Just for fun, I'm going back to the problem of trying to classify or cluster the camera locations using just the provided remote sensing data. There are almost seven years of data. The average number of images per month for each location is 11.3. I'm combining each location's data on a monthly basis, creating composite images, in order to remove clouds and shadows. However, there is a cluster of 17 locations in Ecuador that have no data for March or June in any year. I've check the satellite image boundary maps and all 17 locations fall well within image boundaries.
\nI am wondering if this is a NASA problem, a Google Earth Engine problem, or an intentional anomaly introduced for this competition. @sbeery, would you be able to comment?
", "messages": [ { "id": "1345611", "postDate": "06/11/2021 17:47:06", "content": "Just for fun, I'm going back to the problem of trying to classify or cluster the camera locations using just the provided remote sensing data. There are almost seven years of data. The average number of images per month for each location is 11.3. I'm combining each location's data on a monthly basis, creating composite images, in order to remove clouds and shadows. However, there is a cluster of 17 locations in Ecuador that have no data for March or June in any year. I've check the satellite image boundary maps and all 17 locations fall well within image boundaries.
\nI am wondering if this is a NASA problem, a Google Earth Engine problem, or an intentional anomaly introduced for this competition. @sbeery, would you be able to comment?
", "rawMarkdown": "Just for fun, I'm going back to the problem of trying to classify or cluster the camera locations using just the provided remote sensing data. There are almost seven years of data. The average number of images per month for each location is 11.3. I'm combining each location's data on a monthly basis, creating composite images, in order to remove clouds and shadows. However, there is a cluster of 17 locations in Ecuador that have no data for March or June in any year. I've check the satellite image boundary maps and all 17 locations fall well within image boundaries.\n\nI am wondering if this is a NASA problem, a Google Earth Engine problem, or an intentional anomaly introduced for this competition. @sbeery, would you be able to comment?", "votes": null }, { "id": "1345929", "postDate": "06/12/2021 00:32:51", "content": "Thanks for pointing this out! This was not intentional on our end.
\nIf you're interested, this code is similar to what I used to collect the remote sensing imagery for the challenge. It's certainly possible that I made a mistake when collecting the imagery, but it would be strange for that sort of bug to only affect locations in a certain area.
\nIt's also possible that the imagery really isn't available at the quality level we impose (\"tier 1\"). One way to get some insight into this problem is to use the official Landsat viewer. To mirror our data collection process, use these settings: Landsat8 OLI only, tier 1 imagery only, cloud cover 0 - 100%. When I look at Ecuador in May for a few randomly chosen years, it does seem like there are often missing patches in the eastern part of the country. Does that line up with your observations?
", "rawMarkdown": "Thanks for pointing this out! This was not intentional on our end. \n\nIf you're interested, [this code](https://github.com/elijahcole/google-earth-engine-example) is similar to what I used to collect the remote sensing imagery for the challenge. It's certainly possible that I made a mistake when collecting the imagery, but it would be strange for that sort of bug to only affect locations in a certain area. \n\nIt's also possible that the imagery really isn't available at the quality level we impose (\"tier 1\"). One way to get some insight into this problem is to use the [official Landsat viewer](https://landsatlook.usgs.gov). To mirror our data collection process, use these settings: Landsat8 OLI only, tier 1 imagery only, cloud cover 0 - 100%. When I look at Ecuador in May for a few randomly chosen years, it does seem like there are often missing patches in the eastern part of the country. Does that line up with your observations?", "votes": null } ], "comments": [ { "id": 1345929, "author_name": "elijahcole", "author_url": "", "post_date": "06/12/2021 00:32:51", "content": "Thanks for pointing this out! This was not intentional on our end.
\nIf you're interested, this code is similar to what I used to collect the remote sensing imagery for the challenge. It's certainly possible that I made a mistake when collecting the imagery, but it would be strange for that sort of bug to only affect locations in a certain area.
\nIt's also possible that the imagery really isn't available at the quality level we impose (\"tier 1\"). One way to get some insight into this problem is to use the official Landsat viewer. To mirror our data collection process, use these settings: Landsat8 OLI only, tier 1 imagery only, cloud cover 0 - 100%. When I look at Ecuador in May for a few randomly chosen years, it does seem like there are often missing patches in the eastern part of the country. Does that line up with your observations?
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1345611": "Just for fun, I'm going back to the problem of trying to classify or cluster the camera locations using just the provided remote sensing data. There are almost seven years of data. The average number of images per month for each location is 11.3. I'm combining each location's data on a monthly basis, creating composite images, in order to remove clouds and shadows. However, there is a cluster of 17 locations in Ecuador that have no data for March or June in any year. I've check the satellite image boundary maps and all 17 locations fall well within image boundaries.\n\nI am wondering if this is a NASA problem, a Google Earth Engine problem, or an intentional anomaly introduced for this competition. @sbeery, would you be able to comment?", "1345929": "Thanks for pointing this out! This was not intentional on our end. \n\nIf you're interested, [this code](https://github.com/elijahcole/google-earth-engine-example) is similar to what I used to collect the remote sensing imagery for the challenge. It's certainly possible that I made a mistake when collecting the imagery, but it would be strange for that sort of bug to only affect locations in a certain area. \n\nIt's also possible that the imagery really isn't available at the quality level we impose (\"tier 1\"). One way to get some insight into this problem is to use the [official Landsat viewer](https://landsatlook.usgs.gov). To mirror our data collection process, use these settings: Landsat8 OLI only, tier 1 imagery only, cloud cover 0 - 100%. When I look at Ecuador in May for a few randomly chosen years, it does seem like there are often missing patches in the eastern part of the country. Does that line up with your observations?" }, "source": "meta" }