Exam 1 Flashcards
(31 cards)
Describe the temperature distribution during the middle of the day in florida
Due to the Bermuda high temperature contours are shifted to western parts of the state
Which models are global
GFS, ECMWF, GALWEM, HAFS-A?
Which models are regional
NAM, RAP, HRRR
What is the future for US modeling
The Unified Forecast System
How to deal with uncertainty
Perturbed (Moved) initial conditions which would lead to a range of outcome
Can also perturb the physics of a model
What do ensembles attempt to do
Cover a range of forecasts
What does the ensembles attempt do for us
- Allows us to estimate the predictability for an event
- Good argument among members usually implies high predictability
- Regardless of errors in initial conditions you can still get a valuable forecast
Ensemble bias
the model is just systematically too cold or too wet
Underdispersion
Estimate of initial conditions errors tend to be too small and the real atmosphere may behave in ways completely outside the realm of our physics equations. Ensembles don’t have enough spread
Cone of uncertainty
Average of errors over the last 5 years that remains the same all year
Machine Learning
Parameters are fit on a training data set in order to optimize a predefined loss function
What is the most common loss function for traditional linear regression
Residual summed squared error (RSS)
What are all ML essentially built on and is the simplest of models
Linear regression
In order to have a good forecast…
- Be consistent with users prior knowledge
- Have good quality (accuracy)
- Be valuable (provide benefit)
Neutral Networks
Are deep webs of information transfer.
Research modeling and its problems
-operational models, determine what improvements can be made
Problems: they are “tuned” and it’s difficult to do ‘true’ tests
Research models
WRF, Cloud Model 1 CM1
What can’t AI capture at the moment
Insintric predictabiltiy of the atmosphere: slightly different conditions may lead to the same “result” which is not physically possible
What are the limits of NWP
- Equations can’t be solved exaclty
- Cannot measure the entire atmosphere
- Data has measurement errors
- Certain physical processes occur at scales below model resolution
- The true state of the atmosphere is never known
Intrinsic Predictability
The ability to predict given nearly perfect representation of the dynamical system (by a forecast model) and nearly perfect initial boundary conditions, an inherent limit due to the chaotic nature of the atmosphere and cannot be extended by any means -> 2 weeks
Practical Predictability
Same as forecast skill, the ability to predict given realistic uncertainties in both the forecast model and initial boundary conditions
Anomaly correlation coefficient
Measures the correlation between forecast and actual observed deviations from climatology
Errors in deterministic forecast accumulate _______
Over time
Boundary conditions
How we handle the first and last grid points near our grid
