{
  "id": 6237,
  "title": "First line in weather files",
  "url": "/competitions/flight2-main/discussion/6237",
  "author_name": "TecS",
  "post_date": "2013-11-05T20:32:34.390000",
  "votes": 0,
  "comment_count": 0,
  "views": 1195,
  "content": "<p>I'd like to check that my understanding of the first line in the weather files used by the simulator is correct. For instance, in weather_20130910_1803 it is<br>{&quot;Interval&quot;:1.0,&quot;Minimum&quot;:-0.064722222222222223,&quot;ElementCount&quot;:8}</p>\n<p>I read this as time interval between successive elements = 1.0 hour; time offset from start of simulation of the first element = -0.0647222... hours; number of elements = 8.</p>\n<p>So there are 8 snapshots of weather conditions in the file, one hour apart, starting 3.8833... minutes before the start of the simulation.</p>\n<p>Note that 3.8833... &gt; 3, which you might expect from the cutoff time in the filename (18:03). All weather files I've looked at have the same property, a time offset larger than the time since the official cutoff by some fraction of a whole minute.</p>\n<p>My guess: this is because Kaggle picks the cutoff time as a real number, but truncates the official cutoff time seen in the filenames down to whole minutes.</p>\n<p>If this is correct, I guess it does not matter much; everybody will be training on weather data that is off by the same fractions of a minute. Or did I miss a more precise source of time information somewhere else, either in Kaggle's files or in the RAP files?</p>\n<p>Edit: indirectly answered <a href=\"http://www.gequest.com/c/flight2-main/forums/t/6115/how-the-time-in-actuallandings-file-is-calculated/33428#post33428\">here</a>.</p>",
  "messages": [
    {
      "id": 33282,
      "postDate": "2013-11-05T20:32:34.390Z",
      "content": "<p>I'd like to check that my understanding of the first line in the weather files used by the simulator is correct. For instance, in weather_20130910_1803 it is<br>{&quot;Interval&quot;:1.0,&quot;Minimum&quot;:-0.064722222222222223,&quot;ElementCount&quot;:8}</p>\n<p>I read this as time interval between successive elements = 1.0 hour; time offset from start of simulation of the first element = -0.0647222... hours; number of elements = 8.</p>\n<p>So there are 8 snapshots of weather conditions in the file, one hour apart, starting 3.8833... minutes before the start of the simulation.</p>\n<p>Note that 3.8833... &gt; 3, which you might expect from the cutoff time in the filename (18:03). All weather files I've looked at have the same property, a time offset larger than the time since the official cutoff by some fraction of a whole minute.</p>\n<p>My guess: this is because Kaggle picks the cutoff time as a real number, but truncates the official cutoff time seen in the filenames down to whole minutes.</p>\n<p>If this is correct, I guess it does not matter much; everybody will be training on weather data that is off by the same fractions of a minute. Or did I miss a more precise source of time information somewhere else, either in Kaggle's files or in the RAP files?</p>\n<p>Edit: indirectly answered <a href=\"http://www.gequest.com/c/flight2-main/forums/t/6115/how-the-time-in-actuallandings-file-is-calculated/33428#post33428\">here</a>.</p>"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "33282": ""
  }
}