The effectiveness of alternative and renewable energy sources, such as solar and wind power stations, is directly dependent on weather conditions. Therefore, for the proper operation of such energy sources, it is necessary to constantly monitor weather forecasts. But, as practice shows, access to such forecasts is either not available for automatic systems, or does not contain the necessary data set.
The purpose of the work is to create a weather forecasting service with an automatic program interface that uses the global GFS forecasting system.
The Global Forecast System (GFS) is a computer model that launches a new global forecast four times a day (00:00, 6:00, 12:00 and 18:00 UTC). GFS produces 16-day forecasts and has a forecast grid of 0.25 degrees or approximately 27 km for the entire surface of the earth.
But the most interesting thing for us is that for the next 3 days or 72 hours, the model produces a weather forecast in increments of 1 hour. For the following days, the step is 3 hours.
At the moment, the forecast accuracy of the selected model is about 90% on the fifth day of the forecast, and is reduced by 2-3% for each next day. The image shows a comparison of the forecast accuracy of the selected model in comparison with other popular models, such as ECMWF and EC Parallel. And although the GFS model is slightly inferior in accuracy of forecasts, many weather sites around the world use this data. GFS also provides very good short-term forecasts for many types of weather, which many weather models do not offer.
When developing the service, it was decided to use the API interface as the main one for receiving data. This is done in order to simplify the work with forecasts in automated systems.
There are currently three types of data requests.
A standard weather forecast that contains data such as wind speed and direction, solar radiation flux and ambient temperature. This request format is chosen to save traffic or processor resources on the receiving side.
A complete weather forecast contains more than 500 different parameters of the GFS model and is presented as a dataset grouped by height layers. The values of the full forecast cover the atmosphere from the surface of the earth to an altitude of 10 km with a certain step.
And, at the end, the third type of query is statistics. Among the routes of statistics are wind speed and direction for heights of 10, 20, 30, 40, 50, 80 and 100 m above ground level, as well as ambient temperature and a downward flow of short-wave radiation.
The analysis of average annual indicators is made 2 times a day. Data is taken from January 1, 2010 to the current moment. This approach allows you to always have up-to-date information regarding the dynamics of changes in average annual indicators. It also allows you to build such beautiful and informative diagrams as in the presented image.