CBM Numerical Weather Prediction and Forecast Models 2 — Questions and Answers
Question 1: Model Output Statistics (MOS) improve upon raw model output primarily by:
- Running the model at higher resolution in real time
- Applying statistical relationships derived from historical model errors to correct systematic biases (Correct answer)
- Replacing numerical forecasts with climatological averages
- Using only the most recent 6-hour model run
Correct answer: Applying statistical relationships derived from historical model errors to correct systematic biases
MOS uses regression equations built from historical pairs of model output and observed weather to correct known systematic biases in raw model data.
Question 2: The High-Resolution Rapid Refresh (HRRR) model is particularly valuable to broadcast meteorologists because it:
- Provides 16-day global forecasts at 3 km resolution
- Runs hourly with 3 km resolution and explicitly resolves convection across the CONUS (Correct answer)
- Is the only model that ingests lightning data
- Offers the longest ensemble forecast period of any operational model
Correct answer: Runs hourly with 3 km resolution and explicitly resolves convection across the CONUS
The HRRR runs every hour with a 3 km grid and uses a convection-allowing configuration, making it highly useful for short-range convective forecasting and live storm coverage.
Question 3: What is 'quantitative precipitation forecast (QPF)' as it appears in NWP model output?
- A qualitative description of precipitation type
- A numerical estimate of the total liquid-equivalent precipitation over a specified time period (Correct answer)
- The probability of any measurable precipitation occurring
- The radar-estimated precipitation over the past hour
Correct answer: A numerical estimate of the total liquid-equivalent precipitation over a specified time period
QPF is the model's numerical prediction of total liquid-equivalent precipitation accumulation over a defined time interval, expressed in inches or millimeters.
Question 4: Which of the following statements about the Rapid Refresh (RAP) model is correct?
- It is a global model with 13 km horizontal resolution updated every 3 hours
- It is a regional hourly-updating model covering North America at approximately 13 km resolution (Correct answer)
- It replaces the HRRR for convective-scale forecasting
- It uses a 1 km grid and is the finest-resolution operational model
Correct answer: It is a regional hourly-updating model covering North America at approximately 13 km resolution
The RAP is a NOAA hourly-cycling regional model covering North America at roughly 13 km resolution, providing frequently updated guidance for short-range forecasting.
Question 5: In NWP, 'convective parameterization' schemes are used when model grid spacing is too coarse to:
- Compute sea-surface temperatures accurately
- Explicitly resolve individual thunderstorm updrafts and downdrafts (Correct answer)
- Assimilate satellite radiance data
- Forecast long-wave radiation at the surface
Correct answer: Explicitly resolve individual thunderstorm updrafts and downdrafts
Convective parameterization approximates the bulk effects of sub-grid-scale convection when grid spacing (typically >4 km) is too large to explicitly simulate individual storm-scale circulations.
Question 6: A broadcast meteorologist comparing the 00Z GFS run to the 12Z GFS run notices significant differences in the forecast. The best way to convey this to the audience is to:
- Always trust the newer 12Z run and not mention uncertainty
- Communicate the range of outcomes and acknowledge forecast uncertainty to viewers (Correct answer)
- Wait until both runs agree before issuing any forecast
- Only present the run that matches climatology
Correct answer: Communicate the range of outcomes and acknowledge forecast uncertainty to viewers
When model runs disagree, communicating uncertainty transparently — including a range of possible outcomes — is the professional standard and builds long-term viewer trust.
Question 7: What is the primary advantage of ensemble mean forecasts over a single deterministic model run?
- The ensemble mean is always more accurate than any individual member
- The ensemble mean reduces the impact of outlier solutions and generally outperforms a single run on average over many cases (Correct answer)
- The ensemble mean provides higher spatial resolution than deterministic models
- The ensemble mean eliminates the need for post-processing with MOS
Correct answer: The ensemble mean reduces the impact of outlier solutions and generally outperforms a single run on average over many cases
By averaging multiple perturbed runs, the ensemble mean smooths out individual member errors, yielding statistically better performance than any single deterministic run over a large sample of cases.
CBM Numerical Weather Prediction and Forecast Models 2
Model Output Statistics (MOS) improve upon raw model output primarily by: