2010 Haiti Earthquake

BACKGROUND

On Tuesday, January 12, 2010, a magnitude 7.0 earthquake struck just miles from Haiti's capital city of Port-au-Prince. In the days following the disaster, humanitarian aid and military organizations from around the world rushed to the small Caribbean nation to search for survivors, provide medical care, help restore order to the devastated city, and plan for long-term recovery.

The World Bank has contracted with IPLER partners ImageCat Inc. and RIT's Laboratory for Imaging Algorithms and Systems to collect high quality aerial multi-spectral imagery and LiDAR information over the affected area to aid in the relief effort. The team has flown the WASP camera system and a LiDAR sensor from Kucera International over Haiti, for a campaign that stretched from 1/21/2010-1/27/2010.

SPONSORS

The project was initiated by the World Bank, through a contract with IPLER partner ImageCat. The USGS also requested LiDAR coverage of the fault line to improve their earthquake risk predictions and to better understand the event.

UPDATE (2/26/2010)

The LIAS-DIRS teams in RIT's Center for Imaging Science have successfully completed seven days (1/21-1/27) of data collection over Haiti Port-au-Prince, Léogâne, Petit- and Grand Goâve, Fermate, Jacmel, and the western and eastern fault line areas. Please see Data Collection Overview for a summary of the campaign. Available data:

  • Imagery (1/21-1/27): Visible (RGB), shortwave-infrared, midwave-infrared, and longwave-infrared imagery for Port-au-Prince, Léogâne, Petit- and Grand Goâve, Fermate, and Jacmel completed using a 1m LiDAR-derived digital elevation model (DEM) for ortho-rectification.
  • LiDAR (1/21-1/27): All areas are available as digital elevation models (DEMs), surface elevation models (SEMs; 1st return surface), at both 1m and 10m spatial resolutions, and LiDAR point clouds (*.las format).

    NOTE: The LiDAR data collection was governed by the primary image collection specifications and flight design. Hence the LiDAR acquisition was not optimized for many applications, such as hydrology, wall-to-wall coverage, etc.

WASP Image of Haiti earthquake disaster: This image was captured at 2,500 feet above ground level and has a 0.15m spatial resolutionWASP Image of Haiti earthquake disaster: This image was captured at 2,500 feet above ground level and has a 0.15m spatial resolution

WASP image over Haiti that captures the need for a rapid response: This image is a zoom-in to a portion of the image on the left.WASP image over Haiti that captures the need for a rapid response: This image is a zoom-in to a portion of the image on the left.

A birds-eye view of the Haitian palace: A LiDAR point cloud of the Haitian palace - note the detailed structureA birds-eye view of the Haitian palace: A LiDAR point cloud of the Haitian palace - note the detailed structure

LiDAR data over the Haitian palace: This LiDAR coverage of the presidential palace shows the first LiDAR returns. Different colors depict different heightsLiDAR data over the Haitian palace: This LiDAR coverage of the presidential palace shows the first LiDAR returns. Different colors depict different heights

Video of WASP visible imagery draped over the LiDAR surface elevation model



Padding

DATA DETAILS AND DISSEMINATION

The data were piped from the University of Puerto Rico to RIT on a daily basis at rates that approached 47MB/s (total time to move 112G of WASP data: ~40 minutes). This is almost unheard of and has been made possible through a collaborative effort between the University of Puerto Rico, Internet2, NYSERNet, and the computing infrastructure folks at RIT. Data were made available to Google, Yahoo!, Microsoft, USGS, ERDAS, Virtual Disaster Viewer (VDV) at University at Buffalo, and UN-SPIDER (hosted at RIT). The data are in the public domain.

  1. Imagery can be accessed directly via the HTTP site at RIT.
  2. RIT's data can also be easily viewed at Telascience.
    - Expand the "Hi Res Aerials Image" tab on the left
    - Check or un-check the three aerial imagery choices under this tab - RIT's data are displayed as "Worldbank"
    - Scroll around the data by "grabbing" the imagery (left-mouse click) and dragging the mouse around while keep the left mouse button pressed
    - Zoom in and out using the mouse wheel
    - You can view the various type of collects by switching the check boxes on the left
    - RIT's imagery is now available for the southern coastal stretch of Port-au-Prince, stretching north to the airport
  3. The UN-SPIDER effort has resulted in a WMS service that also can be used to browse the imagery for interpretation purposes:
    - In ESRI Acrmap, go into ArcCatalog, and add a WMS server (NOTE: You will have to select 1.1 or 1.0 in order to avoid errors).
    - Point your WMS connectors at: "http://wms.cis.rit.edu/cgi-bin/mapserv.csh?map=/local2/home/cnicholas/mapfiles/haiti.map&Request=GetCapabilities".
    - You will have to zoom to a scale of 1:1,000 for the imagery to display properly.
  4. Data can be DOWNLOADED:


    IMAGERY (1/21-1/27 coverage) :

    = Anonymous FTP (not an SFTP) using thor.cis.rit.edu OR the WASP HTTP Site .

    - lwir-ortho-scaled (32-bit calibrated longwave IR radiance; λ=8-9.2µm; 0.83m)
    - mwir-ortho-scaled (32-bit calibrated midwave IR radiance; λ=3-5µm; 0.83m)
    - swir-ortho-scaled (8-bit shortwave IR digital numbers; λ=1-1.7µm; 0.83m)
    - vnir-ortho-scaled (8-bit RGB digital numbers; 0.15m)
    - vnir-ortho-nonscaled (10-bit RGB digital numbers; 0.15m)

    LIDAR (1/21-1/27 coverage) :

    = Anonymous FTP (not an SFTP) using thor.cis.rit.edu OR the WASP HTTP Site .
    = OpenTopography Facility : Please see the 2010 World Bank-ImageCat-RIT Haiti LiDAR data at the OpenTopography Portal

    All Digital Elevation Model (DEM) and Surface Elevation Model (SEM; 1st return surface) are now available (see "LiDAR_Rasters_DEM_SEM" folder)
    - 1/21-1/23 (full Port-au-Prince area): PaP south, central, and north files
    - 1/24-1/25 (Leogane & western fault line section): Leogane and FaultLineWest files
    - 1/26 (Petit- and Grand-Guave): Grand Guave files
    - 1/27 (Jacmel, Fermate, and eastern fault line section): Jacmel, Fermate, and FaultLineEast files
    - In both UTM and GCS
    - At both 1m and 10m resolutions
    - Note README file in respective folders
    Entire campaign LiDAR point clouds (*.las files)
    - Folder: "LiDAR_LAS_Files"
    - Folder: "LAS_Compressed_Files" contains 7-zip compressed files.

    NOTES:

    - All imagery are in GCS (WGS84) format.
    - The thermal (mid- and longwave IR) data are very unique modalities and require expert use and interpretation. The kite-type imagery or "indented image borders" on these are normal and are due to the camera model.
    - The LiDAR data collection was governed by the primary image collection specifications and flight design. Hence the LiDAR acquisition was not optimized for many applications, such as hydrology, wall-to-wall coverage, etc. E.g., the DEMs and SEMs are not seamless, since each day had to be processed separately for quick turnaround.

VALUE-ADDED PRODUCTS

RIT is also in the process of developing value-added products, e.g., tarp detection, building footprint delineation, etc. and has partnered with researchers from Purdue University, Virginia Tech (LiDAR products), University of Texas (geologists), and University at Buffalo (building damage assessment). Products will be disseminated via the Virtual Disaster Viewer, and the RIT Haiti Product Coordination Application (zoom to a scale of about 1:2,500 for the imagery to display properly). Examples of some of these products can be found at:

  • Blue Tarp Detection Scheme (Created by Dr. Dave Messinger from the Center for Imaging Science at RIT)
      Description: This detection scheme is based off the simple idea that the blue tarps in the imagery, indicative of Internally Displaced Persons (IDP's), are both rare and high in contrast to the rest of the image. A simple Mahalanobis Distance Classifier was developed to detect the presence of blue tarps in the imagery. Upon detection, because the imagery contain the geographic location information, the locations of the tarps can be determined. These locations are output as KML files that can be displayed by visualization tools such as Google Earth. A word of caution when using these in visualization products: the results are based on the imagery collected for the World Bank by RIT. These may not be the same imagery displayed by the visualization software you use. In particular, if the imagery displayed were not collected on the same date as the imagery used for the analysis, the tarps may have been moved, taken down, erected, etc. Consequently, there may be detections provided in the KML file that are not actually in the imagery you are using, or, you may see blue tarps that were not detected by this scheme. Actual detection planes and the GeoTIFF imagery used for the analysis are available upon request. These results are meant only to be a demonstration of how a semi-automated detection scheme can be used to identify signatures of Internally Displaced Persons, and how those results can be displayed visually.

PRESS COVERAGE

The World Bank - ImageCat - RIT effort has received coverage via a variety of press outlets:

SPONSORS