Weather forecast differences

[Luke.]

I just jumped on to SkippySky and Cloud Free Night tonight and I noticed that they are reporting almost completely opposite conditions in Melbourne.
At the moment there is 100% low cloud cover here.

I was just wondering why two sites can be so different at the exact same time?

[cfn]

You are looking at two different models and the map scales are slightly different.

SkippySky is using the global GFS model.

CFN is using both the global GFS model and the local ACCESS model which is more accurate for Australian forecasts.

If you want to switch CFN to GFS viewing like SkippySky just run your finger over the model's name or click on the GFS button as illustrated in this post. Then everything will look like SkippySky but still in 3h increments up to 120h.

For further information have a look at CFN map help and read the topic about "Changing Map Model".

[Luke.]

Thanks for that! I wonder why SkippySky wouldn't use the more accurate version for Australia as defualt. I will probably just stick to CFN in that case anyway
Do you know which model clearoutside uses?

[cfn]

Originally Posted by Luke.

Thanks for that! I wonder why SkippySky wouldn't use the more accurate version for Australia as defualt. I will probably just stick to CFN in that case anyway

Almost every weather website and app you have ever heard of is providing data from the United States (NCEP) GFS model, because it is freely available, even for commercial use. Cloud Free Night is the only service currently providing detailed cloud forecasts information from the Australian (Bureau of Meteorology) ACCESS model for the benefit of Australian astronomers

Originally Posted by Luke.

Do you know which model clearoutside uses?

Sources used for clearoutside.

[Atmos (Colin)]

CloudFree Night is pretty accurate but I tend to take a statical between CFN, Skippy, Meteoblue & yr.no (they use another model entirely). Skippy & Meteoblue both use the same data (GFS) but tend to have slightly different interpretations on the data.

[multiweb (Marc)]

I just use CFN now. It's got everything in one place. All the others seem to be using GFS and ACCESS is more refined for us aussies anyway so why try harder.

[rustigsmed (Russell)]

interesting i'll have to check that out.
i've been using mostly clearoutside app. which was really good about 6 months ago but not so much these last couple of months. it was telling me there was cloud i look outside and it was clear clear clear! i ended up comparing sky skippy and clearoutside and they were a bit different too. but then again the weather has been crazy so i don't know if these are apps' faults.

[Wavytone]

Luke,

As alluded to earlier there are three kinds of apps:

a) those that use data from a computer model,
b) those that use ACTUAL LIVE DATA from weather stations that have been installed at appropriate locations and calibrated properly by the BOM;
c) those that use data from amateur weather stations that are uncalibrated, and usually in hopeless installations in terms of exposure to sun (temperature, humidity) and wind.

Type (b) latter are rare because the developer must pay a periodic fee to the Bureau of Meteorology to use the live data from their weather stations, whereas the computer model data is free.

Type (c) might be OK for what's happening on the owners balcony, but not an accurate representative of the conditions for the locality. But various weather services are happy to take the data anyway.

The difference can be very great and easily tested - wait for a cold front to pass your location with a sudden temperature drip and wind squall. The computer models have a very poor idea when this happened and the actual temperature and wind change, whereas the apps that use the BOM data from weather stations will accurately show the time, the real temperature drop and the wind change accurately.

As a paraglider pilot the wind and weather are absolutely critical - my ability to fly safely depend on it, and I learnt a great deal about weather and what goes on above by flying PG, where you get to experience it directly in your face.

[cfn]

CFN gets its data from the following sources:

GFS: United States National Center for Environmental Prediction (NCEP) Global operational Forecast System (GFS) model with 3-hourly forecast parameters out to 8 days (+192 hours), updated 4 times per day (00, 06, 12, 18 UTC), freely available (in real-time) for commercial use at NOMADS (NOAA Operational Model Archive and Distribution System).

ACCESS: Australian Bureau of Meteorology Australian Community and Earth-System Simulator (ACCESS) models - ACCESS-R, with hourly forecast parameters out to 3 days (+72 hours), updated 4 times per day (00, 06, 12, 18 UTC); and ACCESS-G, with three hourly forecast parameters out to 10 days (+240 hours), updated 2 times per day (00, 12 UTC), freely available (in real-time) for research use at Bureau of Meteorology OPeNDAP server - NMOC Atmosphere Data Library.

Himawari-8: Satellite image originally processed by the Bureau of Meteorology from the geostationary satellite Himawari-8 operated by the Japan Meteorological Agency, updated every ten minutes, freely available (in real-time) for non-commercial use at Bureau of Meteorology FTP Server.

To further increase your forecast confidence, check out the Norwegian weather agency, the only freely available service for detailed cloud forecasts from the European global model (ECMWF) for Australian locations. Links to the yr.no (meteogram) forecast, Bureau of Meteorology official 7-day forecast, Weatherzone three-hourly cloud cover probability for the next 48 hours, and sunrise, sunset and moon times are included on the meteogram page for each particular location.

[OzEclipse (Joe Cali)]

For solar eclipse chasing, weather forecasting is crucial.

In 2012, I engaged a commercial meteorological forecast company (Weatherwatch) to do the forecasting. They were fantastic.

However for most offshore eclipses, we have to do our own local forecasting in the days prior to the eclipse. We organize & plan our travels and select the general observing location months to years in advance using average climate statistics.

Listed below are general rules of thumb I use when looking at and interpreting forecasting models, not absolute dogma. During a solar eclipse, we have the added local issue of rapid cooling of the atmosphere that none of the models take into account.

1. Divergent Models
If different forecast sites that use different weather models (already listed in other posts) are indicating opposing forecasts, this is an indication that the forecasts are unreliable - could go either way. This may be due to unstable conditions.

2. Flipping Model
Models are published (calculated) at regular intervals (6,12, 24 hourly) If the same model is flipping, changing from fine to clear from publication to publication, this is an indication that the forecasts are unreliable - could go either way. This may again be due to unstable conditions.

3. Consistent Model
If the same model is consistent from publication to publication and give the same or similar forecasts, this is an indication that the forecast is probably more reliable.

4. Convergent Models
If the two different models are convergent and give the same or similar forecasts, this is an indication that the forecast is probably more reliable.

As mentioned different web apps publish data on different scales, different time bases etc and this can make comparison difficult.

I'd welcome hearing others opinions experiences re-the above rules of thumb.

regards

Joe

[multiweb (Marc)]

+1 makes a lot sense, especially when committing to an O/S trip. If two models say the same thing it's all good, otherwise sounds fishy.