Introduction to various QPF Techniques

11/16/01


Click here to start


Table of Contents

Introduction to various QPF Techniques

Forecasting QPF

QPF

HOW MUCH

QUESTIONS TO ASK WHEN PREPARING A SUBJECTIVE QPF

For short range QPF (0-3h to 0-6 hour), we rely heavily on

Automated short range forecasts

Auto-nowcaster provides 0-2 hour place specific forecasts of thunderstorms

Forecast parameters combines using fuzzy logic

A FEW IDEAS TO HELP DETERMINE HOW BIG AN AREA OF RAINFALL TO FORECAST

START BY LOOKING AT SYNOPTIC SCALE (THE BIG PICTURE)

THE CURRENT GENERATION OF MODELS OFTEN HAVE PROBLEMS HANDLING MESOSCALE FEATURES

INTELLIGENT USE OF THE MODEL REQUIRES THAT THE FORECASTER

PATTERN RECOGNITION, IS THIS A MADDOX FRONTAL TYPE EVENT?

BOUNDARY LAYER WIND AND TEMPERATURE FORECAST V.T. 00Z 18 JULY

A STRONG LOW LEVEL JET IS PRESENT WITH LOTS OF MOISTURE

OOZ 18 JULY FORECASTS OF

IS THIS A GOOD QPF? DO YOU THINK THE RAINFALL IS ORIENTED CORRECTLY

REMEMBER TO LOOK FOR LOW-LEVEL BOUNDARIES. NOTICE THE THERMAL GRADIENT OVER IL

HOW DID YOU DO? THIS IS FAIRLY TYPICAL OF OUR HANDLING OF MCCS. WE OFTEN KNOW WHEN ONE WILL FORM BUT USUALLY MISS THE EXACT LOCATION OF THE HEAVIEST RAINFALL.

Initiation: moisture, instability and lifting is needed for deep convection to develop

A number of factors make the QPF associated with mesoscale convective systems very difficult

Why models have forecast problems

Understanding how the physics may impact a forecast is tough because the atmosphere is complicated and acts in a non-linear fashion. For example, whenever the parameterization for convection kicks in, it

Models that have grid spacing that is too wide to allow the explicit handling of convection

In a very unstable airmass, the eta predicts almost 5 inches of rain from “grid scale” processes and less than an inch from its convective parameterization scheme. Does this look reasonable?

Bias for various models during September 2000 for northeast

A few other things that are parameterized are:

Probabilistic QPF

Krzysztofowicz approach

STATISTICAL METHODS THAT CAN BE USED TO DEVELOP PROBABILITIES

PROBABILITIES USING STATISTICAL METHODS ARE WELL CALIBRATED ESPECIALLY FOR LOWER THRESHOLDS.

VERIFICATION BUT DIFFERENT MONTH

STATISTICS METHODS

Sometimes the MOS assumptions can be violated when a model changes

MOS format

VERY USEFUL BUT TREND TOWARDS CLIMATOLOGY AT LONGER RANGES

HPC MEDIUM RANGE PREDICTIONS OF POPS OFTEN IMPROVES ON MOS POPS

Ensemble forecasting

THERE IS ALWAYS UNCERTAINTY IN THE INITIAL CONDITIONS

Importance of initial analysis 12Z Jan 24 Eta 250 mb analysis of speed (color fill) and observed RAOB speed. Note the 62 m/s observation at ATL where the initial analysis thought the speed was 30 m/s

500 Forecasts valid 12Z 25 Jan. 2000 (purple), Observed (orange)

PPT Slide

PPT Slide

PPT Slide

PPT Slide

Eta spread for “Surprise” snowstorm

Total 12h precip (in.) (00Z-12 Jan 25 from 12Z Jan 24

PPT Slide

Available on EMC homepage. Look for SREF. Probabilities are uncalibrated!!

Can show where at least 60% of the members exceeded a threshold. Again giving you some idea of the probability of a threshold assuming the model is unbiased

PPT Slide

ARPS 3km Forecast - AR Tornadoes

6 February 1999 - Bust!

Forecaster Challenges: Intelligent Use of non-hydrostatic Model Guidance

Because higher amounts or thresholds are relatively rare, it is difficult getting a big enough sample to calibrate forecasts using traditional statistical techniques. Verification of 24 hour QPF for various thresholds

The probability of 1” in 6 hours (heavy rainfall) is low (from Charba 1985).

NOTE LOGARITHMIC DECAY OF RAINFALL RATES

The HPC bias for light amounts is a little too high, the bias for .50” or greater in any 6 hour period is too low

Accuracy decreases rapidly as threshold increases and as the scale of the event decreases

Then how do we approach forecasting extreme rainfall , Probabilistically?

So how do I predict subjectively quantitative precipitation. There is no one magic method. What works for light amounts may not work for heavier ones

Statistical methods for predicting QPF References

Statistical methods for predicting QPF References (continued)

References that apply to QPF

References that apply to QPF

Author: Norman Junker

Email: njunker@ncep.noaa.gov

Home Page: http://www.hpc.ncep.noaa.gov/research/res2.html

Download presentation source