Geoinformatics Center - MODIS Fire Information System

Geoinformatics Center of the Asian Institute of Technology (AIT), Bangkok, Thailand acquires MODIS satellite data from MODIS sensors aboard Terra and Aqua Satellites on a daily basis. The data are archived for the purposes of research on the global environmental change monitoring and management/prevention of disasters of different types in the Southeast Asian region. In this context, the authors having identified the need for an information system to better manage and control the disasters caused by forest fires, have developed a system referred to as the MODIS Fire Information System (FIS). The MODIS FIS is based on MODIS Active Fire Product (MOD14) Production Code, version 4.3.2, developed by NASA and automatically generates fire pixels or hotspot information with other physical parameters which would be useful to establish a better understanding of the fire occurrence phenomenon. The generated information is stored in a database, which could be accessed with ease via Internet. Details of the system, the associated algorithm producing the fire pixels information, and database access through the Internet are described in the following sections.

Overview of the MODIS Fire Information System

The AIT MODIS Fire Information System is a near-realtime automatic system, which consists of the following modules: MODIS Fire Product generation, MODIS fire visualization, MODIS Fire Database & Statistical analysis and MODIS fire validation systems. Click here to see the flowchart of the system (197 KB).

The purpose of this initial module of the system is to generate fire pixels using MODIS data using the software developed by NASA based on the NASA ATBD-MOD14 algorithm. The software could be downloaded freely from NASA Directout website. MODIS L1B data geolocation data is used as the input. The output data is the fire mask in HDF format. The fire mask includes the pixels which are classified into the following classes: missing data, cloud, water, non-fire, fire and unknown. From the fire mask, fire pixels could be extracted and stored in a text file. The fire pixels in the text file are arranged into columns of physical parameters which are granule lines and samples, geographic latitude and longitude of the file pixel, band 2 reflectance of pixel fire, band 21 or 22 brightness temperature of fire pixel, band 31 brightness temperature of fire pixel, total emitted power and the detection confidence of the fire pixels. The information from fire pixels text file are plotted to a base map in which the fire pixels are classified into three categories based on the fire detection confidence level. This information, including the base map is uploaded into the website daily. The fire pixels text file of each MODIS pass generated in this module would be important in the sense that it is to be utilized as input data for data processing in the subsequent modules of the system.

To understand and make use the fire phenomenon detected by MODIS sensors in different areas, it is necessary to have a tool supporting visual interpretation of the MODIS image in the interested area. Therefore, a visualization module is developed. The important task of this module is the generation of true color image with a resolution of 250 m, based on the True Color 250 m algorithm developed by NASA. The algorithm generates true color MODIS image of 250 m resolution in jpg format from Level 1B data. This image will be used to observe the smoke plume, haze and other fire-related phenomena. The true color MODIS image is used with high-resolution satellite images available in Google Earth to visualize the fire pixels and interpret the land cover. The advantage of this module is the combination of near-real time true color MODIS image of 250 m, which the surface reflectance correction was performed similarly to the MODIS product MOD09 except the effect of the aerosols intact, and the high-resolution satellites images available in Google Earth 3D viewer. The fire pixels text files are converted to kml format that could be overlaid with MODIS images and high-resolution images in Google Earth. See for an example of True Color MODIS image of 250m (800KB) which shows the smokes from forest fire in Southern part of Sumatra, Indonesia on October 06, 2006. More MODIS fire images can be seen here.

In order to perform further statistical analyses, the physical parameters in MODIS fire pixels are stored in a database designed to provide facilities for continued archiving and data retrieval. Also being web based, the database could conveniently be accessed via the Internet.

Click here to see a sample of statistical information related to the MODIS fire pixels distribution by countries for one scene (49.4KB).

It is possible develop a rapid response system to produce hotspots and to setup a web based validation system to verify the product by means of field inspection. Near real-time hotspot verification system is a kind of robust system producing geo-coded hotspot points rapidly including the field verification form to verify the spots via internet. It is totally based on open source technologies such as Minnesota Mapserver, Apache, PostgresSQL and PostGIS. From an economic perspective the system has the advantage that its development does not incur any cost for the software and once developed, the only need the stakeholders would be left with is to have access to internet.

Mr. Tun Tun Win (Database System), Mr. Kyaw Saan Oo, Mr. Win Myint Aung and Pornthip Limlahapun (Validation System)

References