how to win “forecaster of the year” or how to be more like me. if you really want to. really...
TRANSCRIPT
How to win “Forecaster of the How to win “Forecaster of the Year”Year”
OrOrHow to be more like me.How to be more like me.
If you really want to.If you really want to.REALLY want to.REALLY want to.
What is forecasting?What is forecasting?
AnalysisAnalysis DiagnosisDiagnosis PrognosisPrognosis
What is What is notnot forecasting? forecasting?
Rules of thumbRules of thumb Following the model blindlyFollowing the model blindly
– ““Meteorological cancer”Meteorological cancer”
FeelingsFeelings
So what do we do?So what do we do?
Forecast (in the mind, on a screen, or on Forecast (in the mind, on a screen, or on paper) the motion of large-scale featurespaper) the motion of large-scale features– LowsLows– HighsHighs– FrontsFronts– Vorticity centresVorticity centres– MoistureMoisture– Jet streams / streaksJet streams / streaks
Keeping in mindKeeping in mind
Geography / climatology of the areaGeography / climatology of the area– Valleys, upslope areasValleys, upslope areas– TreelineTreeline
Recent events and surface conditionsRecent events and surface conditions– SnowpackSnowpack– Recent rainfallsRecent rainfalls
Now we have a handle on the Now we have a handle on the large-scale featureslarge-scale features
Let’s get down to the nitty grittyLet’s get down to the nitty gritty(but not the dirt band)(but not the dirt band)
Temperature forecastingTemperature forecasting
Think process – what affects the Think process – what affects the temperature?temperature?– Cloud cover / sunshineCloud cover / sunshine– PrecipitationPrecipitation– AdvectionAdvection
What works for temperatureWhat works for temperature
Analysis (tephis, surface, upper plots)Analysis (tephis, surface, upper plots) Gerlyand curvesGerlyand curves Temperature trendsTemperature trends 850 850 ΔΔT/T/ΔΔt superimposed on yesterday’s t superimposed on yesterday’s
valuesvalues BarthrumBarthrum Synoptic correlationSynoptic correlation
Cloud cover forecastingCloud cover forecasting
Once more, think process – what causes Once more, think process – what causes cloud?cloud?– Cloud shows ascent – so the focus is on what is Cloud shows ascent – so the focus is on what is
causing that ascentcausing that ascent FrontsFronts TroughsTroughs Vorticity centresVorticity centres Outflow boundariesOutflow boundaries Jet streaksJet streaks
What works for sky conditionWhat works for sky condition
Analysis (tephis, surface, upper plots)Analysis (tephis, surface, upper plots) PersistencePersistence Satellite images (visible, 4 Satellite images (visible, 4 μμm, blended)m, blended) Water vapour imagesWater vapour images
Precipitation forecastingPrecipitation forecasting
This is the trickiest, but it boils down to the This is the trickiest, but it boils down to the same thing – think processsame thing – think process
Precipitation tells you there is significant lift Precipitation tells you there is significant lift occurring along with ample moistureoccurring along with ample moisture– So this is essentially the same as for cloud So this is essentially the same as for cloud
cover forecastingcover forecasting
What works for precipitationWhat works for precipitation
Analysis (tephis, surface, upper plots)Analysis (tephis, surface, upper plots) RADARRADAR Satellite imagerySatellite imagery
– Looking for convectionLooking for convection– Looking for vort centresLooking for vort centres– Looking for subsidenceLooking for subsidence
What works for windWhat works for wind
SCRIBESCRIBE– Also analysis (tephis, surface, upper plots)Also analysis (tephis, surface, upper plots)
How do I put it all together?How do I put it all together?
3 primary steps3 primary steps– AnalysisAnalysis– DiagnosisDiagnosis– PrognosisPrognosis
I end up with a forecastI end up with a forecast
And thenAnd then
AfterAfter I have my forecast I have my forecast
Shocker!Shocker!
I look at what the model saysI look at what the model says
I compare SCRIBE to my forecastI compare SCRIBE to my forecast
I look for areas where the model is known I look for areas where the model is known not to fare wellnot to fare well– Boundary layer cloudBoundary layer cloud– Convective precipitationConvective precipitation– Strongly curved flowStrongly curved flow
And make changes accordinglyAnd make changes accordingly
Keep in mindKeep in mind
This important techniqueThis important technique
BE LAZYBE LAZY
(when manipulating SCRIBE)(when manipulating SCRIBE)
What do you mean, be lazy?What do you mean, be lazy?
Verification has shown us thatVerification has shown us that– When we make small changes, we’re right When we make small changes, we’re right
about 50% of the timeabout 50% of the time– When we make big changes, we usually don’t When we make big changes, we usually don’t
make a make a big enoughbig enough change change
So……………..So……………..
““If I’ve ever been sure of anything If I’ve ever been sure of anything in my life, I’m sure of this”in my life, I’m sure of this”
(This change will be a good one)(This change will be a good one)
Only make big changesOnly make big changes
What is a big change?What is a big change?
Temperature changes of Temperature changes of at leastat least 2 degrees 2 degrees Sky condition changes of Sky condition changes of at leastat least 2 2
conditionsconditions Wind changes of Wind changes of at leastat least 20 km/h and 90 20 km/h and 90
degreesdegrees– And then onlyAnd then only
When it’s windy and SCRIBE calls for calm, and vice-When it’s windy and SCRIBE calls for calm, and vice-versaversa
In highly curved flowIn highly curved flow
What is a big change, continuedWhat is a big change, continued
PrecipitationPrecipitation– By 1 POP categoryBy 1 POP category– By 1 phaseBy 1 phase
Points to rememberPoints to remember
Analysis, diagnosis, prognosisAnalysis, diagnosis, prognosis Make up your forecast Make up your forecast beforebefore checking checking
SCRIBESCRIBE Be lazy – only make significant changesBe lazy – only make significant changes