Verification of Icing and Turbulence Forecasts:

1
National Weather ServiceTechnology Infusion
Plan A Science and Technology Roadmapto Weather
Services in 2025 U.S. DEPARTMENT OF
COMMERCE National Oceanic and Atmospheric
Administration National Weather Service Office of
Science and Technology Silver Spring,
Maryland October 2001
Overview of Forecasting Techniques Product
Methodologies Envisioned for the NWS Next and
AfterNext Era Presented by Richard A.
Wagoner National Center for Atmospheric
Research Boulder, Colorado NWS Corporate Board
Meeting 7-9 August, 2001 Boulder, Colorado
2
Some things to ponder

• Forecast and product preparation has been
  primarily a manual process for 125 years.
• Will this change over the next 25 years?
• How will technology drive the human/machine mix?
• How will technology influence the role of NWS
  personnel?
• The 2015 option was considered 20 years ago.

3
Trends provider user perspective

• User requirements are becoming increasingly
  demandingand complex.
• Demand for higher resolution information
• County scale or 1 square mile?
• Section scale (1 square mile) for
  agriculturalists
• ½ hour time resolution instead of several hours
• Demand for geo-located information
• Demand for higher accuracy
• Demand for confidence/uncertainty information
• NWS in current human/machine mix and level of
  technology is not able to meet these new demands.
• Human workforce would be overwhelmed
• Level of automation falls far short of need to
  meet future requirements

4
Current process for producing forecasts/warnings
is a push paradigm

• Paradigm is 125 years old.
• Based on the idea that the NWS knows best what
  users want.
• User requirements are assessed but result is
  always a product suite that meets the average
  user needs.
• Technical constraints also limit the ability to
  meet all user needs.

5
Current Forecast System Architecture
Official Products
Analyses Forecasts Warnings
PostProcessing
NWP
6
NWS AfterNext System Architecture
Auto-verify Subsystem
Dynamic Weights Adjusted
Official
Human-derived Translation Algorithm(Experience)
6-D Database
Human Forecaster
( Selective intervention)
7
Why new verification approaches? An example
In all cases POD0, FAR1, CSI0
8
NWS AfterNext System Architecture
Auto-verify Subsystem
Dynamic Weights Adjusted
Official
Human-derived Translation Algorithm(Experience)
6-D Database
Human Forecaster
( Selective intervention)
9
Primary goal for NWS After-Next period (Holy
Grail)

• Create and maintain a 6-dimensional gridded
  database that specifies all relevant atmospheric
  parameters. The following characteristics define
  the first 4 dimensions
• 0.5 km space resolution
• 5 min temporal resolution
• 1.0 km space resolution
• 30 min temporal resolution
• 2.0 km space resolution
• 1 hr temporal resolution
• 4.0 km space resolution
• 6 hr temporal resolution
• Confidence metrics specified for all parameters

0 6 hrs
6 - 36 hrs
36 - 240 hrs
11 - 180 days
10
Database dimension 5 point data

• Point-oriented plane of the overall database
• Analysis forecast information for any point
  that has a period of continuous recorded data
• Retains higher fidelity made possible by local
  data collected exactly at the forecast location
• Allows dynamic tuning of the forecast variables
  by adjusting weights of data types

11
Database dimension 6 event data

• Captures data valid for moving phenomena
• Hurricane eye locations (lat, long, MSLP, MSPD,
  speed, direction)
• Boundaries (TSTM outflow, microburst, frontal)
• Centroid radius of volcanic plume
• Low and high pressure centers
• Contains primarily post-processed information
  from other planes of the 6-D database
• Can be used as triggering criteria for alerting
  systems or input to decision support systems
• Reminder Official NWS information

12
NWS AfterNext System Architecture
Auto-verify Subsystem
Dynamic Weights Adjusted
Official
Human-derived Translation Algorithm(Experience)
6-D Database
Human Forecaster
( Selective intervention)
13
New pull paradigm for forecasts warnings
build it and they will come

• New pull paradigm allows the end-user to tap
  the 6-D database and specify any set of criteria
  for analyses, forecasts and alerts (thresholds,
  update frequency, lead time, etc.).
• If advertised well in advance and coordinated
  closely with the private sector, new stand-alone
  software products and new interactive services
  will grow rapidly. This vibrant, competitive
  market will serve the Nation with rich, low-cost
  weather information over a broad spectrum of
  specialized users.
• Software development companies will compete for
  the killer weather app.
• Analogous to Navy implementation of GPS.
• NWS will tap the same database to produce
  standard products for general public use.

14
How do we get there?

• By the beginning of the NWS-Next period
• Significant verification program that provides
  quantitative parallel measurements of model,
  IFPS, and hybrid post-processing technique
  performance using advanced verification methods
  at high time/space resolutions.
• Establish several validation programs to
  determine the best combination of human, model,
  and post-processor inputs to the 6-D gridded
  database.
• Begin development of a national boundary
  detection program.

15
How do we get there?

• By the beginning of the NWS After-Next period
• Establish IOC for 6-D database.
• Initiate best estimate of combined human, model
  and post-processor system. Continue to refine and
  adjust based on verification/validation system.
• Implement IOC of national boundary detection
  system.
• Move stepwise toward long term goals for the 6-D
  database (time/space resolution, enhanced model
  output, GOES-R multi-spectral data, polarimetric
  radar, and water vapor fields).

16
The End 303-497-8404 wagoner_at_ucar.edu