Free and open hires GRIB files for the Med are available

N

norulz

New Member
Joined
2 Nov 2014
Messages
6
Location
In the Med
Visit site
The University of Athens, Department of Physics, Atmospheric Modelling and Weather Forecasting Group has provided access to the output of their SKIRON forecast model. The compiled GRIB files are available for free download at wild-silk.org. A site that supports free and open weather resources for leisure sailors in the Mediterranean.

The files have a resolution of 0.1 deg. compared to 0.5 deg. for GFS gribs. This is 25 time higher than the GFS gribs. The SKIRON model covers the entire Mediterranean. It is a local model that well reflects local weather effects. This makes it suitable for sailing forecasts in the Med both coastal and off-shore. The well known weather pages of the University of Athens are based on the outputs of the SKIRON model.

The downloaded GRIB files can be viewed in zyGrib (ver. 6.2.4 and up) or in OpenCPN with the Grib plugin.

The attached image shows the coverage of the GRIB files.
grib-index-b.jpg
 
Excellent stuff! Very many thanks.

I was caught out at first by an older copy of zygrib but the latest (6.2.4) displays them perfectly.

Yet another useful weather tool with a different data source.

:)
 
Twister_Ken said:
For those of us not yet Gribbing, any suggestions for software to open .grb files on a Mac?
Click to expand...

No Mac info, but happy to report that I have viewed the Aegean area via "SailGrib free" on Android (Hudl-1) tablet. I had to download and unzip separately, so not as convenient as all-in-one, but usable...

Mike.
 
norulz said:
The University of Athens, Department of Physics, Atmospheric Modelling and Weather Forecasting Group has provided access to the output of their SKIRON forecast model. The compiled GRIB files are available for free download at wild-silk.org. A site that supports free and open weather resources for leisure sailors in the Mediterranean.

The files have a resolution of 0.1 deg. compared to 0.5 deg. for GFS gribs. This is 25 time higher than the GFS gribs. The SKIRON model covers the entire Mediterranean. It is a local model that well reflects local weather effects. This makes it suitable for sailing forecasts in the Med both coastal and off-shore. The well known weather pages of the University of Athens are based on the outputs of the SKIRON model.

………………….
Click to expand...

On the face of it, that could be a useful service. However, a note of warning; like other non-official GRIB type services, the hi res bit is misleading

If you look at the Athens University site, you will see that they use no more meteorological data than the GFS 0.5 degree output. In other words, they only know about weather detail on a scale of about 200-250 km size. I realise that they put in topographical data on the 0.1 degree grid but, as with most, if not all, non-official GRIB type services, their detailed model forecasts will be compromised accordingly.

I realise that it needs a browser but I would recommend http://212.175.180.126/DTS/sea.php. The output is direct from the ECMWF. They use one of the best global models, if not the best. They use the maximum amount of data possible and they analyse on the scale of their model ie 0.125 degrees. Being a global model they do not have boundary updating problems.
 
franksingleton said:
On the face of it, that could be a useful service. However, a note of warning; like other non-official GRIB type services, the hi res bit is misleading

If you look at the Athens University site, you will see that they use no more meteorological data than the GFS 0.5 degree output. In other words, they only know about weather detail on a scale of about 200-250 km size. I realise that they put in topographical data on the 0.1 degree grid but, as with most, if not all, non-official GRIB type services, their detailed model forecasts will be compromised accordingly.
Click to expand...

Hi,

The GFS model is only one of the input fields for the full fledged SKIRON arithmetic forecast model. See a short description of the model at this link. The model generates data from a broad set of inputs to a resolution of 0.05 degree. The distributed gribs are actually diluted to 0.1 degree.

...they use no more meteorological data than the GFS 0.5 degree output.
Click to expand...
A bit harsh! There are other inputs and then a run of a full fledged forecast model in a super-computer. In no way just a rehash of GFS gribs.

The SKIRON model is considered one of the most authoritative models in Europe and the Mediterranean. About 10 other weather services and oceanic institutes use the SKIRON as the basis for their sub-models.

If you Google "skiron" you will come up with a lot of information and perhaps be a bit less discrediting of the model.
If you wish I can put you in contact with the model developer to help gap your concerns.
 
norulz said:
Hi,

The GFS model is only one of the input fields for the full fledged SKIRON arithmetic forecast model. See a short description of the model at this link. The model generates data from a broad set of inputs to a resolution of 0.05 degree. The distributed gribs are actually diluted to 0.1 degree.

.............
Click to expand...

From the Athens U page,
Input Data
• Atmospheric data : NCEP/GFS initial and lateral boundary conditions (0.5 x 0.5 degrees, 26 preuote)ssure levels)

• Terrain and sea-surface data : Topography 30x30 sec, Vegetation 30x30 sec, Soil Texture 30x30 sec, daily NCEP SSTs 0.5x0.5 degrees.

• Initialization : 12 UTC daily

• Forecast Horizon : 5 days
Click to expand...

The only meteorological data used are from the GFS. I said that they use topographic data. I have no doubt that the model is good. My point is that, in common with other non-official models, with no detailed meteorological data then the ability to predict detail is limited.

The GFS does, in fact, run on a 0.25 degree grid but only makes available data on a 0.5 degree grid. The Turkish site that I referenced gives ECMWF data. ECMWF runs on a 0.125 degree grid and the Turks output the data on that same grid.

Further, the Athens model has to keep updating its horizontal boundaries with the GFS 0-5 degree data. ECMWF has no horizontal boundaries. Effects of the coarse boundary data will affect what happens over the model area. That is why detailed models really should only be used for fairly short times ahead – depending on the areas size, distance of any location from the boundaries, speed of movement of weather features etc.

You will see why I am rather sceptical about the use of even the best and best run meso-scale models beyond a day or so – never mind 5 days..
 
Hi Frank,

I'll try and take a practical look at the merit of models. I believe that the merit of a model is more in its output and what it is trying to achieve Than what went into the pot. The value of the leading global models in not in dispute. They are great for open sea crossings and overview of the big picture but they hold less value for the coastal sailor in the Mediterranean simply as they do not fully take into account many of the local weather effect factors and their output is too low in resolution to reflect wind variations in between the islands in the Aegean for example.

Sailors in the Eastern Med will usually consult more with the HNMS weather site, with Poseidon, with meteo.gr, with the UOA site (where the interactive interface is quite good) and also with the Turkish national weather site. (I agree with you - they do a good job and have good outputs). The local sailors will consult less with Passage Weather and zyGrib (GFS) when they are seeking local coastal forecasts. All in the previous list are using arithmetic forecast models with varying inputs and with varying algorithms in their computing, however, they are all doing their best to provide localised high resolution outputs for an area that has dominant localised weather behaviours.

I totally agree with you that few models, if any, provide much accuracy more than about 2 days ahead (for local weather). In the Mediterranean this is not much of an issue as one is in cellular communications most of the time and crossings with no weather browsing or downloads are up to 3 days at the most

So... where does this leave a slightly frustrated sailor like myself who can find pretty accurate weather resources for the Med but all are browser based. What about Gribs to provide for weather visualization in my chart plotter, download for off-line use and for tinkering around with weather-routing? For this, really only GFS is available - OK for the big picture but not much use in coastal areas.

What I did was to approach owners of forecast models: The Greek National Meteo Service, The Turkish National Meteo Service and the University of Athens Weather Forecasting group and I asked if they would kindly consider releasing output of their respective models in grid format freely and openly for the good of sailors like ourselves. No harm in asking? Right?

Well, I got only one response from Prof. George Kallos of the UOA. He likes the idea of free and open weather resources and kindly allowed access to the Skiron outputs to see if they can be used as is. The daily output is delivered in huge files each covering the whole of Europe and the Med. One file for each hour for atmospheric data on one for sea data. There are also many variables that are only useful for research and would not show up in regular grib readers. All together - Gigabytes.

To make the output useful for the sailing community I decided to set-up a dedicated server in Amazon cloud where a batch runs every morning to download the Skiron and WAM outputs, crop them to smaller relevant areas, concatenate the time layers, remove unneeded variables and merge Skiron and WAM.
The resulting manageable output is listed for free download on the wild-silk.org site which is hosted on the same server. Of course with attribution to the source of the data. If other model owners provide their gribs freely and openly for use - wonderful! We can then choose and pick and each have our own favourite set of gribs for which we can swear to its accuracy. If any model owner needs assistance in preparing the gribs in "sailor format" - I'm more than willing to do the same exercise with them.

In the small world of the Mediterranean there are currently only two sources of high resolution gribs - The French Meto-Consult who's gribs show wind only (BTW - they are, as far as I know, Skiron based as is) and the UOA Skiron gribs hosted on wild-silk.org. I can attest, after several years of using the Skiron interactive output, that it is a pretty accurate model for the Adriatic, Ionian, Aegean, Turkish coast, Cyprus and Israeli coast in comparison to other sources mentioned. I have not yet used it west of Italy.

Back to practicality - these are the resources and they are good for what they are designed for! Skiron is now freely available in Grib format. I can only hope that many others will find the grib files useful.

The proof is in the pudding - if they prove to provide decent forecast, they will be used and appreciated and if not we will continue the search for better gribs.

David
 
Last edited:
OK, David. I see where you are coming from.

I spent 10 years in the Med with GPRS as my only way of getting forecasts off the Internet. That is why I said this Skiron GRIB could be a useful service. I was only questioning the idea that it was a high res service.

The quality of any numerical model depends on the data input; atmospheric and (what I call broadly) topographic. As far as I know none of the “unofficial” models use anything other than GFS (occasionally CMC) 0.5 degree data. All they can do is use some form of interpolation between data points. As I said, they can only start with defining weather on a 200-250 km scale.

There are always ifs and buts in meteorology. In the case of such meso-scale models, there will be situations when the weather is largely driven by local heating. To that extent, such models can produce detail on the scale of around 40-50 km (4 or 5 times the grid length.) That must be the case in the Eastern Med at times but not always so and not, in any case, not generally so in the Med as a whole.

The major modelling centres use grids as small as their computers can handle and the data can support. There are two reasons for smaller grids.

First is for short term prediction of severe weather, such as severe connective storms, squal lines, motorway and airport fog. Such events can be predicted in general terms days ahead but only hours ahead in terms of detail. The finest scale data available are used. Some satellite sensors produce data down to a few km; so does radar. The data are of many different kinds, times, representivity, accuracy, density and frequency. The processing for analysis is enormous and contains results of continuously ongoing experimentation, monitoring and improvement . None of the private firms or universities can handle data in the same ways.

The second, arguably more important, use of short grid lengths to improve analysis for global models. In simple terms, the better you know what is happening now the better and longer ahead will be the end results. ECMWF global model uses a 0.125 degree grid, the UK only slightly larger. They outperform the rest. Shorter grid lengths require greater resolution in the vertical and shorter time steps. The UK uses 70 levels in the vertical up to about 80 km; ECMWF uses over 100 levels.

The UK meso-scale model uses a 4 km grid over a large area (mid-Atlantic to the urals) and 1.5 just for the UK. They only go to 36 hours. Partly. that is because of the boundary issue, even over such a large area, and partly in the light of the lifetimes of small weather detail One of the reasons for their recent computer upgrade is to be able to run ensembles of their meso-scale models. That is a recognition of the limitations of detailed deterministic prediction.

Perhaps you can see why I always have to query claims of great precision made by some firms (not, in all fairness Athens U.) It is also why I challenge claims by some users that such and such a forecast is “very accurate” or “precise.” The weather is neither; nor are forecasts..

In brief, models such as those run by these unofficial organisations will not grossly mislead. Neither will they give high res information.

In my time in the Med, I only had the GFS. By the time we left (2009) we were able to see Mistrals coming 5+ days ahead. We could see Boras coming. We could see strong winds coming several days ahead. That was good enough for me.

Were I there now, I would no doubt use a dongle with 3G to get the ECMWF data from Turkey. That would give me the most soundly based forecast of detail at a size scale that can be sensibly predicted over useful time scales. In all probability, I would use Skiron if I did not have 3G Internet access. I would use the GFS de faut mieux otherwise for outlook/planning purposes.
 
One thing we found is that Greek offshore forecasts are fine but closer to the coast the forecasts can't forecast Katabatic winds, which can be a nightmare. What we worked out is that if there is dew on the deck in the morning there will be Katabatic winds starting at noon and finishing at 17.00. So if we set off somewhere we made sure we were there before noon.
 
KellysEye said:
One thing we found is that Greek offshore forecasts are fine but closer to the coast the forecasts can't forecast Katabatic winds, which can be a nightmare. What we worked out is that if there is dew on the deck in the morning there will be Katabatic winds starting at noon and finishing at 17.00. So if we set off somewhere we made sure we were there before noon.
Click to expand...

I think you're referring to the "sea breeze" an afternoon wind caused by the heating of the land. A katabatic wind is a gravity wind which usually occurs at night as the land cools.
 
>I think you're referring to the "sea breeze" an afternoon wind caused by the heating of the land. A katabatic wind is a gravity wind which usually occurs at night as the land cools.

Sea breeze is a UK expression as you say caused by the land heating and an onshore wind. In Greece Katabatic winds blow offshore and are usually much stronger than a sea breeze, gale force is not unusual. They are caused by the mountains heating up. They can go on most of the night but it is unusual and usually stop, as I said, about 19.00 as the mountains cool as the sun goes down.

Once a Katabatic wind sprang up about 10.00, I was down below and the next minute I was thrown across the boat looking at the water as we got knocked down, we were very close to shore so had no warning. We did radio boats further out to warn them.
 
KellysEye said:
One thing we found is that Greek offshore forecasts are fine but closer to the coast the forecasts can't forecast Katabatic winds, which can be a nightmare. What we worked out is that if there is dew on the deck in the morning there will be Katabatic winds starting at noon and finishing at 17.00. So if we set off somewhere we made sure we were there before noon.
Click to expand...

KellysEye said:
>I think you're referring to the "sea breeze" an afternoon wind caused by the heating of the land. A katabatic wind is a gravity wind which usually occurs at night as the land cools.

Sea breeze is a UK expression as you say caused by the land heating and an onshore wind. In Greece Katabatic winds blow offshore and are usually much stronger than a sea breeze, gale force is not unusual. They are caused by the mountains heating up. They can go on most of the night but it is unusual and usually stop, as I said, about 19.00 as the mountains cool as the sun goes down.

Once a Katabatic wind sprang up about 10.00, I was down below and the next minute I was thrown across the boat looking at the water as we got knocked down, we were very close to shore so had no warning. We did radio boats further out to warn them.
Click to expand...

I don't want to go overboard on this but your first post talked about noon to 17:00. That's the sea breeze, an afternoon onshore wind. As your second post says katabatic winds occur at night (and usually blow offshore). I am in Greece so I know well how bad the katabatic winds can be in the early night, you're right that they can be very strong.
 
Up
Tony Cross said:
I don't want to go overboard on this but your first post talked about noon to 17:00. That's the sea breeze, an afternoon onshore wind. As your second post says katabatic winds occur at night (and usually blow offshore). I am in Greece so I know well how bad the katabatic winds can be in the early night, you're right that they can be very strong.
Click to expand...

I think he got his wires slightly crossed. A katabattic is, as he described in his second post. You can think of it as a rather exaggerated land breeze. Dare I say it - read my book!
 
>I think he got his wires slightly crossed. A katabatic is, as he described in his second post. You can think of it as a rather exaggerated land breeze. Dare I say it - read my book!

A katabatic wind, from the Greek word katabatikos meaning “going downhill” it's certainly not a sea breeze.
 
KellysEye said:
>I think he got his wires slightly crossed. A katabatic is, as he described in his second post. You can think of it as a rather exaggerated land breeze. Dare I say it - read my book!

A katabatic wind, from the Greek word katabatikos meaning “going downhill” it's certainly not a sea breeze.
Click to expand...

That is what I said-or intended to say in agreeing to your description in your second post. Your first was a little misleading and that is why, I think, Tony tried to correct you.

The katabatic effect is caused by descent of air down mountain sides. A Bora is a katabatic on a large scale. The descending air loses potential energy and, therefore, gains kinetic energy- conservation of energy. Hence the strengths of the wind.

We tend to think of the air moving downhill because it is cold and denseer than its surroundings. In fact, the cooling of air by contact with cold ground increases the pressure at the level where the cooling is taking place. This creates a horizontal pressure gradient at the height of the mountain side. Therefore the air moves towards lower pressure which will be downhill towards the sea in the case of Greek islands.

This is pretty small scale stuff in terms of weather models although small scale models should work if the mountains are big enough to be recognised by the model grids. When we were in the Med, the GFS showed the Boras developing quite well. Windfinder should do a little better and finer scale models still eg the Croatian Met Service should do should do even better.

A land breeze works in much the same way. Cooling of land at night increases pressure over land and results in seawards movement. That is why I said that the katabatic was similar to a land breeze in the way that it forms. The sea breeze is the reverse. Warming the ground during a sunny day causes a decrease in pressure over the land and a movement of air from sea to land.

You can get the reverse of the katabatic- the anabatic. This is when slope of mountains get warm, pressure is decreasd and air then moves from high to low pressure ie up the slope.
 
You must log in or register to reply here.

Share this page

Top