{
  "id": 543797,
  "title": "The role of FGS in denoising AIRS-CH0",
  "url": "/competitions/ariel-data-challenge-2024/discussion/543797",
  "author_name": "Andrei Zamfir",
  "post_date": "2024-11-01T13:40:01.137000",
  "votes": 1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>It was very interesting reading through solutions to see a few particular aspects that multiple of us thought about and used in our way of tackling this problem. One thing that strikes me is, unless I missed it, I have yet to see a mention of FGS.</p>\n<p>Has anyone found any success? Any use cases? Could correlations be found and used to increase SNR? </p>\n<p>I lack the domain knowledge and my searches ended after computing a few PSD (power spectral density) plots and playing around with mixing FGS into various AIRS-CH0 frequencies, but nothing ever resulted in any CV increase.</p>",
  "messages": [
    {
      "id": 3033763,
      "postDate": "2024-11-01T13:40:01.137Z",
      "content": "<p>It was very interesting reading through solutions to see a few particular aspects that multiple of us thought about and used in our way of tackling this problem. One thing that strikes me is, unless I missed it, I have yet to see a mention of FGS.</p>\n<p>Has anyone found any success? Any use cases? Could correlations be found and used to increase SNR? </p>\n<p>I lack the domain knowledge and my searches ended after computing a few PSD (power spectral density) plots and playing around with mixing FGS into various AIRS-CH0 frequencies, but nothing ever resulted in any CV increase.</p>",
      "rawMarkdown": "It was very interesting reading through solutions to see a few particular aspects that multiple of us thought about and used in our way of tackling this problem. One thing that strikes me is, unless I missed it, I have yet to see a mention of FGS.\n\nHas anyone found any success? Any use cases? Could correlations be found and used to increase SNR? \n\nI lack the domain knowledge and my searches ended after computing a few PSD (power spectral density) plots and playing around with mixing FGS into various AIRS-CH0 frequencies, but nothing ever resulted in any CV increase.",
      "votes": 1
    },
    {
      "id": 3033788,
      "postDate": "2024-11-01T14:12:04.957Z",
      "content": "<p>I did use it in my final submission, but the gain is limited (~0.003 on training set). The main advantage is that you just get a lot more photons to find the mean of the transit depth correctly, but it's not helping as much there as I hoped for.</p>",
      "rawMarkdown": "I did use it in my final submission, but the gain is limited (~0.003 on training set). The main advantage is that you just get a lot more photons to find the mean of the transit depth correctly, but it's not helping as much there as I hoped for.",
      "votes": 2
    }
  ],
  "comments": [
    {
      "id": 3033788,
      "author_name": "Jeroen Cottaar",
      "author_url": "",
      "post_date": "2024-11-01T14:12:04.957000",
      "content": "<p>I did use it in my final submission, but the gain is limited (~0.003 on training set). The main advantage is that you just get a lot more photons to find the mean of the transit depth correctly, but it's not helping as much there as I hoped for.</p>",
      "votes": 2,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3033763": "It was very interesting reading through solutions to see a few particular aspects that multiple of us thought about and used in our way of tackling this problem. One thing that strikes me is, unless I missed it, I have yet to see a mention of FGS.\n\nHas anyone found any success? Any use cases? Could correlations be found and used to increase SNR? \n\nI lack the domain knowledge and my searches ended after computing a few PSD (power spectral density) plots and playing around with mixing FGS into various AIRS-CH0 frequencies, but nothing ever resulted in any CV increase.",
    "3033788": "I did use it in my final submission, but the gain is limited (~0.003 on training set). The main advantage is that you just get a lot more photons to find the mean of the transit depth correctly, but it's not helping as much there as I hoped for."
  }
}