ps_blue_tarp_detections.ej.data.mdx

---
id: ps_blue_tarp_detections
name: "PlanetScope Blue tarp Detections"
description: "Machine learning generated blue tarp detections using PlanetScope 3-band RGB imagery"
media:
  src: ::file ./ps-bluetarp--dataset-cover.jpg
  alt: Blue tarp detections for Jefferson Parish, LA on February 12, 2022
  author:
    name: NASA
    url:
taxonomy:
  - name: Topics
    values:
      - Disasters
  - name: Source
    values:
      - Planet
infoDescription: |
  ::markdown
    PlanetScope provides 3-band RGB imagery at 3-meter ground resolution which
    can support building-scale analysis of the land surface. In the aftermath of
    natural disasters associated with high wind speeds, homes with damaged roofs
    typically are covered with blue tarps to protect the interior of the home
    from further damage. Using machine learning, blue tarps can be detected from
    the PlanetScope imagery using pre-event cloud free images to detect blue
    pixels and potential impacts after a natural disaster.
layers:
  - id: blue-tarp-detection
    stacCol: blue-tarp-detection
    name: "Blue tarp detections"
    type: raster
    description: "machine learning generated blue tarp detections. Includes copyrighted material of Planet. All rights reserved."
    zoomExtent:
      - 14
      - 20
    sourceParams:
      resampling: cubic_spline
      bidx: 1
      colormap_name: reds
      rescale:
        - 0
        - 400
    info:
      source: NASA
      spatialExtent: Puerto Rico
      temporalResolution: Sub-Annual 
      unit: N/A

  - id: blue-tarp-planetscope
    stacCol: blue-tarp-planetscope
    name: PlanetScope input RGB imagery used for blue tarp detection
    type: raster
    description: "PlanetScope input RGB imagery used for blue tarp detection. Includes copyrighted material of Planet. All rights reserved."
    zoomExtent:
      - 14
    compare:
      datasetId: ps_blue_tarp_detections
      layerId: blue-tarp-detection
      mapLabel: |
        ::js ({ dateFns, datetime, compareDatetime }) => {
          if (dateFns && datetime && compareDatetime) return `${dateFns.format(datetime, 'LLL yyyy')} VS ${dateFns.format(compareDatetime, 'LLL yyyy')}`;
        }
    info:
      source: NASA
      spatialExtent: Puerto Rico
      temporalResolution: Sub-Annual 
      unit: N/A

---

<Block>
  <Prose>
    PlanetScope provides 3-band RGB imagery at 3-meter ground resolution which
    can support building-scale analysis of the land surface. In the aftermath of
    natural disasters associated with high wind speeds, homes with damaged roofs
    typically are covered with blue tarps to protect the interior of the home
    from further damage. Using machine learning, blue tarps can be detected from
    the PlanetScope imagery using pre-event cloud free images to detect blue
    pixels and potential impacts after a natural disaster.
  </Prose>
</Block>

<Block>
  <Prose>
    ## Scientific research Detection of blue tarps from high resolution imagery
    can inform disaster response of the most impacted locations. Additionally,
    given the frequency with which PlanetScope scenes are retrieved from the
    satellite, the rate of recovery for a given location can also be tracked
    over time. This can also support disaster response to provide aid to
    specific locations where recovery efforts are lacking.
  </Prose>
</Block>

<Block type="full">
  <Figure>
    <Map
      datasetId="ps_blue_tarp_detections"
      layerId="blue-tarp-planetscope"
      zoom={17}
      dateTime="2022-02-12"
      compareDateTime="2022-02-12"
      center={[-90.0613, 29.914]}
    />
    <Caption attrAuthor="NASA" attrUrl="https://nasa.gov/">
      Blue tarp detections in Jefferson Parish, LA on February 12, 2022.
    </Caption>
  </Figure>
  <Prose>
    ## Interpreting the data Machine learning-based detections of blue tarps are
    displayed as red pixels on the basemap and are available for both Hurricane
    Maria (2017) and Hurricane Ida (2021). The input true color imagery used to
    produce the detections are also added as an additional layer and can be used
    for qualitative validation of the detections.
  </Prose>
</Block>

<Block>
  <Prose>
    ## Credits

    1. Source data made available through the NASA Commercial Smallsat Data Acquisition (CSDA) Program
   </Prose>
</Block>

<Block>
  <Prose>
    ## Additional resources 
    1. [Machine Learning activities at NASA IMPACT](https://impact.earthdata.nasa.gov/project/ml.html) 
    2. [Commercial Smallsat Data Acquisition Program](https://www.earthdata.nasa.gov/esds/csda)
  </Prose>
</Block>