Soundings _ Australia


Click to access skew-t.pdf


The Skew T Log P Aerological Diagram

Bureau of Meteorology › Aviation Weather Service


Click to access skew-t.pdf




AAO correlated with southern hemisphere synoptic 2019

Here l take snaps of the ACCESS g model southern hemisphere synoptic and also record the AAO/SAM parameter.

Looking for synoptic patterns and if they correlate with shifts in the AAO. I will make a few observation notes as warranted.

AAO link.

ACC g  southern hemisphere synoptic

The first entry is below and then scroll down to comments section for all further additions. Click on the heading to load if necessary

5th may 2019 _AAO

image below 8th may 2019  AAO  going strong positive  Here +0.7 and going up

8th may 2019 polar low displaces cold air _SH

11th may 2019 aao vs SH synoptic


Ken what is the difference between deterministic and ensemble?

Ken Kato is a senior meteorologist on the Qld coast with BOM. He runs a facebook page

south Brisbane storms


is a frequent and long term  contributor to the ‘weatherzone forum’

Many people ask Ken excellent questions

Here a forum member asks Ken about model runs

and here is Kens’ answer


In the context of a single model ensemble like the EC ensemble, it’s when a single model like the EC (or any other model) is run multiple times to generate a range of scenarios (members).
The EC ensemble always generates 50 members. Each member starts off with slightly different initial states of the atmosphere and oceans. This is because we don’t have obs data for every square inch of the whole planet (satellite data covers most of the planet but it has its own degree of uncertainties) – it’s these uncertainties in the exact initial state of the atmosphere and oceans that grow and grow as you look further into the future and are one of the biggest contributions to increasing forecast error over time (chaos theory, butterfly effect, etc). So the model uses a sophisticated system to deliberately inject variations (weighted towards the uncertainties which are most likely to cause the maximum forecast errors) into the starting state of the atmosphere/ocean for each member.
If most ensemble members are tightly clustering around a particular scenario (e.g. max temp of 35C for a particular day, 25-50mm of rain, etc), it generally means that the weather setup concerned is insensitive to influences that can throw that forecast off which implies there’s a high confidence in that model for that scenario. But if there’s a big spread in the ensemble’s members, it implies the setup’s sensitive to even small changes and uncertainty is high.

In contrast, the deterministic version of a model (the forecasts from models like EC, GFS, ACCESS-G, etc whose forecasts you see on most websites) is just a single scenario.

Most deterministic models have an ensemble version.

The advantage of ensembles is that they give a great idea of how confident or uncertain a model is for particular scenarios. Comparing an ensemble to a deterministic version of a model is a bit like asking a big bunch of doctors for a diagnosis on a hard-to-diagnose disease compared to asking just a single doctor.
One disadvantage of ensembles is that they have lower resolution than their deterministic versions (due to the computational resources they take up) so they can sometimes miss smaller scale details, underestimate the intensity of a smaller than normal intense TC, etc.

There’s also multimodel ensembles and grand ensembles… the former are ensembles consisting of multiple deterministic models (WATL and OCF are examples) and the latter are ensembles of ensembles.

So in a nutshell, think of ensembles by their common definition such as that used for furniture i.e. a group of things that are treated as a whole thing. Ensemble = multiple scenarios from a model or models. Deterministic = single scenario from a model.

One thing to note is that a lot of forecast products from ensembles show the average of all the scenarios (e.g. WATL, OCF rainfall amounts, etc). While this is useful, it doesn’t show anything about how the scenarios are distributed, if they’re skewed, what outliers there are, etc. Therefore I prefer to look at probability forecasts from ensembles (percentage of an ensemble’s members going for a particular scenario) and preferably multimodel ensembles because a single model ensemble often tends to be more representative of its deterministic version rather than giving an appreciation of the true range of possible scenarios. ”