Lectures 3 & 4 - Short term forecasting

What is short term forecasting essentially about?

What will the weather be in the coming 6-48h?

What is the most important tool for short term forecasting?

Regional NWP

What are some properties of regional NWP models?

• Highest possible resolution for representing mesoscale weather + topography

• frequent data-assimilation cycles with latest observations

• model output quickly available

• recent years: RUC (rapid update cycles 1 hourly ensembles, blending between hours 2 and 6 with nowcast systems)

Explain what “nesting” is?

Running regional NWP still requires, less-frequently-updating, slower models to provide boundary conditions

Which two regional NWP models are run by GeoSphere?

AROME, C-LAEF

What is the AROME-AUT model?

• deterministic, best guess initial and boundary conditions

• dx = 2.5km

• 90 vertical levels (30m resolution in the lowest 1km)

• 3 hourly

• lead time: 60h

What is C-LAEF?

• 16 member ensemble version of the AROME model

• dx: 2.5km

• 12 hourly

• Lead time: 60h

• pertubed physics, initial and boundary conditions

What is the difference between grid resolution and effective resolution?

effective resolution: what the model can resolve

grid resolution: what the grid spacing is in the model

effective resolution is ~10x greater than grid resolution

What are the grid resolutions of AROME, CLAEF, ICOND2?

dx= 2.5km, 2.5km, 2.2km

What does AROME stand for?

Application of Research to Operations at MEsoscale

What are two key features of AROME?

• Non-hydrostatic

• No parameterized deep convection

How long does AROME-AUT need to run a full forecast?

3+ hours

• Waiting for data to become available ~100min

• Data assimilation ~15min

• Time integration ~25min

What is the idea of an Ensemble?

Turn deterministic (single) NWP into probabilistic forecast by running it multiple times

Which variations are introduced to create an ensemble?

• Initial conditions (measurement uncertainty)

• Boundary conditions (uncertainty from the parent model)

• Model physics pertubations (radiation, shallow convection, turbulence, microphysics)

How do Rapid Update Cycles (RUC) work?

hourly runs that incorporate the latest radar observation, already operational at some national weather institutes

How is Forecast uncertainty definied?

• Forecasting is all about quantifying (un)certainty of specific weather types

• Uncertainty is an essential part of the warning-based forecasting

• High certainty x High impact = high-end warning

Which two possibilities exist for estimating uncertainty in forecasts?

• Ensemble forecast systems

• Conversion of deterministic into an ensemble using spatial and/or temporal filters

What can be used to estimate the maximum potential of precipiation events?

ensembles

What can the ensemble median be used for regarding precipitation?

helps identifying regions of risk of flooding

How does uncertainty of an ensemble evolve with time?

it grows over time

what is the best methods to gauge forecast uncertainty?

using ensemble systems

What are the two most important short term forecast products?

• 2D Maps

• Meteograms

What is a strength of regional NWP?

Resolution compared to medium range forecasts

What would need to be the grid resolution of a model to resolve a thunderstorm (horizontal ~ 1km)?

100m resolution or finer ignoring well representing all physics at these scales

How does vertical resolution look like with increasing height?

Grids are typically increasing in vertical resolution with height,

near surface resolution is approximately 20m

• 30 levels in the lowest 1km (~33m)

• 10 levels from 1 - 2km (~100m)

• remaining 50 levels to the stratosphere (~30km)

which three types of inversions do exist?

• radiation inversion

• subsidence inversion

• frontal inversion

On what does the error of temperature and wind forecasts depend?

• Orography

• time of year

• time of day

Which models are better in flat terrain? Regional vs global models

In flat land regional high resolution models are hardly better than global models

Which models are better in mountainous terrain? Regional vs global models

in complex terrain AROME can beat IFS

Which is better, AROME or IFS, for precipitation <10mm/h?

both are useful on all scales

Which is better, AROME or IFS, for precipitation >10mm/h?

Beyond 10mm/h AROME is best across all scales

What are typical grid scales for regional NWP models?

1-3km horzontal resolution