Scalable spatial-temporal BFAST

Reproduction of the computations on the article "Spatio-Temporal Change Detection from Multidimensional Arrays: detecting deforestation from MODIS time series"

Files:

• LICENSE - License file.
• README.md - This file.
• Docker files:
  • Dockerfile - Script for building a Docker Image.
  • setup.sh - Host script for removing former containers and images from host machine. Then, it creates a Docker image called "sstbfast_img".
• Container files:
  • containerSetup.sh - Script for setting up SciDB. Then it runs the experiment.
  • createArray.afl - Array Functional Language script for creating an array to store MODIS data.
  • downloaddata.sh - Script for downloading MODIS data.
  • removeArrayVersions.sh - Script for removing array's versions.

    <li>SciDB configuration files:
      <ul>
        <li><code>conf</code> - SHIM configuration file. Shim is a web service that allows to query SciDB.</li>
  			<li><code>iquery.conf</code> - IQUERY configuration file.</li>
  			<li><code>scidb_docker.ini</code> - SciDB's configuration file.</li>
        <li><code>setEnvironment.sh</code> - Script for setting environment variables for SciDB's users.</li>
        <li><code>startScidb.sh</code> - Script for starting SciDB.</li>
        <li><code>stopScidb.sh</code> - Script for stopping SciDB.</li>
      </ul>
    </li>
  
    <li>Install additional software:
      <ul>
        <li><code>installBoost_1570.sh</code> - Script for installing the Boost libraries.</li>
        <li><code>installGribModis2SciDB.sh</code> - Script for installing the tool which export MODIS data to SciDB's binary.</li>
        <li><code>installPackages.R</code> - R script for installing R packages.</li>
        <li><code>installParallel.sh</code> - Script for installing GNU Parallel.</li>
        <li><code>installR.sh</code> - Script for installing R.</li>
        <li><code>libr_exec.so</code> - Precompiled <i>r_exec</i> library for SciDB 14.12.</li>
      </ul>
    </li>
  
    <li>Scripts for running the expriement:
      <ul>
        <li><code>reprosarefp.R</code> - Main R script of Scalable spatial-temporal BFAST.</li>
        <li><code>rexec_sar_efp_f.R</code> - Auxiliary R script of Scalable spatial-temporal BFAST.</li>
  				<li><code>SAR_efp_R.R</code> - R script for processing Scalable spatial-temporal BFAST without using SciDB. <b>NOTE</b>: This script is provided for comparing the results.</li>
        </ul>
      </li>
  	</ul>
  </li>
  

  Pre-requisites:

  • Internet access
  • Docker.io
  • SSH
  • The host's ports 49905 and 49905 must be available.

  Instructions:

  NOTE: The execution of these scripts can span for several days.
  
  1. Clone the project and then go to the cloned folder: git clone https://github.com/ifgi/scalable-spatial-temporal-BFAST.git
  2. Modify the configuration file scidb_docker.ini according to your needs and your hardware.
  3. Enable setup.sh for execution chmod +x setup.sh and run it ./setup.sh. This creates a new Docker image from the Dockerfile and then starts a container.
  4. Log into the container: ssh -p 49905 root@localhost. The default password is xxxx.xxxx.xxxx
  5. Execute the script /root/./containerSetup.sh. This script sets SciDB, downloads MODIS data to SciDB, and finally executes the experiment.