• Triple nested, w/ model grid resolution, 18-6-2 km; 61 levels
• Numerical Integration step, 30 sec.
• 7 storms at maximum per cycle.

​

• Ocean Model: 1/12-degree, 42 levels

​

🡺 Computer resources (competing with 25 other operational models)

​

Goal is to produce robust and skillful model guidance.

However, conflicts exist between having right physics (including ocean) and producing on-time better guidance.

4

• Called by

Hurricane/Typhoon/Cyclone

• Characteristics
• Multi-scale processes in spatial and time, ranging from the environmental O(km)/O(decadal) to the microphysical scale O(µm)/O(second), interacting independently and/or collectively.
• The importance of air-sea interactions.

Tropical Cyclone

5

From Here

At Least What is known at PBL and Ocean Upper Layer

From Here

6

Air-Sea Interface

& From Here

Heat Budget + Processes

Momentum Budget + Processes

7

NHC Official Forecast Advisory

e.g., Hurricane Edouard (2014)

To Here

8

The Hurricane Weather Research and Forecast System (HWRF)

Ocean Component in HWRF:

North Atlantic – 3D POM since 2007

Eastern North Pacific – non-coupling before 1D POM (2011), 3D POM (2012 -)

Western North Pacific – non-coupling (2012 - )

9

HYCOM coupling vs. POM coupling

2008 and 2009 North Atlantic Hurricane Forecast Comparisons

HYCOM coupling

🡺 Reduced intensity error by ~13%, and bias by ~43%

(Kim et al. 2014)

10

Verification

HYCOM Coupling vs. Non-Coupled

2012–2013 Western North Pacific Typhoon Forecast Comparisons

(Kim et al. 2015, in review)

HYCOM Coupling

non-balance gradient winds

• air-sea interaction reconsidered
• revisit Initial Conditions for Atm. & Ocn.

11

Comparisons of Forecasts

Example: Soulik (07W) July 7 to 13, 2013

IC=2013/07/10 00Z

HYCOM coupling

• Track – similar/slightly better
• Intensity – reduced by <20 kt

12

Heat Sources

Specific Humidity (left panels) at the Lowest Model Layer and

Air-Sea Temperature Differences (right panels)

Azimuthal average from the storm center out to 300 km

Soulik (2013) IC=2013/07/10 00Z

Specific Humidity

(Latent Heat Flux)

Air-Sea Temperature Difference

(Sensible Heat Flux)

13

SST cooling O(<4^oC)

Improvement of Intensity Forecasts should include SST feedback.

Atm. 🡺 the air-sea interaction parameterization and IC.

Ocg. 🡺 IC better representing the upper structure (MLD, dT/dZ, mixing physics).

Comparisons of SST Cooling

14

The Hurricane Forecast Improvement Project (HFIP)

Ocean Model Impact Tiger Team (OMITT) 2015-2016

​

Specific Charge to the Current Tiger teams

​

Determine the benefit of coupling ocean model of various complexity to operational hurricane atmospheric model, by assessing the impacts of the ocean components on the HWRF forecasts and the sensitivity of these impacts on air-sea interface, surface flux, and atmospheric parameters.

• No change in sea surface conditions from the initial state
• Simple 1D subsurface mixing model
• Various complexities of 3D ocean models
• Impact of various fluxes on hurricane prediction such as waves, sea spray etc.

​

• Process will include:
• Agreement among team participants on:
• Models to be compared – HYCOM, POM and NCOM
• Methods to compare results – build metrics
• Storm selected – 2014: Edouard, Gonzalo, Iselle, Julio and Fengshen; 2013: Soulik; and 2012: Isaac
• Prepare a report to be presented at the 2015 HFIP annual meeting

Hyun-Sook Kim (chair) and George Halliwell (co-chair)

and

~20 scientists from 12 institutions

15

1. Comprehensive, Multi-Institutional Collective Efforts:
1. Coupled Boundary Layers/Air-Sea Transfer (CBLAST) (2004) – air-sea interaction flux at winds < 30 m/s
2. Interaction of Typhoon and Ocean Project (ITOP) & TCS-10 (2010) – air-sea interaction flux at winds ≥ 30 m/s (http://p3rd.as.ntu.edu.tw/2008/data/doc/presentations/20101104/5.4_Jim_Moore.pdf)
3. Hurricane Research Division:
• Hurricane Field Program from 1997 changed to
• Intensity Forecast EXperiment (IFEX) from 2005 –

​

• Ocean Research Field Campaign (varies year by year)

Ocean Obs. – AXBTs, SeaGliders, AXCP, AXCDP, etc

​

3. Special Program supported by the Disaster Relief Act (a.k.a. Sandy Supplemental) fund (2014 – 2015).

​

Hurricane Field Activities

16

Summary

1. Lack of
1. Research
• Air-sea parameters, in particular, are heavily tuned to non-coupling.
• HWRF IC/BC are provided from GFS whose SST is non-eddy resolving and persistent per a 24-h period (no diurnal variation).
2. Validation Activity
3. Observations for Validation
4. Computation Power as higher resolution, multi-moving nest domains, and expanding to global application.

​

• Focusing on Developing/Updating Model Each Year

Requests from NHC

• Better forecast products – Competing against implementing right physics
• Labor intense
1. Tests & Experiments for upcoming version (yearly)

e.g., O(2000 cases) were simulated, which cover the 2011-2014 seasons

• Real-time exercise in parallel w/ operational and experimental systems.

17

Animation:

Coupled HYCOM-HWRF Simulations