COLLABORATION in the IFPS Era

1
COLLABORATIONin the IFPS Era
2
Outline

• Reasons for collaboration
• Influences on collaboration
• Tools
• 12Planet
• ISC
• Collaboration dos and donts
• Collaboration experiences during NDFD prototype
• Unresolved issues

3
Collaboration vs.Coordination

• Collaboration - exchanging ideas to reach a
  common forecast solution
• Coordination - notifying surrounding offices of
  your forecast

4
My forecast vs. NWS forecast
5
How Good is Good Enough?Initial Collaboration
Triggers

• MaxT/MinT/T/Td - 5 deg (7 deg in complex terrain)
• PoP - 20
• Sky - 25 (40 in complex terrain)
• Wind - 10 kts (15 kts in complex terrain)
• QPF - 0.25/6 hr
• Snow - 2 inches (only for homogenous situations)
• Free Air Wind - 15 kts
• Snow Level - 1000 ft
• Wx - Liquid/freezing/frozen (cannot check in ISC
  Discrep)
• http//www-md.fsl.noaa.gov/IFPS/ISC.htm

6
Influences on Collaboration

• Schedules/duties
• Order and method of weather element production
• Forecaster philosophy
• Office philosophy
• The forecasters knowledge of meteorology and
  his/her ability to communicate it

7
Schedules/Duties

• How an office breaks down duties matters
• Routine/Update vs Short/Long term vs
  Public/Aviation
• Adding new day 7 deadline of 1745Z
• Doing days 4-7 at same time is efficient
• Pertinent model data available
• If ensembles later show a change in forecast is
  needed, then collaborate on those aspects
• Need to start at about the same time for most
  effective collaboration
• Everyone working on a subset of entire database

8
An Extended Forecast Timeline

• 13-16z Collaborate major issues and edit grids
• HPC currently experimenting with preliminary
  extended discussion
• 16-17z ISC grid comparisons and final tweaking
• Subject to change, based on availability of
  guidance and customer demands

9
Weather Element Productionand the Impact on
Collaboration

• Edit grids to optimize meteorological reasoning
  so they are internally consistent
• PoP ? Wx ? Sky ???
• Common smart tools facilitate collaboration
• Using diurnal curve tools for T from MaxT/MinT,
  weather from pop, etc.
• Use lapse rate tools to account for terrain

10
Philosophy

• The emphasis an office places on a well
  collaborated forecast makes a difference
• Management team leads the way
• Attitude towards collaboration can differ from
  forecaster to forecaster in an office
• More work for other forecasters - if one
  forecaster does not collaborate then others have
  to fix forecast to be well collaborated
• The forecast style (aggressive or conservative)
  of each forecaster will impact collaboration

11
Tools

• 12Planet
• Intersite Coordination of Grids (ISC)
• Telephone

12
12Planet

• Each office has standardized IDs
• WFOs are XXX_short, XXX_long, XXX_3, XXX_4
• HPCs are hpc_MDLDIAG, hpc_MEDR, hpc_QPF1,
  hpc_QPF2
• RFCs are XXX_1, XXX_2, XXX_3
• SPCs are
• Can block all but your surrounding offices and
  NCEP
• Can talk to all on your subscribe list at once as
  well as just a select few (whisper/private chat)

13
12Planet

• You can set up a word list on which to receive
  audio alarms
• By choosing _ as a word, you will receive an
  audio alarm on each message posted by someone in
  your subscribe list
• You can set up a temporary chat room to talk to a
  few offices or for a regional meeting

Help can be accessed by clicking then Help
while in 12Planet or at http//www-md.fsl.noaa.gov
/IFPS/12p/12Planet.htm
14
12Planet is just the beginning

• Provides initial method for widespread
  collaboration of meteorological thought
• Collaboration strategies and chat tools are sure
  to evolve
• May eventually find a method to enhance or
  replace chat capability

15
Chat Advantages

• Efficient and less intrusive to forecast process
• Focuses discussion on one topic at a time
• Encourages conciseness
• Allows everyone to use same mode of communication
• Facilitates exchange of all viewpoints
• Less potential for personalities to dominate
• Allows earlier collaboration of potential
  headlines

16
Chat Advantages cont

• Allows for continuous sharing of information
• Quicker notification of changes in meteorological
  thinking
• Can also relay technical issues or severe weather
  reports
• Opportunity for science sharing
• Increased interaction with surrounding offices
  and NCEP
• Other uses of posts
• Basis of AFD
• Shift briefing
• Subsequent forecasters can review evolution of
  forecast

17
ISC Grid Collaboration
18
Ways to Send ISC
Send your grids Early and Often - Dont wait
until all grids have been edited! Also do not
send grids you intend to edit again

• You can choose to
• send ISC grids via the
• Save Dialog This should
• be the default

OR
Send ISC Grid option from Right Mouse Button over
a particular grid
19
Send ISC Dialog

• Select Weather Elements you wish to send
• Select the Time Period
• Choose sites you wish to send to
• So if you edited MaxT for Day 3 on the eastern
  boundary, you can just send that grid to the WFO
  that borders on the east

Can choose groups such as Public, Fire, etc.
20
Intersite Coordination Mode

• How to find where forecast differences exceed the
  collaboration thresholds

You can set up a sample point set which will
allow forecasters to get a quick look at whether
ISC thresholds are being met
21
ISC Discrepancy Tools

• There are three GFE tools to help identify
  discrepancies
• Show_ISC_Highlights -- highlights grids that are
  discrepant. This is the tool called by the
  ISC_Discrepancies Procedure
• Show_ISC_Grid -- creates a discrepancy grid to be
  viewed from the Grid Manager
• Show_ISC_Area -- creates an edit area identifying
  where discrepancies were found

Complex Terrain Discrepancy Grid
22
Generating Discrepancy Grids
Via the Consistency Menu
Choose the elements and times to check
Grids with points exceeding the thresholds are
highlighted
23
Generating Discrepancy Grids
Show_ISC_Area
Creates edit areas where thresholds are exceeded
24
Generating Discrepancy Grids
Show_ISC_Grid
Creates grid showing differences between your
forecast at that point and your neighbors Those
exceeding the threshold are highlighted
25
Collaboration Dos

• Collaborate potentially major issues before your
  mind is made up
• Collaborate significant changes to going forecast
  and any potential headlines
• Work to get everyone on board this is one time
  when peer pressure is a good thing!
• Focus on the meteorology when chatting
• Indicate why are you leaning in a particular
  forecast direction
• Especially important in extreme events
• Concentrate on meteorology, not zones or words
• Minimize use of edit areas that conform to
  geopolitical boundaries

26
Collaboration Donts

• No need to chat about minor tweaks if you are
  going to be within tolerances/thresholds
• Open ended questions are often not as effective,
  need to be specific
• SEW - concerned abt shrtwv on tues. May need
  pops. Whats your thoughts?
• OUN/TUL - Eta a lot drier than GFS w/H85 mstr
  tngt? Which way are you leaning?
• Dont neglect the chat that frustrates
  surrounding offices
• Dont go with a particular forecast solution
  without backing it up (e.g. blindly going with
  model guidance)

27
What if You Do Not Agree?How to Work Around a
Potential Conflict

• If after stating your science and experience you
  still cannot come to a consensus, then try these
  workarounds
• Agree to wait for more data which helps to
  clarify the meteorological situation
• For the time being, agree to keep any changes
  closest to what is currently forecast until new
  data can support a particular solution
• Have an impartial third office or HPC break the
  tie
• If this is a recurring issue (e.g. temps on a
  particular boundary), conduct a local study or
  have a multi-office meeting on the matter

28
Collaboration experiences during NDFD prototype
29
Collaboration Successes During the NDFD Prototype

• Summer 2002 during significant cold frontal
  passage, HPC provided specific ECMWF sfc data
• HPC occasionally provided web addresses to
  datasets they were using along with additional
  explanation
• Collaboration makes offices focus on edges of CWA
  which can improve service to those areas
• Several occasions when ICT CWA corner forecasts
  were improved via ISC grids
• RLX forecasters benefited from RNK forecasters
  expertise in a border valley which improved
  service to that area

30
Problems Encountered During the NDFD Prototype

• Numerous cases where offices
• Waited until the last minute to send any ISC
  grids
• Werent in chat room or never responded to
  specific inquiries
• Forecasters who insist on forecasting a seemingly
  small chance event, refuse to give some to
  consensus yet they failed to justify their
  solution well enough to persuade others
• Seen particularly in the medium range
• Forecasters who wont stray from guidance

31
More Problems Encountered During the NDFD
Prototype

• Cases where HPC
• Medium range models were stratified into two
  clusters yet when asked which one they favored,
  HPC responded, Both solutions are equally
  likely
• Would be better to know their gut feeling as to
  which would verify
• Ignored several WFOs who differed on
  location/timing of nocturnal convection in Plains

32
(No Transcript)
33
Evolution of CollaborationChat room excerpts one
year later

• WFO SGF - with regard to the upcoming system wed
  nt into thursday...given the eta trend to sink
  500 mb height field....layer rh field...and vort
  track about 150 miles south over the last 2
  runs...think that a winter storm watch may be
  needed over the northern 2/3 of my cwa. initial
  proposed boundaries are in wwa
• tsa_longTerm - Looking with each successive model
  run that impact will be greater over ne oklahoma
  and nw arkansas wed night as well...but
  preliminary thoughts would be low end advisory
  amounts along ks border. Will have to wait and
  see.
• WFO SGF - if in fact we do get several inches of
  snowpack...fridays high temps may need to be
  trimmed back substantially as waa not really
  kicking in yet
• eax-1 - all prelim grids have been sent. lowered
  max temps quite a bit friday for areas which
  should receive significant snow accumulations.
• ICT (Routine) - 18z mesoeta is even stronger with
  isentropic lift...this fits the pattern of all
  models continuing to deepen the system every run.
  will extend the watch a bit more to the west
• Topeka Long Term - Is anyone concerned with the
  clipper system progged to come across Iowa on
  Saturday? Models are fairly consistent with the
  short wave, but again moisture is the wild card.
• ICT (Routine) - despite impressive forcing/lift,
  the antecedent dry air should preclude sgfnt qpf
  til 00z in sern ks.
• Topeka Long Term - Was thinking that an advisory
  would be warrented in east central zones. Am not
  excited about warning criteria being met. The
  best lift is used to saturate the airmass and
  then scoots east pretty quickly with the best
  accumulation occurring between 00z and 08z.
• ICT (Routine) - the limiting/troubling factor is
  wedge of drier air in the low lvls that seems to
  get wrapped up into sys acrs parts of the cwa.
• eax-4 - quick look at 12Z initialization of ETA
  makes me think the ETA may be a little too far
  south with position of 500mb low, based on WV
  imagery and profiler data.

34
The New Forecast MethodologyCollaboration
Smoothes the Way

• Only make changes where needed
• Examine new information as though current
  forecast is good, until majority of data suggests
  otherwise
• Enables forecasters to come to a synoptic scale
  understanding before making major changes to
  existing grids
• Helps to avoid potential problems and wasted time
• Allows forecasters to share scientific techniques
  
• Use ISC to identify any boundary issues
• Convergence toward a solution is better than
  yo-yo effect

35
Typical Forecast Shift

• Shift briefing should mention potential problems
  (meteorology and grids)
• If substantial changes could be needed, reach
  consensus on large scale evolution first
• Edit grids and share via ISC
• Collaborate to resolve finer scale details

36
Ponder thisAs Yet Unanswered Questions

• How do you do a rolling database over multiple
  offices?
• Do you forecast/chase thunderstorm clusters and
  outflow boundaries and their effect on
  pops/temps/winds/sky etc.?
• At what point do you update?
• What is consensus according to our customers
  and partners?

37
Things to Keep in Mind

• Extreme events continue to be the most important
• Lets work together to nail them
• Is latest model run always the best choice if
  large changes are indicated
• Maybe trends are better
• Consensus forecast has superior accuracy? Yes for
  long term average, not specific forecasts
• Consensus is not the best forecast, but the
  least wrong over the long haul C. Doswell
• But it is still important to have meteorological
  consistency as it affects NWS credibility
• So to gain credibility as an agency, we must
• reach consensus with grids

38
Which Forecasts Are Most Credible?Imagine
Yourself as a Customer
39
Final ThoughtsKeys for Successful Collaboration

• Discussion needs to begin early in the forecast
  process as new information (i.e. model data)
  arrives
• Interaction needs to continue throughout the
  forecast process
• Focus on the meteorology
• In both chat and grids
• Consensus is critical