Sandaysoft

I am developing a tool to use your weather station's data to make a forecast out to twelve hours. I plan to write the software by using current conditions and building a archive of data. The program will then read these archives and pull out some average trends. After this it will take the "average trends" and draw a forecast out to twelve hours. I plan to have an hour by hour output of temperature, humidity, dew point, and eventually wind (wind will be slightly more difficult).

The reason I am posting is because I would like to see if anyone with a real meteorological background would like to assist me in helping build better algorithms for more accurate forecasts. I am just one that is very interested in weather and know a lot about it, but certainly no meteorologist. I do my coding in VB.NET and I know my way around, again, not an expert, but I have never hit a wall I could not cross when coding. Right now all I have the program doing is grabbing realtime.txt from the web and displaying the readings on the GUI.

So just so me some interest in a product like this or even offer me some help, I just ask that you keep me motivated to work on it. Who knows, maybe someday it will become your own personal computer model!

FYI - I have nothing against sharing the code it can be 100% open source, just let me know if you want the code.

_________________
My Site: http://tropwx.com/
My Sourceforge: http://sourceforge.net/users/tobol7

Dear cgwx1,

Rather than reinventing the wheel, I should recommend you work on some forecasting model available freely on the net, like GFS or WRF.

Do a Google search with these terms and they will return a bunch of related sites.

_________________
Marco-Luis
YV1HX
Cabimas, Zulia, Venezuela
http://www.twitter.com/meteoven

Well, "reinventing the wheel" is a long term goal of mine, so I figured why not start small with this, a very simple 12 hour temperature forecast. I also want this software to be able to run based on one weather station, not using a network of stations.

As for how far I have gotten, I now have the archiving step of the program done. It records the temperature increase/drop every hour to a document. Next step is to use the values and create averages for each hour of the day. After this I will have to build the actual forecasting module using these averages to see the future . Looks like it wont take long to get that far... Only 36 builds so far and already half way there! I will keep you updated on releases in this thread.

Anyone else see any interest in this software? Maybe if I were to add a way to output to your website the forecast that was made? Add graphics? Please give me suggestions on what to do to make the program better for you.

_________________
My Site: http://tropwx.com/
My Sourceforge: http://sourceforge.net/users/tobol7

While I do get frustrated when people complain about the Cumulus/Zambretti forecast, and try not to get involved in discussions about it, that's not because I'm not interested in this sort of thing; I do find it fascinating, and I encourage you to continue with your work.

Unfortunately I'm not currently in a position to offer you any help (and I'm no meteorologist anyway); I don't have time to do any work on Cumulus at the moment, let alone anything else. But if you find that it would help if Cumulus did something slightly differently, or provided something additional that would help you, please do let me know and I'll see what I can do.

_________________
Steve
Sanday Weather
----------------------------------------------------------------------------------------------------------------------------------
Like Cumulus and want to support it? Please donate!

You might not be reinventing the wheel as hinted by Marco-Luis, but...

For the past four years, I have been using a (shareware) software to model/forecast the weather in my area - it is called WXSIM - its forecasts are of professional quality, often as good as or even better than the forecasts provided by Environment Canada, especially since customization allowed the forecasts to be highly specific for the area where my weather station is located - Tzouhalem-Maple Bay Weather - WXSIM Weather Forecast Details.

WXSIM is described by its designer as an "interactive local atmospheric modeler" - it is initialized with current data, downloaded from the internet, then activated to model (forecast) the weather for the next few days, with the option to interact with the software and mix in some of one's own forecasting skill and knowledge.

Here are some of the variables modeled or used by WXSIM: Date, time, sun angle, distance from sun, latitude, longitude, elevation, proximity to large bodies of water, climatological temperature and dew point data, heat capacity of the surface, latent heat of condensation, incoming shortwave solar radiation, outgoing longwave terrestrial radiation, cloud albedo and emissivity, mixing due to winds and convection, advection of temperature and dew point, upper level temperatures and dew points (in 5 atmospheric layers), formation of dew, frost, and fog, formation of sea breezes, accumulation and melting of snow, and much more.

In addition to WXSIM's main job of modeling/forecasting, it performs a variety of other useful functions, including:

- Conversions among temperature, dew point, relative humidity, wet bulb temperature, vapor pressure, and mixing ratio.
- Calculations of heat index and wind chill.
- Sunrise and sunset times and sunrise azimuth for any point on earth any day of the year.
- Sun altitude and azimuth at any time and date for any point on earth.
- Solar irradiance and UV index, as functions of time, place, cloud cover, haze, and fog.
- Approximate normal temperature and dew point for any point on earth on any day of the year.
- Display and limited analysis of imported and interoplated RAOB soundings (TTAA data only).
- Graphical display (with printout option) of imported and interpolated NGM and Eta FOUS forecast data, including the option of weighted averaging of the two.

Unfortunately, WXSIM can only import weather data from the following weather software (with no future plan to include additional software): Virtual Weather Station, Weather Display and WeatherLink.

Good luck with your endevour

As has been suggested once you get past a bit of seaweed, forecasting becomes very hit and miss.

The Zambretti style forecaster is in place because it doesn't need a seperate data log.
(A Sager style should be an improvement if eg. we had a sense of wind veer over last 4 hours)

However, both those forecasters are only really intended for the Temperate Zones (cyclonically driven weather)

Mentioning location .... how will your model cope with eg. Northern / Southern hemispheres etc.

Not trying to put you off, just highlighling a few possible pitfalls

_________________

......................Imagine, what you will KNOW tomorrow !

Well, it is extremely simple at the time, it just records the average increase and drop of temperatures for each hour of the day and compiles that into a forecast. I do have plans to take this much further. You have to leave it running 24/7 as it currently has no other way to sample the weather (it does not read archives). It is very likely that you will run into a bug, and if you do please post it here so I can fix it. Thank you everyone who have given me tips on this project.

@Steve - I am in no way complaining about cumulus's forecast method, in fact I plan to include it in my forecast software to try to improve the forecasts. Also cumulus is great the way it is! It is what enabled me to start this project since it gives data in simple formats easy to work with.

--And whoever mentioned WXSIM, yes I have heard of that, but it is far to advanced for the average weather enthusiast. I am looking to make something that does almost 100% of the work for you.

THE FILE BELOW IS OLD, PLEASE GO TO THE END OF THE THREAD FOR THE LATEST VERSION.

_________________
My Site: http://tropwx.com/
My Sourceforge: http://sourceforge.net/users/tobol7

Thanks for this, it may help to have a few instructions on setup etc.
I've downloaded and got it running. To those that are interested you will need Net Framework 4, which you may need to download from MS, quite easy. I'm off out now so will see what I have when I get back home.

When I got back after a couple of hours the programme had failed, no reason given other than it needed to close.

_________________
Neil
Hurstpierpoint Weather

The are no data stored in the folder after running it for 45 minutes ?
edit after an hour it was created but the first forecast gives a temperature forecast in the next 15 minutes of 52,5 ...?? I hope this is not in Celsius the current temperature was 5.1 Celcius

_________________
Ron



Apeldoorn, The Netherlands

There should be no forecast if the software has been running for less than 24 hours... Obviously an issue there, I am going to look into that right now. Can you attach the data files that my software outputted? So I can see what went wrong.
You should see something like this...

_________________
My Site: http://tropwx.com/
My Sourceforge: http://sourceforge.net/users/tobol7

Thank you for the update. I am going to recode a little bit to give detailed error output so you can post the detailed error here so that I can fix it. Look for an update within the next day or so on this thread.

_________________
My Site: http://tropwx.com/
My Sourceforge: http://sourceforge.net/users/tobol7

I was able to figure out why the forecast was being created even though there was insufficient data, but not why it gave you an outrageous number. Changes as follows with this update.

------------------------------------------
Changes... Build 69 2/16/2012 2:04 AM EDT
1. Added error support. The software should (instead of crashing) give you a detailed error report that you can send to me so I can find solutions.
2. Changed UI to reflect the above.
3. Changed a value in the forecast method to prevent forecasting when there is insufficient data.
4. Minor fix to archiving section to prevent an error in an archived file.
--------------------------------------
Remember, this program requires access to the internet 24/7 and needs to be running 24/7 to work properly. Please continue with the feedback, it is great to know what is going wrong so I can get it fixed. I am also working on a better set of instructions.



If you have an issue with the software it is recommended that you erase all data files created by the previous version of the software since there were some issues with the code that could have allowed erroneous data to be added to those files. If this has happened and you do not clear your data files it is likely that you will experience incorrect forecasts.

_________________
My Site: http://tropwx.com/
My Sourceforge: http://sourceforge.net/users/tobol7

just had this in the error box


|| Length cannot be less than zero.
Parameter name: length mscorlib at System.String.InternalSubStringWithChecks(Int32 startIndex, Int32 length, Boolean fAlwaysCopy)
at System.String.Substring(Int32 startIndex, Int32 length)
at Weather_Forecaster.hr12temp.nextval(Int32 spaces)
at Weather_Forecaster.hr12temp.work(Object sender, EventArgs e) || Length cannot be less than zero.
Parameter name: length mscorlib at System.String.InternalSubStringWithChecks(Int32 startIndex, Int32 length, Boolean fAlwaysCopy)
at System.String.Substring(Int32 startIndex, Int32 length)
at Weather_Forecaster.hr12temp.nextval(Int32 spaces)
at Weather_Forecaster.hr12temp.work(Object sender, EventArgs e)

hope it helps

_________________
http://www.wunderground.com/weatherstation/WXDailyHistory.asp?ID=IARGYLLA4

The same error as Bruce occurred by me a few minutes ago.

_________________
Ron



Apeldoorn, The Netherlands

I've just had a similar error, after running for about 20mins.

_________________
Cumulus Real Time Data Grabber
http://bit.ly/WjBh5z

Wokingham Berkshire - weather station
http://www.wokinghamweather.co.uk

MRD Photography
http://www.mrdphotography.co.uk