MODIS C6 Active Fire Product User's Guide User A
User Manual:
Open the PDF directly: View PDF 
.
Page Count: 64
MODIS Collection 6 Active Fire Product User’s Guide
Revision A
Louis Giglio
Department of Geographical Sciences
University of Maryland
18 March 2015
Contents
1 Introduction 7
2 Summary of Collection 6 Algorithm and Product Changes 7
3 Overview of the MODIS Active Fire Products 8
3.1 Terminology...................................... 8
3.1.1 Granules ................................... 8
3.1.2 Tiles...................................... 8
3.1.3 Climate Modeling Grid (CMG) . . . . . . . . . . . . . . . . . . . . . . . 9
3.1.4 Collections .................................. 9
3.2 Level 2 Fire Products: MOD14 (Terra) and MYD14 (Aqua) . . . . . . . . . . . . . 9
3.3 Level 2G Daytime and Nighttime Fire Products: MOD14GD/MOD14GN (Terra)
and MYD14GD/MYD14GN (Aqua) . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.4 Level 3 8-Day Daily Composite Fire Products: MOD14A1 (Terra) and MYD14A1
(Aqua)......................................... 10
3.5 Level 3 8-Day Summary Fire Products: MOD14A2 (Terra) and MYD14A2 (Aqua) 11
3.6 Climate Modeling Grid Fire Products (MOD14CMQ, MYD14CMQ, etc.) . . . . . 12
3.7 Global Monthly Fire Location Product (MCD14ML) . . . . . . . . . . . . . . . . 12
3.8 Near Real-Time MODIS Imagery and Fire Products . . . . . . . . . . . . . . . . . 13
3.9 LDOPE Global Browse Imagery . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4 Obtaining the MODIS Active Fire Products 15
4.1 LAADS ........................................ 15
4.2 LP-DAAC....................................... 16
4.3 University of Maryland ftp Server . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.3.1 MODIS CMG Active-Fire Products . . . . . . . . . . . . . . . . . . . . . 18
4.3.2 MODIS Monthly Fire Location Product . . . . . . . . . . . . . . . . . . . 18
4.3.3 Documentation................................ 18
4.3.4 Collection 4 Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.3.5 VIRS Monthly Fire Product . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.4 LANCE Worldview and Rapid Response . . . . . . . . . . . . . . . . . . . . . . . 18
4.5 LANCEFIRMS.................................... 18
5 Detailed Product Descriptions 19
5.1 MOD14andMYD14................................. 19
5.1.1 FireMask................................... 19
5.1.2 Collection 6 Water Processing . . . . . . . . . . . . . . . . . . . . . . . . 19
5.1.3 Detection Confidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5.1.4 Algorithm Quality Assessment Bits . . . . . . . . . . . . . . . . . . . . . 20
5.1.5 FirePixelTable................................ 21
5.1.6 Metadata ................................... 22
5.1.7 ExampleCode ................................ 23
5.2 MOD14A1andMYD14A1.............................. 24
5.2.1 FireMask................................... 24
5.2.2 QA ...................................... 24
2
5.2.3 MaximumFRP................................ 24
5.2.4 ScanSample ................................. 24
5.2.5 Metadata ................................... 25
5.2.6 Level 3 Tile Navigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
5.2.7 ExampleCode ................................ 28
5.3 MOD14A2andMYD14A2.............................. 30
5.3.1 FireMask................................... 30
5.3.2 QA ...................................... 30
5.3.3 Level 3 Tile Navigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.3.4 ExampleCode ................................ 30
5.4 CMG Fire Products (MOD14CMQ, MYD14CMQ, etc.) . . . . . . . . . . . . . . . 33
5.4.1 CMG Naming Convention . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.4.2 DataLayers.................................. 34
5.4.3 GlobalMetadata ............................... 34
5.4.4 Climate Modeling Grid Navigation . . . . . . . . . . . . . . . . . . . . . 34
5.4.5 ExampleCode ................................ 35
5.5 Global Monthly Fire Location Product (MCD14ML) . . . . . . . . . . . . . . . . 36
5.5.1 Naming Convention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
5.5.2 ExampleCode ................................ 37
6 Validation of the MODIS Active Fire Products 39
6.1 Validation Based on ASTER Imagery . . . . . . . . . . . . . . . . . . . . . . . . 39
6.2 OtherValidation.................................... 39
7 Caveats and Known Problems 40
7.1 Caveats ........................................ 40
7.1.1 Fire Pixel Locations vs. Gridded Fire Products . . . . . . . . . . . . . . . 40
7.2 Collection 6 Known Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
7.2.1 Pre-November 2000 Data Quality . . . . . . . . . . . . . . . . . . . . . . 40
7.2.2 Detection Confidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
7.3 Collection 5 Known Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
7.3.1 False Alarms in Small Forest Clearings . . . . . . . . . . . . . . . . . . . 41
7.3.2 False Alarms During Calibration Maneuvers . . . . . . . . . . . . . . . . 42
8 Frequently Asked Questions 43
8.1 Terra and Aqua Satellites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
8.1.1 Where can I find general information about the Terra and Aqua satellites? . 43
8.1.2 When were the Terra and Aqua satellites launched? . . . . . . . . . . . . . 43
8.1.3 How can I determine overpass times of the Terra and Aqua satellites for a
particularlocation?.............................. 43
8.2 General MODIS Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
8.2.1 Where can I find Algorithm Technical Basis Documents (ATBDs) for the
MODISlandproducts?............................ 43
8.2.2 Do the MODIS sensors have direct broadcast capability? . . . . . . . . . . 43
8.3 General Fire Product Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
8.3.1 How are the fires and other thermal anomalies identified in the MODIS fire
productsdetected? .............................. 43
3
8.3.2 What is the smallest fire size that can be detected with MODIS? What about
thelargest? .................................. 44
8.3.3 Why didn’t MODIS detect a particular fire? . . . . . . . . . . . . . . . . . 44
8.3.4 How well can MODIS detect understory burns? . . . . . . . . . . . . . . . 44
8.3.5 Can MODIS detect fires in unexposed coal seams? . . . . . . . . . . . . . 44
8.3.6 How do I obtain the MODIS fire products? . . . . . . . . . . . . . . . . . 44
8.3.7 What validation of the MODIS active fire products has been performed? . . 45
8.3.8 I don’t want to bother with strange file formats and/or an unfamiliar ordering
interface and/or very large data files. Can’t you just give me the locations
of fire pixels in plain ASCII files and I’ll bin them myself? . . . . . . . . . 45
8.3.9 I want to estimate burned area using active fire data. What effective area
burned should I assume for each fire pixel? . . . . . . . . . . . . . . . . . 45
8.3.10 Why are some of the MODIS fire products not available prior to November
2000?..................................... 45
8.3.11 Why then are the Level 2 swath and Level 3 tiled fire products available
before November 2000? . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
8.4 Level2FireProducts................................. 45
8.4.1 Why do the Level 2 product files vary in size? . . . . . . . . . . . . . . . . 45
8.4.2 How should the different fire detection confidence classes be used? . . . . 46
8.4.3 How are the confidence values in the “FP confidence” SDS related to the
confidence classes assigned to fire pixels? . . . . . . . . . . . . . . . . . . 46
8.4.4 How can I take data from the fire-pixel-table SDSs (i.e., the one-dimensional
SDSs with the prefix “FP ”) and place the values in the proper locations of
a two-dimensional array that matches the swath-based “fire mask” and “al-
gorithmQA”SDSs? ............................. 46
8.4.5 Why are the values of the fire radiative power (FRP) in the Collection 4
Level 2 product inconsistent with those in the Collection 5 Level 2 product? 47
8.4.6 What is the area of a MODIS pixel at the Earth’s surface? . . . . . . . . . 48
8.4.7 Can I use cloud pixels identified in the Level 2 fire product as a general-
purpose cloud mask for other applications? . . . . . . . . . . . . . . . . . 48
8.5 Level 3 Tiled Fire Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
8.5.1 Why do coastlines in the tile-based Level 3 products looked so warped? . . 49
8.5.2 Is there an existing tool I can use to reproject the tiled MODIS products into
a different projection? . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
8.5.3 Why do some MOD14A1 and MYD14A1 product files have fewer than
eight daily data layers? . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
8.5.4 How can I determine the date associated with each daily composite in the
MOD14A1 and MYD14A1 products when fewer than eight days of data are
present?.................................... 49
8.5.5 How do I calculate the latitude and longitude of a grid cell in the Level 3
products?................................... 50
8.5.6 How do I calculate the tile and grid cell coordinates of a specific geographic
location (latitude and longitude)? . . . . . . . . . . . . . . . . . . . . . . 50
8.5.7 What size are the grid cells of Level 3 MODIS sinusoidal grid? . . . . . . 50
8.6 Level 3 CMG Fire Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4
8.6.1 I need to reduce the resolution of the 0.25◦CMG fire product to grid cells
that are a multiple of 0.25◦in size. How do I go about doing this? . . . . . 51
8.6.2 Why don’t you distribute a daily CMG fire product? . . . . . . . . . . . . 53
8.6.3 Why don’t you distribute the CMG fire products as plain binary (or ASCII)
files? ..................................... 53
8.6.4 Where can I find information about the FITS file format? . . . . . . . . . . 53
8.6.5 What software libraries are available for reading FITS files? . . . . . . . . 53
8.6.6 How can I display images in FITS files? . . . . . . . . . . . . . . . . . . . 53
8.6.7 Does the last 8-day CMG product for each calendar year include data from
the first few days of the following calendar year? . . . . . . . . . . . . . . 57
8.6.8 Where can I find details about the different corrections performed on some
of the data layers in the CMG fire products? . . . . . . . . . . . . . . . . . 57
8.6.9 Are persistent hot spots filtered out of the MODIS CMG fire products? . . 57
8.6.10 Is there an easy way to convert a calendar date into the ordinal dates (day-
of-year) used in the file names of the 8-day fire products? . . . . . . . . . . 57
8.7 Global Monthly Fire Location Product . . . . . . . . . . . . . . . . . . . . . . . . 59
8.7.1 Can I use the MCD14ML fire location product to make my own gridded fire
dataset?.................................... 59
8.7.2 How many lines are in each MCD14ML product file? . . . . . . . . . . . . 59
8.7.3 Are persistent hot spots filtered out of the fire location product? . . . . . . 59
8.7.4 The MCD14ML ASCII product files have fixed-width, space-delimited fields.
Is there an easy way to convert these to comma-separated values (CSV) files? 59
8.7.5 How can I compute the scan angle given the sample number in the MCD14ML
product?.................................... 59
8.8 Hierarchical Data Format (HDF) . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
8.8.1 WhatareHDFfiles? ............................. 60
8.8.2 How do I read HDF4 files? . . . . . . . . . . . . . . . . . . . . . . . . . . 60
8.8.3 Can’t I just skip over the HDF header and read the data directly? . . . . . . 60
8.8.4 How can I list the contents of HDF4 files? . . . . . . . . . . . . . . . . . . 60
8.8.5 How can I display images in HDF4 files? . . . . . . . . . . . . . . . . . . 60
9 References 61
10 Relevant Web and FTP Sites 63
5
List of Tables
1 MODIS fire product availability. . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2 MOD14/MYD14 fire mask pixel classes. . . . . . . . . . . . . . . . . . . . . . . . 19
3 Summary of Level-2 fire product pixel-level QA bits. . . . . . . . . . . . . . . . . 20
4 Collection 6 Level 2 fire product SDSs comprising the “fire pixel table”. MAD =
“mean absolute deviation”. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
5 MODIS Level 2 fire product metadata stored as standard global HDF attributes. . . 22
6 QA values in the Collection 6 MODIS Level 3 tiled active fire products. . . . . . . 24
7 MOD14A1 and MYD14A1 fire product metadata stored as standard global HDF
attributes. ....................................... 26
8 Summary of data layers in the CMG fire products. . . . . . . . . . . . . . . . . . . 34
9 Summary of columns in the MCD14ML fire location product. . . . . . . . . . . . 36
10 Fire-pixel confidence classes associated with the confidence level Ccomputed for
eachfirepixel...................................... 46
11 Sizes of grid cells in Level 3 tiled MODIS sinusoidal grid. . . . . . . . . . . . . . 50
12 Calendar dates (month/day) corresponding to the day-of-year (DOY) beginning
each 8-day time period for which the 8-day fire products are generated. Dates for
non-leap years and leap years are shown separately. . . . . . . . . . . . . . . . . . 58
6
1 Introduction
This document contains the most current information about the Collection 6 Terra and Aqua Moder-
ate Resolution Imaging Spectrometer (MODIS) fire products. It is intended to provide the end user
with practical information regarding their use and misuse, and to explain some of the more obscure
and potentially confusing aspects of the fire products and MODIS products in general.
2 Summary of Collection 6 Algorithm and Product Changes
1. Processing has been extended to oceans and other large water bodies to detect offshore gas
flaring.
2. Reduced incidence of false alarms caused by small forest clearings.
3. Improved detection of small fires.
4. Expanded sun-glint rejection.
5. Slightly improved cloud masking.
6. Slight adjustment of detection confidence calculation.
7. FRP retrieval now performed using Wooster et al. (2003) approach.
8. Expanded fire pixel table in Level 2 product.
9. Additional granule-level metadata in Level 2 product.
10. Simplified layer date information in 1-km Level 3 daily fire product.
11. Simplified QA layer in 1-km Level 3 8-day and daily fire products.
12. 0.25◦CMG products (available summer 2015).
13. Improved elimination of static hot-spot sources from CMG products (available summer 2015).
14. New hot-spot type attribute in MCD14ML fire location product (available summer 2015).
7
3 Overview of the MODIS Active Fire Products
Here we provide a general overview of the MODIS active fire products. More detailed descriptions
of these products and example ingest code can be found in Section 4.
3.1 Terminology
Before proceeding with a description of the MODIS fire products, we briefly define the terms gran-
ule,tile, and collection, and climate modelling grid in the context of these products.
3.1.1 Granules
Agranule is an unprojected segment of the MODIS orbital swath containing about 5 minutes of
data. MODIS Level 0, Level 1, and Level 2 products are granule-based.
3.1.2 Tiles
MODIS Level 2G, Level 3, and Level 4 products are defined on a global 250-m, 500-m, or 1-km
sinusoidal grid (the particular spatial resolution is product-dependent). Because these grids are un-
manageably large in their entirety (43200 ×21600 pixels at 1 km, and 172800 ×86400 pixels at
250 m), they are divided into fixed tiles approximately 10◦×10◦in size. Each tile is assigned a
horizontal (H) and vertical (V) coordinate, ranging from 0 to 35 and 0 to 17, respectively (Figure 1).
The tile in the upper left (i.e. northernmost and westernmost) corner is numbered (0,0).
Figure 1: MODIS tiling scheme.
8
3.1.3 Climate Modeling Grid (CMG)
MODIS Level 3 and Level 4 products can also be defined on a coarser-resolution climate modelling
grid (CMG). The objective is to provide the MODIS land products at consistent low resolution spa-
tial and temporal scales suitable for global modeling. In practice, there is a fair amount of variation
in the spatial and temporal gridding conventions used among the MODIS land CMG products.
3.1.4 Collections
Reprocessing of the entire MODIS data archive is periodically performed to incorporate better cali-
bration, algorithm refinements, and improved upstream data into all MODIS products. The updated
MODIS data archive resulting from each reprocessing is referred to as a collection. Later collections
supersede all earlier collections.
For the Terra MODIS, Collection 1 consisted of the first products generated following launch.
Terra MODIS data were reprocessed for the first time in June 2001 to produce Collection 3. (Note
that this first reprocessing was numbered Collection 3 rather than, as one would expect, Collec-
tion 2.) Collection 3 was also the first version produced for the Aqua MODIS products. Collec-
tion 4 reprocessing began in December 2002 and was terminated in December 2006. Production
of the Collection 5 products, which commenced in mid-2006, will continue until the Collection 6
reprocessing is complete. Production of the “Tier-1” Collection 6 MODIS products, which includes
the active fire products, commenced in February 2015.
3.2 Level 2 Fire Products: MOD14 (Terra) and MYD14 (Aqua)
This is the most basic fire product in which active fires and other thermal anomalies, such as volca-
noes, are identified. The Level 2 product is defined in the MODIS orbit geometry covering an area
of approximately 2340 ×2030 km in the along-scan and along-track directions, respectively. It is
used to generate all of the higher-level fire products, and contains the following components:
•An active fire mask that flags fires and other relevant pixels (Figure 2);
•a pixel-level quality assurance (QA) image that includes 19 bits of QA information about each
pixel;
•a fire-pixel table which provides 27 separate pieces of radiometric and internal-algorithm
information about each fire pixel detected within a granule;
•extensive mandatory and product-specific metadata;
•a grid-related data layer to simplify production of the Climate Modeling Grid (CMG) fire
product (Section 3.6).
Product-specific metadata within the Level 2 fire product includes the number of cloud, water,
non-fire, fire, unknown, and other pixels occurring within a granule to simplify identification of
granules containing fire activity.
9
Figure 2: Example MOD14 (Terra MODIS) swath-
level fire mask for granule acquired 5 September
2002 at 07:20 UTC, with water shown in blue,
clouds in purple, non-fire land pixels in grey, and
fire pixels in white. The along-track direction points
toward the bottom of the page. The large land mass
on the right is Madagascar.
3.3 Level 2G Daytime and Nighttime Fire Products: MOD14GD/MOD14GN (Terra)
and MYD14GD/MYD14GN (Aqua)
The Level 2 active fire products sensed over daytime and nighttime periods are binned without
resampling into an intermediate data format referred to as Level 2G. The Level 2G format provides
a convenient geocoded data structure for storing granules and enables the flexibility for subsequent
temporal compositing and reprojection. The Level 2G fire products are a temporary, intermediate
data source used solely for producing the Level 3 fire products and are consequently not available
from the permanent MODIS data archive.
3.4 Level 3 8-Day Daily Composite Fire Products: MOD14A1 (Terra) and MYD14A1
(Aqua)
The MODIS daily Level 3 fire product is tile based, with each product file spanning one of the 460
MODIS tiles, 326 of which contain land pixels. The product is a 1-km gridded composite of fire
pixels detected in each grid cell over each daily (24-hour) compositing period. For convenience,
eight days of data are packaged into a single file.
Figure 3 shows the Terra fire mask for day 7 (4 September 2001) of the 29 August–5 September
2001 daily Level 3 fire product. The tile is located in Northern Australia (h31v10).
10
Figure 3: Example of 4 September 2001 Col-
lection 6 MOD14A1 daily fire mask for tile
h31v10, located in Northern Australia. Wa-
ter is shown in blue, clouds in violet, non-fire
land grid cells in grey, and fire grid cells in
white. Grid cells lacking data are shown in
black.
3.5 Level 3 8-Day Summary Fire Products: MOD14A2 (Terra) and MYD14A2 (Aqua)
The MODIS daily Level 3 8-day summary fire product is tile-based, with each product file spanning
one of the 460 MODIS tiles, of which 326 contain land pixels. The product is a 1-km gridded
composite of fire pixels detected in each grid cell during the 8-day compositing period.
Figure 4 shows the 8-day summary fire mask from the 8-day Level 3 Terra fire product spanning
29 August–5 September 2001. As in the previous example, the tile is located in Northern Australia
(h31v10). The 8-day composite is the maximum value of the individual Level 2 pixel classes that
fell into each 1-km grid cell over the entire 8-day compositing period.
Figure 4: Example of 8-day MOD14A2
summary fire mask for MODIS tile h31v10
spanning 29 August–5 September 2001. The
color scale is the same as that of Figure 3.
11
3.6 Climate Modeling Grid Fire Products (MOD14CMQ, MYD14CMQ, etc.)
The CMG fire products are gridded statistical summaries of fire pixel information intended for use
in regional and global modeling. The products are now generated at 0.25◦spatial resolution for time
periods of one calendar month (MOD14CMQ and MYD14CMQ) and eight days (MOD14C8Q and
MYD14C8Q). An example of the corrected fire pixel count layer of the product is shown in Figure 5.
Figure 5: Example of the corrected fire pixel count data layer from the January 2001 Terra MODIS
monthly CMG fire product.
3.7 Global Monthly Fire Location Product (MCD14ML)
For some applications it is necessary to have the geographic coordinates of individual fire pixels.
New for Collection 5 is the global monthly fire location product (MCD14ML), which contains this
information for all Terra and Aqua MODIS fire pixels in a single monthly ASCII file.
12
3.8 Near Real-Time MODIS Imagery and Fire Products
The Land Atmosphere Near Real-time Capability for EOS (LANCE) Rapid Response System pro-
duces near-real time global imagery including true- and false-color corrected reflectance superim-
posed with fire locations (Figure 6), Normalized Difference Vegetation Index (NDVI), and land
surface temperature. Near-real time locations of Terra and Aqua MODIS fire pixels are also avail-
able as text files.
Figure 6: MODIS Rapid Response true color imagery of fires and smoke in southeast Australia (10
December 2006, 03:45 UTC).
13
3.9 LDOPE Global Browse Imagery
The MODIS Land Data Operational Product Evaluation (LDOPE) provides interactive daily global
browse imagery of many MODIS land products from the MODIS Land Global Browse Images web
site1in near-real time (Figure 7). For most products (including the fire products) the browse im-
agery is generated using only the daytime overpasses. The site allows you to arbitrarily zoom into
any region of the globe and examine features of interest in more detail.
Figure 7: Example Collection-6 Aqua MODIS active fire global browse image for 28 December
2002 showing all daytime overpasses. Fire pixels are shown in red, cloud pixels are shown in light
blue, and areas lacking data are shown in white. Browse image courtesy of the LDOPE.
1http://landweb.nascom.nasa.gov/cgi-bin/browse/browse.cgi
14
4 Obtaining the MODIS Active Fire Products
All MODIS products are available to users free of charge through several different sources (Table 1).
Not all products are available from each source.
Table 1: MODIS fire product availability.
Product Source
Level 2 and Level 3 fire products: LAADS (Section 4.1) and LP-DAAC
MOD14, MYD14 (Section 4.2)
MOD14A1, MYD14A1
MOD14A2, MYD14A2
CMG fire products: University of Maryland (Section 4.3)
MOD14CMQ, MYD14CMQ
MOD14C8Q, MYD14C8Q
Global fire location product: University of Maryland (Section 4.3)
MCD14ML
Near real-time fire and corrected reflectance im-
agery.
LANCE Worldview and Rapid Response
(Section 4.4)
Geographic subsets of near real-time active fire
locations in various GIS-compatible formats.
LANCE FIRMS (Section 4.5)
4.1 LAADS
The MODIS Level 1, atmosphere, and land products may be obtained from the Level 1 and Atmo-
sphere Archive and Distribution System (LAADS), available here:
http://ladsweb.nascom.nasa.gov/index.html
15
4.2 LP-DAAC
Most of the MODIS land products may be obtained from the Land Processes Distributed Active
Archive Center (LP-DAAC) using a Web-based interface known as Reverb, a replacement for the
Warehouse Inventory Search Tool (WIST), and before that the older EOS Data Gateway. Reverb
may be found at the following URL:
https://reverb.echo.nasa.gov/
A Reverb tutorial is available here:
http://www.echo.nasa.gov/reverb/tutorial/Tutorial.html
Figure 8: The Reverb interface.
16
4.3 University of Maryland ftp Server
At present the active-fire CMG products and the MCD14ML product are distributed from an ftp
server at the University of Maryland. Log in using the following information:
Server: fuoco.geog.umd.edu
Login name: fire
Password: burnt
Once connected, you will have access to the following directory tree:
.
|-- db
| ‘-- MCD64A1
|-- gfed -> gfed4
|-- gfed3
| ‘-- monthly
| ‘-- hdf
|-- gfed4
| |-- daily
| ‘-- monthly
|-- modis
| |-- C4
| |-- C5
| | |-- cmg
| | | |-- 8day
| | | | |-- fits
| | | | ‘-- hdf
| | | ‘-- monthly
| | | |-- fits
| | | ‘-- hdf
| | ‘-- mcd14ml
| |-- C6
| | |-- cmg
| | | |-- 8day
| | | | |-- fits
| | | | ‘-- hdf
| | | ‘-- monthly
| | | |-- fits
| | | ‘-- hdf
| | ‘-- mcd14ml
| ‘-- docs
‘-- virs
‘-- monthly
17
4.3.1 MODIS CMG Active-Fire Products
The current MODIS CMG fire products are located in the directory modis/C6/cmg. For con-
venience, these products are distributed in multiple, standard data formats. Currently HDF and
Flexible Image Transport System (FITS) files are available; additional formats may be produced in
the future.
4.3.2 MODIS Monthly Fire Location Product
The current MCD14ML product is located in the directory modis/C6/mcd14ml.
4.3.3 Documentation
The most recent version of the Active Fire Product User’s Guide for each Collection is archived in
the directory modis/docs.
4.3.4 Collection 4 Products
To help ensure traceability and replicability the server provides an archive of the older Collection 4
CMG fire products in the directory modis/C4.These products are obsolete and should not be
used for new research or analyses.
4.3.5 VIRS Monthly Fire Product
Although unrelated to MODIS, the ftp server also hosts an archive of the 0.5◦Tropical Rainfall Mea-
suring Mission (TRMM) Visible and Infrared Scanner (VIRS) monthly fire product in the directory
virs/monthly. See the documentation in the directory virs for details.
4.4 LANCE Worldview and Rapid Response
Near real-time fire and corrected reflectance imagery are available from the NASA’s Land, At-
mosphere Near real-time Capability for EOS (LANCE) Worldview and Rapid Response systems,
located here:
https://earthdata.nasa.gov/lance
Note: In general, the near real-time MODIS products should not be used for time series analyses or
long-term studies.
4.5 LANCE FIRMS
Near real-time MODIS fire locations are available in a variety of formats (ASCII, shapefile, KML,
or WMS) from the LANCE Fire Information for Resource Management System (FIRMS), located
here:
https://earthdata.nasa.gov/data/near-real-time-data/firms
Note: In general, the near real-time MODIS fire locations should not be used for time series analyses
or long-term studies; for such purposes the standard MCD14ML product is more appropriate. By
special request, FIRMS will distribute spatial subsets of MODIS fire locations extracted from the
standard MCD14ML product via the Archive Download Tool.
18
5 Detailed Product Descriptions
5.1 MOD14 and MYD14
MOD14/MYD14 is the most basic fire product in which active fires and other thermal anomalies,
such as volcanoes, are identified. The Level 2 product is defined in the MODIS orbit geometry
covering an area of approximately 2340 by 2030 km in the across- and along-track directions, re-
spectively. It is used to generate all of the higher-level fire products.
5.1.1 Fire Mask
The fire mask is the principle component of the Level 2 MODIS fire product, and is stored as an
8-bit unsigned integer Scientific Data Set (SDS) named “fire mask”. In it, individual 1-km pixels
are assigned one of nine classes. The meaning of each class is listed in Table 2.
Table 2: MOD14/MYD14 fire mask pixel classes.
Class Meaning
0 not processed (missing input data)
1 not processed (obsolete; not used since Collection 1)
2 not processed (other reason)
3 non-fire water pixel
4 cloud (land or water)
5 non-fire land pixel
6 unknown (land or water)
7 fire (low confidence, land or water)
8 fire (nominal confidence, land or water)
9 fire (high confidence, land or water)
5.1.2 Collection 6 Water Processing
For Collection 6, oceans and other large water bodies are now processed to detect offshore gas
flaring. As such, the cloud,unknown, and fire pixel classes can now occur over water. While it
was safe to use the fire mask as a rudimentary land/water mask prior to Collection 6, users must
now appeal to bits 0–1 of the pixel-level algorithm QA layer (Section 5.1.4) to unambiguously
discriminate land from water pixels.
5.1.3 Detection Confidence
A detection confidence intended to help users gauge the quality of individual fire pixels is included
in the Level 2 fire product. This confidence estimate, which ranges between 0% and 100%, is used to
assign one of the three fire classes (low-confidence fire,nominal-confidence fire, or high-confidence
fire) to all fire pixels within the fire mask.
In some applications errors of commission (or false alarms) are particularly undesirable, and
for these applications one might be willing to trade a lower detection rate to gain a lower false alarm
rate. Conversely, for other applications missing any fire might be especially undesirable, and one
19
might then be willing to tolerate a higher false alarm rate to ensure that fewer true fires are missed.
Users requiring fewer false alarms may wish to retain only nominal- and high-confidence fire pixels,
and treat low-confidence fire pixels as clear, non-fire, land pixels. Users requiring maximum fire
detectability who are able to tolerate a higher incidence of false alarms should consider all three
classes of fire pixels.
5.1.4 Algorithm Quality Assessment Bits
Pixel-level QA is stored in a 32-bit unsigned integer SDS named “algorithm QA”, with individual
fields stored in specific bits (Table 3). For details, please see the MODIS Level 2 Fire Product file
specification.
Table 3: Summary of Level-2 fire product pixel-level QA bits.
Bit(s) Meaning
0-1 land/water state (00 = water, 01 = coast, 10 = land, 11 = unused)
2 3.9 µm high-gain flag (0 = band 21, 1 = band 22)
3 atmospheric correction performed (0 = no, 1 = yes)
4 day/night algorithm (0 = night, 1 = day)
5 potential fire pixel (0 = false, 1 = true)
6 spare (set to 0)
7-10 background window size parameter
11-16 individual detection test flags (0 = fail, 1 = pass)
17-19 spare (set to 0)
20 adjacent cloud pixel (0 = no, 1 = yes)
21 adjacent water pixel (0 = no, 1 = yes)
22-23 sun glint level (0–3)
24-28 individual rejection test flags (0 = false, 1 = true)
29-31 spare (set to 0)
20
5.1.5 Fire Pixel Table
The fire pixel table is simply a collection of SDSs containing relevant information about individual
fire pixels detected within a granule. Due to HDF file format and library limitations, the Fire Pixel
Table is stored as 27 separate SDSs. A brief summary of these SDSs is provided in Table 3.
Table 4: Collection 6 Level 2 fire product SDSs comprising the “fire pixel table”. MAD = “mean
absolute deviation”.
SDS Name Data Type Units Description
FP line int16 - Granule line of fire pixel.
FP sample int16 - Granule sample of fire pixel.
FP latitude float32 degrees Latitude at center of fire pixel.
FP longitude float32 degrees Longitude at center of fire pixel.
FP R2 float32 - Near-IR (band 2) reflectance of fire pixel (daytime
only).
FP T21 float32 K Channel 21/22 brightness temperature of fire
pixel.
FP T31 float32 K Channel 31 brightness temperature of fire pixel.
FP MeanT21 float32 K Background channel 21/22 brightness tempera-
ture.
FP MeanT31 float32 K Background channel 31 brightness temperature.
FP MeanDT float32 K Background brightness temperature difference.
FP MAD T21 float32 K
FP MAD T31 float32 K
FP MAD DT float32 K
FP power float32 MW Fire radiative power.
FP AdjCloud uint8 - Number of adjacent cloud pixels.
FP AdjWater uint8 - Number of adjacent water pixels.
FP WinSize uint8 - Background window size.
FP NumValid int16 - Number of valid background pixels.
FP confidence uint8 % Detection confidence estimate.
FP land uint8 - Land flag (0 = water pixel, 1 = land pixel).
FP MeanR2 float32 - Background channel 2 reflectance.
FP MAD R2 float32 - Background channel 2 reflectance MAD.
FP ViewZenAng float32 degrees View zenith angle.
FP SolZenAng float32 degrees Solar zenith angle.
FP RelAzAng float32 degrees Relative azimuth angle.
FP CMG row int16 - CMG row.
FP CMG col int16 - CMG column.
21
5.1.6 Metadata
Every MODIS product carries with it ECS-mandated metadata stored in the HDF global attributes
CoreMetadata.0 and ArchiveMetadata.0. Each attribute is an enormous string of ASCII characters
encoding many separate metadata fields in Parameter Value Language (PVL). Among other infor-
mation, the ArchiveMetadata.0 attribute usually contains product-specific metadata included at the
discretion of the PI. However, since the PVL is awkward to read and tedious to parse, we have stored
many of the product-specific metadata fields as standard HDF global attributes. These are summa-
rized in Table 5. Descriptions of the product-specific metadata stored in the ECS ArchiveMetadata.0
attribute may be found in the MOD14/MYD14 Fire Product file specification (see Section 10).
Table 5: MODIS Level 2 fire product metadata stored as standard global HDF attributes.
Attribute Name Description
FirePix Number of fire pixels detected in granule.
MissingPix Number of pixels in granule lacking valid data for processing.
LandPix Number of land pixels in granule.
WaterPix Number of water pixels in granule.
CoastPix Number of coast pixels in granule.
WaterAdjacentFirePix Number of fire pixels that are adjacent to one or more water pixels.
CloudAdjacentFirePix Number of fire pixels that are adjacent to one or more cloud pixels.
UnknownLandPix Number of land pixels assigned a class of unknown.
UnknownWaterPix Number of water pixels assigned a class of unknown.
LandCloudPix Number of land pixels obscured by cloud in granule.
WaterCloudPix Number of water pixels obscured by cloud in granule.
GlintPix Number of pixels in granule contaminated with sun glint.
GlintRejectedPix Number of tentative fire pixels that were rejected due to apparent
sun glint contamination.
CoastRejectedPix Number of tentative fire pixels that were rejected due to apparent
water contamination of the contextual neighborhood.
HotSurfRejectedPix Number of tentative fire pixels that were rejected as apparent hot
desert surfaces.
ClearingRejectedPix Number of tentative fire pixels rejected as apparent forest clear-
ings.
CoastRejectedWaterPix Number of tentative fire pixels rejected due to apparent land con-
tamination of contextual background.
DayPix Number of daytime pixels in granule.
NightPix Number of nighttime pixels in granule.
Satellite Name of satellite (“Terra” or “Aqua”).
ProcessVersionNumber Production code version string (e.g. “6.2.3”).
MOD021KM input file File name of MOD021KM (Terra) or MYD021KM (Aqua)
Level 1B radiance input granule.
MOD03 input file File name of MOD03 (Terra) or MYD03 (Aqua) geolocation input
granule.
22
5.1.7 Example Code
Example 1: IDL code for reading the “fire mask” SDS in the MODIS Level 2 fire product.
mod14_file = ’MOD14.A2002177.1830.005.2008192223417.hdf’
; open the HDF file for reading
sd_id = HDF_SD_START(mod14_file, /READ)
; find the SDS index to the MOD14 fire mask
index = HDF_SD_NAMETOINDEX(sd_id, ’fire mask’)
; select and read the entire fire mask SDS
sds_id = HDF_SD_SELECT(sd_id, index)
HDF_SD_GETDATA, sds_id, fire_mask
; finished with SDS
HDF_SD_ENDACCESS, sds_id
; finished with HDF file
HDF_SD_END, sd_id
23
5.2 MOD14A1 and MYD14A1
The MOD14A1 and MYD14A1 daily Level 3 fire products are tile-based, with each product file
spanning one of the 460 MODIS tiles, of which 326 contain land pixels. The product is a 1-km
gridded composite of fire pixels detected in each grid cell over each daily (24-hour) compositing
period. For convenience, eight days of data are packaged into a single file.
5.2.1 Fire Mask
The fire mask is stored as an 8 (or less) ×1200 ×1200, 8-bit unsigned integer SDS named “Fire-
Mask”. (For historical reasons this layer was named “most confident detected fire” prior to Collec-
tion 5.) The SDS contains eight successive daily fire masks for a specific MODIS tile. Each of these
daily masks is essentially a maximum value composite2of the Level 2 fire product pixel classes
(Table 2) for those swaths overlapping the MODIS tile during that day. Product files containing less
than eight days of data will occasionally be encountered during time periods of missing data and
should be handled in ingest software.
5.2.2 QA
Each of the daily fire masks has a corresponding simple QA layer. Each layer is a 1200 ×1200
8-bit unsigned integer array. Only seven unique QA values are possible, with the meanings shown
in Table 6. Note that for missing-data grid cells (bit pattern 11 in bits 0-1), bit 2 will always be clear.
Table 6: QA values in the Collection 6 MODIS Level 3 tiled active fire products.
Bit(s) Meaning
0-1 land/water state (00 = water, 01 = coast, 10 = land, 11 = missing data)
2 day/night observation (0 = night, 1 = day)
5.2.3 Maximum FRP
The maximum fire radiative power of all fire pixels falling within each grid cell is provided on a
daily basis in the “MaxFRP” SDS. Here the FRP values have been scaled by a factor of 10 and
stored as a 32-bit signed integer. Multiply these scaled values by 0.1 to retrieve the maximum FRP
in MW.
5.2.4 Scan Sample
For all grid cells assigned one of the fire pixel classes (values 7, 8, or 9), the position of the fire pixel
within the scan is recorded on a daily basis in a 1200 ×1200 16-bit unsigned integer SDS named
“sample”. Sample values have a range of 0 to 1353. All grid cells assigned one of the non-fire
classes in the “FireMask” SDS will be filled with a sample value of 0.
2Due to the introduction of water processing in Collection 6, the original maximum value compositing scheme had to
be modified slightly to prevent cloud-obscured water grid cells from having precedence over cloud-free water grid cells.
24
5.2.5 Metadata
As with the Level 2 fire products, the MOD14A1 and MYD14A1 products contain global
metadata stored in the ECS CoreMetadata.0 and ArchiveMetadata.0 global attributes. Also
like the Level 2 products, a subset of these metadata are written as standard HDF global
attributes for convenience (see Table 7). Here are example values for the product file
“MOD14A1.A2002297.h31v10.006.2015058030918.hdf”, listed using ncdump (see Section
8.8.4):
:FirePix = 482, 319, 329, 386, 660, 739, 1096, 477 ;
:CloudPix = 76933, 83531, 61942, 67212, 36943, 6980, 46727,
15949 ;
:UnknownPix = 450, 168, 113, 96, 711, 190, 192, 225 ;
:MissingPix = 53920, 155798, 4, 163180, 0, 163250, 0, 149596 ;
:Dates = "2002-10-24 2002-10-25 2002-10-26 2002-10-27
2002-10-28 2002-10-29 2002-10-30 2002-10-31" ;
:MaxT21 = 425.2746f ;
:ProcessVersionNumber = "6.0.1" ;
:StartDate = "2002-10-24" ;
:EndDate = "2002-10-31" ;
:HorizontalTileNumber = 31s ;
:VerticalTileNumber = 10s ;
Notice that the first four fields (FirePix,CloudPix,UnknownPix, and MissingPix) are
one-dimensional arrays (or vectors) nominally having eight elements. Each element corresponds
to a single day in the 8-day time period covered by the product. Note that the Scientific Data Sets
in the product file (“FireMask”, “MaxFRP”, etc.) will contain fewer than 8 planes when there are
no valid MODIS observations during one or more days spanned by the product. In such cases, the
vector metadata fields will have fewer than eight elements.
25
Table 7: MOD14A1 and MYD14A1 fire product metadata stored as standard global HDF attributes.
Attribute Name Description
FirePix Number of 1-km tile cells containing fires (8-element array).
CloudPix Number of 1-km tile cells assigned a class of cloud after composit-
ing (8-element array).
UnknownPix Number of 1-km tile cells assigned a class of unknown after com-
positing (8-element array).
MissingPix Number of 1-km tile cells lacking valid data (8-element array).
Dates Date of each plane in three-dimensional fire mask array.
MaxT21 Maximum band 21 brightness temperature (K) of all fire pixels in
tile.
ProcessVersionNumber Production code version string (e.g. “6.0.1”).
StartDate Start date of 8-day time period spanned by product (YYYY-MM-
DD).
EndDate End date of 8-day time period spanned by product (YYYY-MM-
DD).
HorizontalTileNumber Horizontal tile coordinate (H).
VerticalTileNumber Vertical tile coordinate (V).
26
5.2.6 Level 3 Tile Navigation
Navigation of the tiled MODIS products in the sinusoidal projection can be performed using the for-
ward and inverse mapping transformations described here. We’ll first need to define a few constants:
R= 6371007.181 m, the radius of the idealized sphere representing the Earth;
T= 1111950 m, the height and width of each MODIS tile in the projection plane;
xmin = -20015109 m, the western limit of the projection plane;
ymax = 10007555 m, the northern limit of the projection plane;
w=T/1200 = 926.62543305 m, the actual size of a “1-km” MODIS sinusoidal grid cell.
Forward Mapping
Denote the latitude and longitude of the location (in radians) as φand λ, respectively. First compute
the position of the point on the global sinusoidal grid:
x=Rλ cos φ(1)
y=Rφ. (2)
Next compute the horizontal (H) and vertical (V) tile coordinates, where 0≤H≤35 and 0≤
V≤17 (Section 3.1.2):
H=x−xmin
T(3)
V=ymax −y
T,(4)
where ⌊⌋ is the floor function. Finally, compute the row (i) and column (j) coordinates of the grid
cell within the MODIS tile:
i=(ymax −y) mod T
w−0.5(5)
j=(x−xmin) mod T
w−0.5.(6)
Note that for the 1-km MOD14A1 and MYD14A1 products (indeed, all 1-km MODIS products on
the sinusoidal grid) 0≤i≤1199 and 0≤j≤1199.
27
Inverse Mapping
Here we are given the row (i) and column (j) in MODIS tile H,V. First compute the position of
the center of the grid cell on the global sinusoidal grid:
x= (j+ 0.5)w+HT +xmin (7)
y=ymax −(i+ 0.5)w−V T (8)
Next compute the latitude φand longitude λat the center of the grid cell (in radians):
φ=y
R(9)
λ=x
Rcos φ.(10)
Applicability to 250-m and 500-m MODIS Products
With the following minor changes the above formulas are also applicable to the higher resolution
250-m and 500-m MODIS tiled sinusoidal products.
250-m grid: Set w=T /4800 = 231.65635826 m, the actual size of a “250-m” MODIS sinusoidal
grid cell. For 250-m grid cells 0≤i≤4799 and 0≤j≤4799.
500-m grid: Set w=T /2400 = 463.31271653 m, the actual size of a “500-m” MODIS sinusoidal
grid cell. For 500-m grid cells 0≤i≤2399 and 0≤j≤2399.
5.2.7 Example Code
Example 2: MATLAB code to read the Level 3 MODIS daily fire mask, using the MATLAB routine
hdfread. This is probably the easiest way to read individual HDF SDSs in MATLAB.
mod14a1_file = ’MOD14A1.A2008281.h31v10.005.2008292070548.hdf’
% read entire "FireMask" SDS in one shot
fire_mask = hdfread(mod14a1_file, ’FireMask’);
% display fire mask for the first day in MOD14A1/MYD14A1
% note how image is transposed so that North appears on top
imagesc(fire_mask(:,:,1)’);
28
Example 3: IDL code to read some of the global attributes and SDSs in the Level 3 daily fire
product.
mod14a1_file = ’MOD14A1.A2007241.h08v05.005.2007251120334.hdf’
sd_id = HDF_SD_START(mod14a1_file, /READ)
; read "FirePix" and "MaxT21" attributes
attr_index = HDF_SD_ATTRFIND(sd_id, ’FirePix’)
HDF_SD_ATTRINFO, sd_id, attr_index, DATA=FirePix
attr_index = HDF_SD_ATTRFIND(sd_id, ’MaxT21’)
HDF_SD_ATTRINFO, sd_id, attr_index, DATA=MaxT21
index = HDF_SD_NAMETOINDEX(sd_id, ’FireMask’)
sds_id = HDF_SD_SELECT(sd_id, index)
HDF_SD_GETDATA, sds_id, FireMask
HDF_SD_ENDACCESS, sds_id
index = HDF_SD_NAMETOINDEX(sd_id, ’MaxFRP’)
sds_id = HDF_SD_SELECT(sd_id, index)
HDF_SD_GETDATA, sds_id, MaxFRP
HDF_SD_ENDACCESS, sds_id
HDF_SD_END, sd_id
help, FirePix
print, FirePix, format = ’(8I8)’
print, MaxT21, format = ’("MaxT21:",F6.1," K")’
help, FireMask, MaxFRP
The code produces the following output:
FIREPIX LONG = Array[8]
18 48 19 1 18 11 100 32
MaxT21: 468.1 K
FIREMASK BYTE = Array[1200, 1200, 8]
MAXFRP LONG = Array[1200, 1200, 8]
29
5.3 MOD14A2 and MYD14A2
The MOD14A2 (Terra) and MYD14A2 (Aqua) daily Level 3 8-day summary fire products are tile-
based, with each product file spanning one of the 460 MODIS tiles, 326 of which contain land
pixels. The product is a 1-km gridded composite of fire pixels detected in each grid cell over each
8-day compositing period.
5.3.1 Fire Mask
The fire mask is stored as a 1200 ×1200 8-bit unsigned integer SDS named “FireMask”. (For
historical reasons this layer was named “most confident detected fire” prior to Collection 5.) This
summary fire mask is essentially a maximum value composite of the Level 2 fire product pixel
classes (Table 2) for those swaths overlapping the MODIS tile during the eight-day compositing
period.
5.3.2 QA
The QA layer contains pixel-level quality assessment information stored in a 1200 ×1200 8-bit
unsigned integer image. The possible QA values are the same as those for the MOD14A1 and
MYD14A1 products (see Table 6).
5.3.3 Level 3 Tile Navigation
Forward and inverse mapping of the MODIS sinusoidal tile grid used for the MOD14A2 and
MYD14A2 products is the same as for the MOD14A1 and MYD14A1 products. See Section 5.2.6
for details.
5.3.4 Example Code
Example 4: MATLAB code to read the Level 3 MODIS 8-day fire mask, using the MATLAB routine
hdfread. This is probably the easiest way to read individual HDF SDSs in MATLAB.
mod14a2_file = ’MYD14A2.A2004193.h08v08.005.2007207151726.hdf’
% read entire "FireMask" SDS in one shot
fire_mask = hdfread(mod14a2_file, ’FireMask’);
% display fire mask (transposed so that North appears on top)
imagesc(fire_mask’);
30
Example 5: Longer version of MATLAB code to read the Level 3 MODIS 8-day fire mask. This is
probably the better approach to use if multiple subsets of an SDS will be read in sequence since the
HDF file will be opened and closed only once. (The shorter approach using hdfread requires that
the file be opened and closed for each read.)
mod14a2_file = ’MYD14A2.A2004193.h08v08.005.2007207151726.hdf’
sd_id = hdfsd(’start’, mod14a2_file, ’DFACC_RDONLY’);
sds_index = hdfsd(’nametoindex’, sd_id, ’FireMask’);
sds_id = hdfsd(’select’, sd_id, sds_index);
% prepare to read entire SDS (always 1200 x 1200 pixels in size)
start = [0,0];
edges = [1200,1200];
[fire_mask, status] = hdfsd(’readdata’, sds_id, start, [], edges);
status = hdfsd(’endaccess’, sds_id);
status = hdfsd(’end’, sd_id);
% display fire mask (transposed so that North appears on top)
imagesc(fire_mask’);
31
Example 6: C code for reading Level 3 MODIS 8-day fire mask using HDF library functions.
#include <stdio.h>
#include <stdlib.h>
#include "mfhdf.h"
#define ROWS 1200
#define COLS 1200
main(int argc, char **argv)
{
int32 sd_id, sds_index, sds_id;
int32 rank, data_type, nattr, dim_sizes[MAX_VAR_DIMS];
int32 start[2], int32 edges[2];
char *infile;
int i, j;
long nfire;
uint8 fire_mask[ROWS][COLS];
infile = "MOD14A2.A2008265.h31v10.005.2008275132911.hdf";
if ((sd_id = SDstart(infile, DFACC_READ)) == FAIL) exit(1);
start[0] = start[1] = 0;
edges[0] = ROWS;
edges[1] = COLS;
if ((sds_index = SDnametoindex(sd_id, "FireMask")) == FAIL) exit(2);
if ((sds_id = SDselect(sd_id, sds_index)) == FAIL) exit(3);
if (SDgetinfo(sds_id, (char *) NULL, &rank, dim_sizes, &data_type,
&nattr) == FAIL) exit(4);
/*check rank and data type */
if (rank != 2) exit(5);
if (data_type != DFNT_UINT8) exit(6);
if (SDreaddata(sds_id, start, NULL, edges,
(void *) fire_mask) == FAIL) exit(7);
if (SDendaccess(sds_id) == FAIL) exit(8);
if (SDend(sd_id) == FAIL) exit(9);
/*simple example: count grid cells containing fires */
nfire = 0;
for (i = 0; i < ROWS; i++) {
for (j = 0; j < COLS; j++)
if (fire_mask[i][j] >= 7) nfire++;
}
printf("%d grid cells containing fires.\n", nfire);
exit(0);
}
32
5.4 CMG Fire Products (MOD14CMQ, MYD14CMQ, etc.)
The CMG fire products are gridded statistical summaries of fire pixel information intended for use in
regional and global modeling, and other large scale studies. For Collection 6, the products are gener-
ated at 0.25◦spatial resolution for time periods of one calendar month (MOD14CMQ/MYD14CMQ)
and eight days (MOD14C8Q/MYD14C8Q).
At present the CMG products are distributed from the University of Maryland via anonymous
ftp (see Section 4.3). For convenience, these products are distributed in multiple, standard data
formats. Currently, HDF and Flexible Image Transport System (FITS) files are available.
5.4.1 CMG Naming Convention
Monthly CMG fire products. The file names of the monthly CMG product files have the structure
M?D14CM?.YYYYMM.CCC.VV.XXX, where M?D14CM? is a prefix3encoding the satellite
and product spatial resolution (see Figure 9), YYYY is the four-digit product year, MM is the
two-digit calendar month, CCC denotes the Collection (see Section 3.1.4), VV denotes the
product version within a Collection, and XXX is a suffix indicating the file format.
Eight-day CMG fire products. The file names of the 8-day CMG product files have the structure
M?D14C8?.YYYYDDD.CCC.VV.XXX, where M?D14C8? is a prefix encoding the satellite
and product spatial resolution (see Figure 9), YYYY is the four-digit product year, DDD is the
two-digit calendar month, CCC denotes the Collection (see Section 3.1.4), VV denotes the
product version within a Collection, and XXX is a suffix indicating the file format.
Satellite
‘O’ = Terra
‘Y’ = Aqua
‘C’ = combined Terra/Aqua
M?D14C??
Temporal Resolution
‘M’ = monthly
‘8’ = 8 days
Spatial Resolution
‘H’ = 0.5˚
‘Q’ = 0.25˚
Figure 9: MODIS CMG fire product naming prefix (ESDT) convention.
3In MODIS-speak this prefix is usually referred to as an Earth Science Data Type (ESDT).
33
5.4.2 Data Layers
The CMG fire products contain eight separate data layers summarized in Table 8.4For the 0.25◦prod-
ucts each layer is a 1440 ×720 numeric array.
Table 8: Summary of data layers in the CMG fire products.
Layer Name Data Type Units Description
CorrFirePix int16 - Corrected number of fire pixels.
CloudCorrFirePix int16 - Corrected number of fire pixels, with an addi-
tional correction for cloud cover.
MeanCloudFraction int8 - Mean cloud fraction.
RawFirePix int16 - Uncorrected count of fire pixels.
CloudPix int32 - Number of cloud pixels.
TotalPix int32 - Total number of pixels.
MeanFRP float32 MW Mean fire radiative power.
NumPixFRP int16 - Number of fire pixels used to compute mean FRP.
5.4.3 Global Metadata
Global metadata are stored as global attributes in the HDF product files, and primary-HDU key-
words in the FITS product files.
5.4.4 Climate Modeling Grid Navigation
Forward navigation. Given the latitude and longitude (in degrees) of a point on the Earth’s surface,
the image coordinates (x,y) of the 0.25◦CMG grid cell containing this point are computed as fol-
lows:
y = floor((90.0 - latitude) / 0.25)
x = floor((longitude + 180.0) / 0.25),
where floor is the floor function, e.g., floor(2.2) = 2. These equations yield image coordinates
satisfying the inequalities 0 ≤x≤1439, 0 ≤y≤719, which are appropriate for programming
languages using zero-based array indexing such as C and IDL; for languages using one-based array
indexing (e.g. Fortran, MATLAB) add 1.
Inverse navigation. Given coordinates (x,y) of a particular grid cell in the 0.25◦CMG fire products,
the latitude and longitude (in degrees) of the center of the grid cell may be computed as follows:
latitude = 89.875 - 0.25 ×y
4An additional fire persistence layer will likely be added in time for the summer 2015 public release.
34
longitude = -179.875 + 0.25 ×x
Here, xand yare again zero-based image coordinates; for one-based image coordinates first subtract
1 from both xand y.
5.4.5 Example Code
Example 7: IDL code for reading the cloud-corrected fire pixel layer within the Collection 6 MODIS
CMG monthly and 8-day fire products (HDF4 format).
; read "CloudCorrFirePix" array in CMG product (HDF4 format)
cmg_file = ’MYD14CMQ.200412.006.01.hdf’
sd_id = HDF_SD_START(cmg_file, /READ)
index = HDF_SD_NAMETOINDEX(sd_id, ’CloudCorrFirePix’)
sds_id = HDF_SD_SELECT(sd_id, index)
HDF_SD_GETDATA, sds_id, CloudCorrFirePix
HDF_SD_ENDACCESS, sds_id
HDF_SD_END, sd_id
Example 8: IDL code for reading the cloud-corrected fire pixel layer within the Collection 6 MODIS
CMG monthly and 8-day fire products (FITS format).
; read "CloudCorrFirePix" array in CMG product (FITS format)
cmg_file = ’MYD14CMQ.200412.006.01.fits’
FITS_OPEN, cmg_file, fcb
ihdu = FITS_FIND_HDU(fcb, ’CloudCorrFirePix’)
FITS_READ_ARRAY, fcb, ihdu, CloudCorrFirePix, ndims, dims
FITS_CLOSE, fcb
35
5.5 Global Monthly Fire Location Product (MCD14ML)
The monthly fire location product contains the geographic location, date, and some additional infor-
mation for each fire pixel detected by the Terra and Aqua MODIS sensors on a monthly basis. For
convenience, the product is distributed as a plain ASCII (text) file with fixed-width fields delimited
with spaces. The first line of each file is a header containing the abbreviated names of each column
(field). As an example, here are the first nine lines of the December 2008 product file:
YYYYMMDD HHMM sat lat lon T21 T31 sample FRP conf type
20081201 0051 T -12.029 143.019 321.8 289.6 681 15.1 0 1
20081201 0051 T -12.030 143.028 317.9 287.9 682 10.4 0 1
20081201 0051 T -12.039 143.027 356.4 289.1 682 75.6 0 0
20081201 0051 T -12.048 143.026 346.3 286.7 682 52.9 0 0
20081201 0051 T -12.055 141.969 320.9 291.4 571 15.9 0 0
20081201 0051 T -12.558 142.061 317.8 293.3 592 10.1 47 0
20081201 0051 T -12.981 143.487 330.4 301.2 752 20.2 83 0
20081201 0051 T -12.982 143.496 325.1 300.9 753 12.5 55 0
A brief description of each data column is provided in Table 9.
Table 9: Summary of columns in the MCD14ML fire location product.
Column Name Units Description
1 YYYYMMDD - UTC year (YYYY), month (MM), and day (DD).
2 HHMM - UTC hour (HH) and minute (MM).
3 sat - Satellite: Terra (T) or Aqua (A).
4 lat degrees Latitude at center of fire pixel.
5 lon degrees Longitude at center of fire pixel.
6 T21 K Band 21 brightness temperature of fire pixel.
7 T31 K Band 31 brightness temperature of fire pixel.
8 sample - Sample number (range 0-1353).
9 FRP MW Fire radiative power (FRP).
10 conf % Detection confidence (range 0-100).
11 type - Inferred hot spot type:
0 = presumed vegetation fire
1 = active volcano
2 = other static land source
3 = offshore
Hot-spot types 0–2 are reserved exclusively for land pixels; hot spots detected over water (presum-
ably offshore gas flares) will always be assigned a type of 3 (offshore).
5.5.1 Naming Convention
The names of the MCD14ML product files have the structure MCD14ML.YYYYMM.CCC.VV.asc,
where YYYY is the four-digit product year, MM is the two-digit calendar month, CCC denotes the
Collection (see Section 3.1.4), and VV denotes the product version within a Collection.
36
5.5.2 Example Code
Example 9: IDL code for reading a single monthly fire location product file while still compressed
(note the COMPRESS keyword when the file is opened).
infile = ’MCD14ML.200904.006.01.asc.gz’
header = ’’
year = 0
month = 0B
day = 0B
hour = 0B & minute = 0B
sat = ’’
lat = 0.0 & lon = 0.0
T21 = 0.0 & T31 = 0.0
sample = 0
FRP = 0.0
confidence = 0B
type = 0B
fmt = ’(I4.4,2I2,1X,2I2,1X,A1,F8.3,F9.3,2F6.1,I5,F8.1,I4,I2)’
openr, 2, infile, /COMPRESS
; skip header
readf, 2, header
while not EOF(2) do begin
readf, 2, year, month, day, hour, minute, sat, $
lat, lon, T21, T31, sample, FRP, confidence, type, $
FORMAT = fmt
; do something with values here
endwhile
close, 2
37
Example 10: R/S-Plus code for reading a single monthly fire location product file and plotting
separate histograms of band 21 brightness temperature for Terra and Aqua fire pixels.
z <- read.table("MCD14ML.200711.006.01.asc", header=T)
# two plots on page
par(mfrow=c(2,1))
xstr <- "Band 21 Brightness Temperature"
# Terra fire pixels
hist(z$T21[z$sat == "T"], xlab = xstr)
# Aqua fire pixels
hist(z$T21[z$sat == "A"], xlab = xstr)
38
6 Validation of the MODIS Active Fire Products
In this section we provide a brief overview of the validation status of the MODIS active fire prod-
ucts. A more detailed overview may be found in the active fire section of the MODIS Land Team
Validation web site5.
6.1 Validation Based on ASTER Imagery
Validation of the Terra MODIS active fire product has primarily been performed using coincident,
high resolution fire masks derived from Advanced Spaceborne Thermal Emission and Reflection
Radiometer (ASTER) imagery. See Morisette et al. (2005a,b), Csiszar et al. (2006), and Schroeder
et al. (2008) for details. A very brief (though now somewhat obsolete) discussion of the general
validation procedure, with some early results, can be found in Justice et al. (2002). For information
about the methodology for producing the ASTER fire masks, see Giglio et al. (2008).
More recent work by Schroeder et al. has achieved Stage 3 validation of the Level 2 Terra
MODIS fire product using 2500 ASTER scenes distributed globally and acquired from 2001 through
2006 (Figure 10). The results of this exhaustive effort will be published in a forthcoming paper.
Figure 10: Spatial coverage and distribution of 2,500 ASTER scenes (red patches) used in the
Stage 3 validation of MOD14. Image courtesy of Wilfrid Schroeder.
6.2 Other Validation
Independent validation of the Collection 5 Terra and Aqua MODIS active fire products without
ASTER has been performed by de Klerk (2008) and Hawbaker et al. (2008). These approaches
have at least two advantages over ASTER-based methods: 1) They can be applied to both MODIS
sensors (not just the Terra MODIS), and 2) they are not restricted to the near-nadir portion of the
MODIS swath.
5http://landval.gsfc.nasa.gov/ProductStatus.php?ProductID=MOD14
39
7 Caveats and Known Problems
7.1 Caveats
7.1.1 Fire Pixel Locations vs. Gridded Fire Products
We urge caution in using fire pixel locations in lieu of the 1-km gridded MODIS fire products. The
former includes no information about cloud cover or missing data and, depending on the sort of
analysis that is being performed, it is sometimes possible to derive misleading (or even incorrect)
results by not accounting for these other types of pixels. It is also possible to grossly misuse fire
pixel locations, even for regions and time periods in which cloud cover and missing observations
are negligible. Some caveats to keep in mind when using MODIS fire pixel locations:
•The fire pixel location files allow users to temporally and spatially bin fire counts arbitrarily.
However, severe temporal and spatial biases may arise in any MODIS fire time series analysis
employing time intervals shorter than about eight days.
•Known fires for which no entries occur in the fire-pixel location files are not necessarily
missed by the detection algorithm. Cloud obscuration, a lack of coverage, or a misclassifica-
tion in the land/sea mask may instead be responsible, but with only the information provided
in the fire location files this will be impossible to determine.
7.2 Collection 6 Known Problems
7.2.1 Pre-November 2000 Data Quality
Prior to November 2000, the Terra MODIS instrument suffered from several hardware problems
that adversely affected all of the MODIS fire products. In particular, some detectors were rendered
dead or otherwise unusable in an effort to reduce unexpected crosstalk between many of the 500 m
and 1 km bands. The dead detectors are known to introduce at least three specific artifacts in the
pre-November 2000 fire products: striping, undetected small fires, and undetected large fires. In
some very rare instances severe miscalibration of band-21 in the first weeks of the MODIS data
archive (February and March 2000) will cause entire scan lines to be identified as fire.
7.2.2 Detection Confidence
A detection confidence intended to help users gauge the quality of individual fire pixels is included
in the Level 2 fire product. This confidence estimate, which ranges between 0% and 100%, is
used to assign one of the three fire classes (low-confidence fire,nominal-confidence fire, or high-
confidence fire) to all fire pixels within the fire mask. In the Collection 4 fire product this confidence
estimate did not adequately identify highly questionable, low confidence fire pixels. Such pixels,
which by design should have a confidence close to 0%, were too often assigned much higher con-
fidence estimates of 50% or higher. While an adjustment implemented in the Collection 5 code
partially mitigated this problem, some highly questionable fire pixels are still classified as nominal-
confidence fires. An second minor adjustment was implemented for Collection 6 to help correct this
problem.
40
7.3 Collection 5 Known Problems
The frequency of certain problems known to affect the Collection 5 MODIS fire products was re-
duced for Collection 6. Two such problems are described here.
7.3.1 False Alarms in Small Forest Clearings
Extensive validation of the Collection-5 Level-2 Terra MODIS fire product by Schroeder et al. (2008)
found that small clearings within rainforest were a source of persistent false alarms in the Amazon.
An example is shown in Figure 11. For Collection 6, the frequency of this type false alarm was
reduced using an additional rejection test.
Figure 11: Example false alarm (red square with cross) in the Collection 5 product from 23
May 2002 (14:03 UTC) in an Amazonian rainforest clearing, with approximate edges of 1-km
MODIS pixels (black grid) superimposed on a high resolution ASTER image. Source: Schroeder
et al. (2008).
41
7.3.2 False Alarms During Calibration Maneuvers
A bug in the Level 1B calibrated-radiance production code occasionally produces spurious radiance
values in the thermal bands during lunar roll calibration maneuvers. This can produce spurious
stripes of fire pixels across the entire swath in up to ∼20 scans during these periods. The bug causes
similar striping in several other MODIS products, in particular the cloud mask.
While most of the affected Level 2 granules were deleted from the Collection 5 archive, a small
number were missed during quality assurance and subsequently propagated “arcs” of fire pixels into
the Collection 5 CMG and MCD14ML fire products. An example for the Aqua MODIS is shown
in Figure 12. The bug was fixed in late 2009, and the corrected Level 1B production code is now
being used for the Collection 6 reprocessing.
Figure 12: Example of a spurious arc of false fire pixels (red dots) in the Collection-5 8 December
2008 Aqua daily global browse imagery caused by spurious mid-infrared radiance values in the
Level 1B input data during a lunar calibration maneuver at 22:35 UTC. Cloud pixels are shown in
light blue, and areas lacking data are shown in white. Browse image courtesy of the LDOPE.
42
8 Frequently Asked Questions
8.1 Terra and Aqua Satellites
8.1.1 Where can I find general information about the Terra and Aqua satellites?
See NASA’s Terra and Aqua web sites for a start:
http://terra.nasa.gov/
http://aqua.nasa.gov/
8.1.2 When were the Terra and Aqua satellites launched?
18 December 1999 and 4 May 2002, respectively.
8.1.3 How can I determine overpass times of the Terra and Aqua satellites for a particular
location?
Both historical and predicted orbit tracks for Terra and Aqua are available from the University of
Wisconsin-Madison Space Science and Engineering Center (SSEC)6.
8.2 General MODIS Questions
8.2.1 Where can I find Algorithm Technical Basis Documents (ATBDs) for the MODIS land
products?
ATBDs for all of the MODIS land products are available from MODARCH7. Note that some are
not up to date and predate the launch of both the Terra and Aqua satellites.
8.2.2 Do the MODIS sensors have direct broadcast capability?
Yes, and there is a large community of MODIS direct broadcast data users. More information is
available from the NASA Direct Readout Laboratory8.
8.3 General Fire Product Questions
8.3.1 How are the fires and other thermal anomalies identified in the MODIS fire products
detected?
Fire detection is performed using a contextual algorithm (Giglio et al., 2003) that exploits the strong
emission of mid-infrared radiation from fires (Dozier, 1981; Matson and Dozier, 1981). The algo-
rithm examines each pixel of the MODIS swath, and ultimately assigns to each one of the following
classes: missing data,cloud,water,non-fire,fire, or unknown.
Pixels lacking valid data are immediately classified as missing data and excluded from further
consideration. Cloud and water pixels are identified using cloud and water masks, and are assigned
the classes cloud and water, respectively. Processing continues on the remaining clear land pixels.
6http://www.ssec.wisc.edu/datacenter/
7http://modarch.gsfc.nasa.gov/data/atbd/
8http://directreadout.sci.gsfc.nasa.gov/
43
A preliminary classification is used to eliminate obvious non-fire pixels. For those potential fire pix-
els that remain, an attempt is made to use the neighboring pixels to estimate the radiometric signal
of the potential fire pixel in the absence of fire. Valid neighboring pixels in a window centered on
the potential fire pixel are identified and are used to estimate a background value. If the background
characterization was successful, a series of contextual threshold tests are used to perform a relative
fire detection. These look for the characteristic signature of an active fire in which both 4 µm bright-
ness temperature and the 4 and 11 µm brightness temperature difference depart substantially from
that of the non-fire background. Relative thresholds are adjusted based on the natural variability of
the background. Additional specialized tests are used to eliminate false detections caused by sun
glint, desert boundaries, and errors in the water mask. Candidate fire pixels that are not rejected
in the course of applying these tests are assigned a class of fire. Pixels for which the background
characterization could not be performed, i.e. those having an insufficient number of valid pixels, are
assigned a class of unknown.
See Giglio et al. (2003) for a detailed description of the detection algorithm.
8.3.2 What is the smallest fire size that can be detected with MODIS? What about the
largest?
MODIS can routinely detect both flaming and smoldering fires ∼1000 m2in size. Under very good
observing conditions (e.g. near nadir, little or no smoke, relatively homogeneous land surface, etc.)
flaming fires one tenth this size can be detected. Under pristine (and extremely rare) observing
conditions even smaller flaming fires ∼50 m2can be detected.
Unlike most contextual fire detection algorithms designed for satellite sensors that were never
intended for fire monitoring (e.g. AVHRR, VIRS, ATSR), there is no upper limit to the largest
and/or hottest fire that can be detected with MODIS.
8.3.3 Why didn’t MODIS detect a particular fire?
This can happen for any number of reasons. The fire may have started and ended in between satellite
overpasses. The fire may be too small or too cool to be detected in the 1 km2MODIS footprint.
Cloud cover, heavy smoke, or tree canopy may completely obscure a fire. Occasionally the MODIS
instruments are inoperable for extended periods of time (e.g. the Terra MODIS in September 2000)
and can of course observe nothing during these times.
8.3.4 How well can MODIS detect understory burns?
The likelihood of detection beneath a tree canopy is unknown but probably very low. Understory
fires are usually small, which already makes MODIS less likely to detect them, but with the addition
of a tree canopy to obstruct the view of a fire, detection becomes very unlikely.
8.3.5 Can MODIS detect fires in unexposed coal seams?
In general, no. The detection algorithm is not tuned to look for the subtle temperature changes in
the overlying soil that is characteristic of such fires.
8.3.6 How do I obtain the MODIS fire products?
See Section 4.
44
8.3.7 What validation of the MODIS active fire products has been performed?
Validation of the Terra MODIS active fire product has primarily been performed using coincident,
high resolution fire masks derived from Advanced Spaceborne Thermal Emission and Reflection
Radiometer (ASTER) imagery. See Section 6.
8.3.8 I don’t want to bother with strange file formats and/or an unfamiliar ordering interface
and/or very large data files. Can’t you just give me the locations of fire pixels in plain
ASCII files and I’ll bin them myself?
You can use the MCD14ML monthly fire location product, or obtain MODIS fire pixel locations
via the Web Fire Mapper, but this doesn’t necessarily mean that fire pixel locations are the most
appropriate source of fire-related information. The fire pixel location files include no information
about cloud cover or missing data, and depending on the sort of analysis you are performing, it
is sometimes possible to derive misleading (or even incorrect) results by effectively ignoring these
other types of pixels. In many cases it is more appropriate to use one of the 1-km Level 3 or CMG
fire products. See Section 7.1.1 for more information about this issue.
8.3.9 I want to estimate burned area using active fire data. What effective area burned
should I assume for each fire pixel?
Pulling this off to an acceptable degree of accuracy is generally not possible due to nontrivial spa-
tial and temporal sampling issues. For some applications, however, acceptable accuracy can be
achieved, although the effective area burned per fire pixel is not simply a constant, but rather varies
with respect to several different vegetation- and fire-related variables. See Giglio et al. (2006b) for
more information.
8.3.10 Why are some of the MODIS fire products not available prior to November 2000?
Although the Terra MODIS first began acquiring data in February 2000, crosstalk and calibration
remained problematic until early November 2000 (see Section 7.2.1). Among other problems, this
compromises the integrity and consistency of the earliest MODIS fire products, in particular the
CMG fire products which are almost always used for time series analyses. For this reason we do not
distribute those products (specifically, the CMG and fire-location products) which were rendered
particularly inconsistent during the pre-November 2000 time period.
8.3.11 Why then are the Level 2 swath and Level 3 tiled fire products available before Novem-
ber 2000?
Because these products are not totally useless despite the early calibration problems. In addition,
these products are less often used for time series analysis, where a lack of consistency is likely to
be more problematic.
8.4 Level 2 Fire Products
8.4.1 Why do the Level 2 product files vary in size?
Level 2 granules can contain slightly different numbers of scans. More importantly, internal HDF
compression is used to reduce the size of the files.
45
8.4.2 How should the different fire detection confidence classes be used?
Three classes of fire pixels (low confidence, nominal confidence, high confidence) are provided in
the fire masks of the MODIS Level 2 and Level 3 fire products. Users requiring fewer false alarms
may wish to consider only nominal- and high-confidence fire pixels, and treat low-confidence fire
pixels as clear, non-fire, land pixels. Users requiring maximum fire detectability, who are able to
tolerate a higher incidence of false alarms, should consider all three classes of fire pixels.
8.4.3 How are the confidence values in the “FP confidence” SDS related to the confidence
classes assigned to fire pixels?
The confidence class assigned to a fire pixel (low,nominal, or high) is determined by thresholding
the confidence value (C) calculated for the fire pixel. These thresholds are listed in Table 10.
Table 10: Fire-pixel confidence classes associated with the confidence level Ccomputed for each
fire pixel.
Range Confidence Class
0% ≤C < 30% low
30% ≤C < 80% nominal
80% ≤C≤100% high
8.4.4 How can I take data from the fire-pixel-table SDSs (i.e., the one-dimensional SDSs with
the prefix “FP ”) and place the values in the proper locations of a two-dimensional
array that matches the swath-based “fire mask” and “algorithm QA” SDSs?
1. Open a MOD14/MYD14 Level 2 granule for reading using your favorite programming lan-
guage.
2. Determine the number of fire pixels in the granule. The easiest way to do this is to read
the global HDF attribute “FirePix”. (If you are a masochist you can read and parse the ECS
CoreMetadata.0 string for the product specific attribute FIREPIXELS instead.) If the num-
ber of fire pixels is zero, all of the “FP ” SDSs will have length zero, and there’s nothing left
to process, so close the file and go on to whatever else you’d like to do.
3. Find the number of lines in the granule. Call this number nlines. In the product this
quantity corresponds to the dimension number of scan lines. Since it is difficult to
determine the value of a named dimension directly with the HDF library, you must in-
stead determine the dimensions of an SDS for which the named dimension applies. You
can use either the “fire mask” or “algorithm QA” SDSs for this as they both have dimen-
sions number of scan lines by pixels per scan line. The HDF library function
SDgetinfo returns this information (in IDL use HDF SD GETINFO). You can determine the
number of samples as well (pixels per scan line), if you like, but the value of this
dimension will always be 1354.
4. Read the the “FP line” and “FP sample” SDSs in their entirety. These arrays contain pixel co-
ordinates within the granule for all of the quantities in the other “FP ” SDSs. Hereafter we’ll
46
assume these have been read and stored in internal arrays named FP line and FP sample,
respectively.
5. Create a 2-D array to hold whatever “FP ” quantity it is that you’d like to use. Assuming you
want the band 21/22 brightness temperature (“FP T21”), then in IDL you could do this:
T21 = fltarr(nlines, 1354)
6. Read the entire “FP ” SDS that you’d like to use. In the above example this is “FP T21”.
Following our earlier convention, we’ll assume this SDS is read into an internal array named
FP T21.
7. Populate pixels in the T21 array by indexing it with FP line and FP sample. In IDL you
would do this in one shot:
T21 = FP_T21[FP_line, FP_sample]
In non-vector-based languages you’d have to write an explicit loop. In C, for example, do
this:
for (i = 0; i < num_fire_pixels; i++)
T21[FP_line[i]][FP_sample[i]] = FP_T21[i];
Note that the coordinates in “FP line” and “FP sample” are zero-based. In a language like
Fortran (with the first array element numbered 1) you’d have to add 1 to all values in FP line
and FP sample.
8. Do whatever you want with the 2-D T21 array – it can now be indexed just like the fire mask
and QA SDSs would be if you had read them from the file. Note, though, that the newly
created T21 array will only contain data in those pixels where fires were detected. This is
true for 2-D arrays created from any of the other “FP ” SDSs as well.
9. Go back to step 4 for the remaining “FP ” quantities you want to use.
10. Close the HDF file.
8.4.5 Why are the values of the fire radiative power (FRP) in the Collection 4 Level 2 product
inconsistent with those in the Collection 5 Level 2 product?
The “FP power” SDS in the Collection 4 Level 2 product actually contained radiative power per
unit area, despite the fact that the units attribute of this SDS is assigned a value of “megawatts”
(this is an error). These values had to be multiplied by the appropriate pixel area (at the surface of
the Earth) to obtain the FRP, like this:
FRP (MW) = power values stored in the Collection 4 Level 2 product ×pixel area (km2)
Note that the area of a MODIS pixel varies with its position in the MODIS scan; see the next ques-
tion for details. Note also that starting with Collection 5 the Level 2 products have this multiplication
performed during processing and therefore contain the correct FRP.
47
8.4.6 What is the area of a MODIS pixel at the Earth’s surface?
The area of a MODIS pixel is nominally 1 km2but grows away from nadir. To find the approximate
pixel area, calculate the along-scan and along-track pixel dimensions (∆Sand ∆T, respectively).
The pixel area is then the product ∆S×∆T. General formulas for the pixel dimensions (in km)
can be found in Ichoku and Kaufman (2005) and are reproduced here:
∆S=Res
cos θ
q(Re/r)2−sin2θ
−1
(11)
∆T=rs cos θ−q(Re/r)2−sin2θ,(12)
where Re= 6378.137 km (Earth radius), r=Re+h,h= 705 km (satellite altitude), s= 0.0014184397,
and θis the scan angle. The scan angle (in radians) can be calculated from the granule sample SDS
(“FP sample”) included in the Level 2 fire product as follows:
θ=s×(sample −676.5) (13)
Note that the errors in the above approximations are smaller than the error entailed by treating the
pixel as having sharp edges.
8.4.7 Can I use cloud pixels identified in the Level 2 fire product as a general-purpose cloud
mask for other applications?
Cloud pixels are identified in the Level 2 fire products using simple, fixed brightness-temperature
and reflectance thresholds. While adequate for identifying optically thick cloud cover, this scheme
often fails to identify cloud edges and thin cirrus. It is also likely to misclassify snow and sometimes
desert as cloud. While adequate for the fire detection algorithm, which can tolerate these limitations
but cannot tolerate fires being mislabeled as cloud, these characteristics probably render the internal
cloud mask inadequate for most other applications.
48
8.5 Level 3 Tiled Fire Products
8.5.1 Why do coastlines in the tile-based Level 3 products looked so warped?
The tile-based Level 3 products are defined on a global sinusoidal grid which preserves areas but
greatly distorts the shape of land masses at longitudes far from the prime meridian.
8.5.2 Is there an existing tool I can use to reproject the tiled MODIS products into a different
projection?
The MODIS Reprojection Tool (MRT) can reproject the tiled MODIS products into many different
projections; see Section 10.
8.5.3 Why do some MOD14A1 and MYD14A1 product files have fewer than eight daily data
layers?
Days for which no MODIS data was acquired at all will not have a “plane” in the three-dimensional
Scientific Data Sets included in the MOD14A1 and MYD14A1 products.
8.5.4 How can I determine the date associated with each daily composite in the MOD14A1
and MYD14A1 products when fewer than eight days of data are present?
There are two ways to do this. Let the number of days of valid data (nominally 8) in the product be
Ndays.
Method 1: Using the product-specific Dates global attribute.
The Dates attribute (a string) will contain the date for each plane of the three-dimensional
SDS, with each of the Ndays dates encoded as 10-character substring (format YYYY-MM-DD),
separated by a space. Here’s an example for the normal case of Ndays = 8:
"2002-10-24 2002-10-25 2002-10-26 2002-10-27 2002-10-28 2002-10-29 2002-10-30 2002-10-31"
Method 2: Using the ECS ArchiveMetadata.0 global attribute.
You may alternatively search the ArchiveMetadata.0 global attribute for the DAYSOFYEAR
entry. This field will contain exactly Ndays dates, with each date corresponding to the date of each
plane in the three-dimensional Scientific Data Sets. For example, a time period having no days of
100% missing data will contain eight dates in the DAYSOFYEAR entry, like this:
" OBJECT = DAYSOFYEAR\n",
" NUM_VAL = 1\n",
" VALUE = \"2009-08-29, 2009-08-30, 2009-08-31,
2009-09-01, 2009-09-02, 2009-09-03,
2009-09-04, 2009-09-05\"\n",
" END_OBJECT = DAYSOFYEAR\n",
And here’s an example when only six days of data are present:
49
" OBJECT = DAYSOFYEAR\n",
" NUM_VAL = 1\n",
" VALUE = \"2001-06-10, 2001-06-11, 2001-06-12,
2001-06-13, 2001-06-14,
2001-06-15\"\n",
" END_OBJECT = DAYSOFYEAR\n",
8.5.5 How do I calculate the latitude and longitude of a grid cell in the Level 3 products?
You can use the online MODLAND Tile Calculator9, or perform the calculation as described in
Section 5.2.6.
8.5.6 How do I calculate the tile and grid cell coordinates of a specific geographic location
(latitude and longitude)?
You can use the online MODLAND Tile Calculator8, or perform the calculation as described in
Section 5.2.6.
8.5.7 What size are the grid cells of Level 3 MODIS sinusoidal grid?
The Level 3 MODIS products generated on the MODIS sinusoidal grid are colloquially referred to
as having “1 km”, ”500 m”, and “250 m” grid cells, but the actual cell sizes are shown in Table 11.
Table 11: Sizes of grid cells in Level 3 tiled MODIS sinusoidal grid.
Colloquial Size Actual Size (m)
“1 km” 926.62543305
“500 m” 463.31271653
“250 m” 231.65635826
9http://landweb.nascom.nasa.gov/cgi-bin/developer/tilemap.cgi
50
8.6 Level 3 CMG Fire Products
8.6.1 I need to reduce the resolution of the 0.25◦CMG fire product to grid cells that are a
multiple of 0.25◦in size. How do I go about doing this?
For all pixel-count data layers simply sum the values of the 0.25◦grid cells that lie within the larger
grid cells. Be sure to handle grid cells flagged with the missing data value of -1. At the very
least this entails excluding the negative missing data values from the resulting sum, but, depending
upon the application, it may be more appropriate to flag the coarser grid cell as lacking valid data
entirely. When coarsening the mean fire radiative power layer (MeanPower) you should weight
the individual 0.25◦mean FRP values by the corrected fire pixel counts (CorrFirePix), handling
(by at least excluding) missing FRP values of 0 in the process. A few examples are shown in the
following figures:
0.25˚ CorrFirePix
100 200
300 400
1000
0.5˚ CorrFirePix
Rebinning corrected fire pixel counts from 0.25◦grid cells (left) to a 0.5◦grid cell (right). The result
is simply the sum of the pixel counts of the four 0.25◦grid cells (100 + 200 + 300 + 400 = 1000
fire pixels) nested within the 0.5◦grid cell.
51
0.25˚ CorrFirePix
-1 -1
300 400
-1
0.5˚ CorrFirePix
Rebinning corrected fire pixel counts from 0.25◦grid cells (left) to a 0.5◦grid cell (right) when
missing data values of -1 are present. In this case we flag the entire 0.5◦grid cell as lacking valid
data which is appropriate when, for example, we are going to compare independent gridded fire
products that won’t generally have missing data values in exactly the same grid cells. For other
applications it would be sufficient to simply exclude the missing values from the sum, yielding a
result of 300 + 400 = 700 fire pixels in the 0.5◦grid cell.
0.25˚ MeanPower (MW)
10 20
30 40
30
0.5˚ MeanPower (MW)
Rebinning the mean FRP from 0.25◦grid cells (left) to a 0.5◦grid cell (right). The result is the
average of the FRP in the four 0.25◦grid cells nested within the 0.5◦grid cell, weighted by their
individual corrected fire pixel counts. Using the corrected fire pixel counts from the first example
(above) this yields:
10 MW×100 + 20 MW×200 + 30 MW×300 + 40 MW×400
100 + 200 + 300 + 400 = 30 MW
52
8.6.2 Why don’t you distribute a daily CMG fire product?
Because a MODIS product at daily temporal resolution will be plagued by extremely large sampling
bias errors. At most latitudes a single MODIS instrument simply does not sample the Earth’s surface
adequately in time periods shorter than about 8 days to “average out” most of the sampling bias.
8.6.3 Why don’t you distribute the CMG fire products as plain binary (or ASCII) files?
1) The number of files one must deal with balloons since most users request that individual data
layers be written to separate files; 2) it is difficult to include useful metadata without writing separate
header files, increasing the total number of files to handle even further; 3) it is possible for data and
its accompanying metadata to become separated; and 4) production, ingest, and analysis software is
much more likely to break when changes are made to the product.
8.6.4 Where can I find information about the FITS file format?
Extensive information about FITS is available from NASA’s FITS Support Office10. For a more
general overview see the FITS entry on Wikipedia11. You may also find the TRMM VIRS Monthly
Fire Product User’s Guide (Giglio and Kendall, 2003) helpful, particularly Appendix 2.
8.6.5 What software libraries are available for reading FITS files?
An extensive list, which includes packages for C, Fortran, IDL, Java, Perl, Python, and R, is main-
tained by the FITS Support Office12.
8.6.6 How can I display images in FITS files?
Two good choices are SAOimage13 (Figure 13) and DS914 (Figure 14). Use the “-ul” switch with
SAOimage and the “-orient y” switch with DS9 to orient North upwards, otherwise the grid
will appear upside down. If you’re willing to tolerate an upside-down orientation (i.e., South on
top), then Fv15 (Figure 15) is another good choice.
10http://fits.gsfc.nasa.gov/
11http://en.wikipedia.org/wiki/FITS
12http://fits.gsfc.nasa.gov/fits libraries.html
13http://tdc-www.harvard.edu/software/saoimage.html
14http://hea-www.harvard.edu/RD/ds9/
15http://heasarc.nasa.gov/docs/software/ftools/fv/
53
Figure 13: SAOimage displaying the CorrFirePix data layer in the February 2003 Collection 5
MOD14CMH product.
54
Figure 14: DS9 displaying the CorrFirePix data layer in the February 2003 Collection 5
MOD14CMH product. The longitude and latitude, respectively, at the center of the pixel beneath
the cursor is shown in the upper left hand corner of the window, in the two numeric fields to the
right of the word “LINEAR”.
55
Figure 15: Fv displaying the CorrFirePix data layer in the January 2009 Collection 5
MYD14CMH product. Note the “upside down” orientation of the global image.
56
8.6.7 Does the last 8-day CMG product for each calendar year include data from the first
few days of the following calendar year?
Yes. The last 8-day CMG product for each calendar year, which begins on day 361, includes the
first three days (two days for leap years) of the following calendar year. For example, the 8-day
Aqua MODIS CMG product MYD14C8Q.2004361.006.01.hdf is produced using observa-
tions from days 361–366 of 2004 and days 1–2 of 2005.
8.6.8 Where can I find details about the different corrections performed on some of the data
layers in the CMG fire products?
See Giglio et al. (2006a).
8.6.9 Are persistent hot spots filtered out of the MODIS CMG fire products?
Yes, static, persistent hot spots are excluded during production of the CMG fire products (Figure 16).
See Giglio et al. (2006a) for details. For Collection 6, the static source database will be updated
bi-annually since the Terra MODIS data record now spans more than 15 years in duration.
Figure 16: Locations of static hot spots excluded from the Collection 5 CMG fire products.
8.6.10 Is there an easy way to convert a calendar date into the ordinal dates (day-of-year)
used in the file names of the 8-day fire products?
Yes. Try the Unix (or Linux) cal command (with the -j switch), or use Table 12.
57
Table 12: Calendar dates (month/day) corresponding to the day-of-year (DOY) beginning each 8-
day time period for which the 8-day fire products are generated. Dates for non-leap years and leap
years are shown separately.
Non-Leap Leap Non-Leap Leap
DOY Date Date DOY Date Date
1 01/01 01/01 185 07/04 07/03
9 01/09 01/09 193 07/12 07/11
17 01/17 01/17 201 07/20 07/19
25 01/25 01/25 209 07/28 07/27
33 02/02 02/02 217 08/05 08/04
41 02/10 02/10 225 08/13 08/12
49 02/18 02/18 233 08/21 08/20
57 02/26 02/26 241 08/29 08/28
65 03/06 03/05 249 09/06 09/05
73 03/14 03/13 257 09/14 09/13
81 03/22 03/21 265 09/22 09/21
89 03/30 03/29 273 09/30 09/29
97 04/07 04/06 281 10/08 10/07
105 04/15 04/14 289 10/16 10/15
113 04/23 04/22 297 10/24 10/23
121 05/01 04/30 305 11/01 10/31
129 05/09 05/08 313 11/09 11/08
137 05/17 05/16 321 11/17 11/16
145 05/25 05/24 329 11/25 11/24
153 06/02 06/01 337 12/03 12/02
161 06/10 06/09 345 12/11 12/10
169 06/18 06/17 353 12/19 12/18
177 06/26 06/25 361 12/27 12/26
58
8.7 Global Monthly Fire Location Product
8.7.1 Can I use the MCD14ML fire location product to make my own gridded fire data set?
Yes, but please see the caveats in Section 7.1.1 first. If done carelessly you may end up with severe
temporal and spatial biases in your gridded data.
8.7.2 How many lines are in each MCD14ML product file?
This size of each product file depends on the number of fire pixels detected each month but typically
varies between 200,000 and 500,000 lines.
8.7.3 Are persistent hot spots filtered out of the fire location product?
No. Unlike the CMG fire product, static, persistent hot spots are not removed from the MCD14ML
product. For Collection 6, you can use the new type attribute to identify such fire pixels in the
MCD14ML product.
8.7.4 The MCD14ML ASCII product files have fixed-width, space-delimited fields. Is there
an easy way to convert these to comma-separated values (CSV) files?
Yes. In Unix, Linux, or Mac OS-X you can use the tr command to do this. Here’s an example:
tr -s ’ ’ , < MCD14ML.200805.005.01.asc > MCD14ML.200805.005.01.csv
8.7.5 How can I compute the scan angle given the sample number in the MCD14ML prod-
uct?
The scan angle θ(in radians) can be calculated from the value in the sample column as follows:
θ=s×(sample −676.5),(14)
where s= 0.0014184397. (See section 8.4.6 if you are also interested in the approximate size of the
pixel at the Earth’s surface.)
59
8.8 Hierarchical Data Format (HDF)
8.8.1 What are HDF files?
The Hierarchical Data Format (HDF), developed at the National Center for Supercomputing, is one
of various file formats used to portably archive and distribute scientific data. HDF files are more or
less “self-describing” in that they can include extensive metadata about the data stored within the
file. Note that there are two incompatible flavors of HDF in use: HDF4, the format in which all
MODIS products are stored, and HDF5, which is actually a completely different file format that is
not backwards-compatible with HDF4. See the NCSA HDF web site for more information16.
8.8.2 How do I read HDF4 files?
If you are writing your own software in a “traditional” programming language such as C or Fortran,
you will need obtain the HDF4 library from NCSA17. Some commercial software packages, how-
ever, including MATLAB, IDL, and ENVI, have the HDF library built-in, in which case you will
not need to install the library.
8.8.3 Can’t I just skip over the HDF header and read the data directly?
Put any thought of reading or writing HDF files without the HDF library out of your head. HDF
was intended to be not so much a physical file format, but instead an “application interface”. As
such, the format is fairly complicated (and has in fact changed over time) and it would be very time
consuming (and risky) to roll your own HDF ingest code. The physical file format is nothing like
the typical header-followed-by-data common to many other formats, and it is not easy to simply
skip over the metadata fragments in an HDF file.
8.8.4 How can I list the contents of HDF4 files?
The NCSA HDF4 Library includes a utility named ncdump which will do this. Be sure to use the
switch -h otherwise you will be inundated with ASCII dumps of all numeric arrays in the file.
8.8.5 How can I display images in HDF4 files?
Commercial software packages that can display the data layers in most MODIS products (which are
generally stored as HDF4 “Scientific Data Sets”) include ENVI18 and ERDAS Imagine19.
Freely available display software includes HDFView20 and the older HDFLook21 (pre-compiled
binaries only).
16http://hdf.ncsa.uiuc.edu/
17http://hdf.ncsa.uiuc.edu/hdf4.html
18http://www.ittvis.com/ProductServices/ENVI.aspx
19http://www.erdas.com/
20http://www.hdfgroup.org/hdf-java-html/hdfview/
21http://www-loa.univ-lille1.fr/Hdflook/hdflook gb.html
60
9 References
Csiszar, I., Morisette, J. T., and Giglio, L., 2006, Validation of active fire detection from moderate
resolution satellite sensors: the MODIS example in Northern Eurasia. IEEE Transactions on Geo-
sciences and Remote Sensing, 44, 1757-1764, doi:10.1109/TGRS.2006.875941.
Giglio, L., Csiszar, I., and Justice, C. O., 2006a, Global distribution and seasonality of active fires as
observed with the Terra and Aqua MODIS sensors. Journal of Geophysical Research, 111, G02016,
doi:10.1029/2005JG000142.
Giglio, L., Csiszar, I., Rest´
as, ´
A., Morisette, J. T., Schroeder, W., Morton, D., and Justice, C. O.,
2008, Active fire detection and characterization with the Advanced Spaceborne Thermal Emission
and Reflection Radiometer (ASTER). Remote Sensing of Environment, 112:3055-3063.
Giglio, L., Descloitres, J., Justice, C. O., and Kaufman, Y. J., 2003, An enhanced contextual fire
detection algorithm for MODIS. Remote Sensing of Environment, 87:273-282.
Giglio, L., and Kendall, J., 2003, TRMM VIRS Monthly Fire Product User’s Guide, Revision B.
(ftp://lake.nascom.nasa.gov/data/TRMM/VIRS Fire/docs/VIRS Fire Users Guide B.pdf).
Giglio, L., van der Werf, G. R., Randerson, J. T., Collatz, G. J., and Kasibhatla, P., 2006b, Global es-
timation of burned area using MODIS active fire observations. Atmospheric Chemistry and Physics,
6:957-974.
Ichoku, C., and Kaufman, Y. J., 2005, A method to derive smoke emission rates from MODIS fire
radiative energy measurements. IEEE Transactions on Geoscience and Remote Sensing, 43:2636-
2649.
Justice, C. O., Giglio, L., Korontzi, S., Owens, J., Morisette, J., Roy, D., Descloitres, J., Alleaume,
S., Petitcolin, F., and Kaufman, Y. J., 2002, The MODIS fire products. Remote Sensing of Environ-
ment, 83:244-262.
Kaufman, Y. J., Ichoku, C., Giglio, L., Korontzi, S., Chu, D. A., Hao, W. M., Li, R.-R., and Jus-
tice, C. O., 2003, Fires and smoke observed from the Earth Observing System MODIS instrument
– products, validation, and operational use. International Journal of Remote Sensing, 24:1765-1781.
Kaufman, Y. J., Justice, C. O., Flynn, L. P., Kendall, J. D., Prins, E. M., Giglio, L., Ward, D. E.,
Menzel, W. P., and Setzer, A. W., 1998, Potential global fire monitoring from EOS-MODIS. Journal
of Geophysical Research, 103:32215-32238.
Hawbaker, T. J., Radeloff, V. C., Syphard, D., Zhu, Z., and Stewart, S., 2008, Detection rates of the
MODIS active fire product in the United States. Remote Sensing of Environment, 112:2656-2664.
de Klerk, H., 2008, A pragmatic assessment of the usefulness of the MODIS (Terra and Aqua) 1-km
active fire (MOD14A2 and MYD14A2) products for mapping fires in the fynbos biome. Interna-
tional Journal of Wildland Fire, 17:166-178.
61
Morisette, J. T., Giglio, L., Csiszar, I., and Justice, C. O., 2005a, Validation of the MODIS Ac-
tive fire product over Southern Africa with ASTER data. International Journal of Remote Sensing,
26:4239-4264.
Morisette, J. T., Giglio, L., Csiszar, I., Setzer, A., Schroeder, W., Morton, D., and Justice, C. O.,
2005b, Validation of MODIS active fire detection products derived from two algorithms. Earth In-
teractions, 9(9):1-25.
Roy, D. P., Borak, J. S., Devadiga, S., Wolfe, R. E., Zheng, M., and Descloitres, J., 2002, The
MODIS Land product quality assessment approach. Remote Sensing of Environment, 83:62-76.
Schroeder, W., Prins, E., Giglio, L., Csiszar, I., Schmidt, C., Morisette, J. T., and Morton, D., 2008,
Validation of GOES and MODIS active fire detection products using ASTER and ETM+ data. Re-
mote Sensing of Environment, 112:2711-2726.
Wooster, M. J., Zhukov, B., and Oertel, D., 2003, Fire radiative energy for quantitative study of
biomass burning: derivation from the BIRD experimental satellite and comparison to MODIS fire
products, Remote Sensing of Environment, 86:83-107.
62
10 Relevant Web and FTP Sites
•MODIS Fire and Thermal Anomalies: General information about the MODIS Fire (Ther-
mal Anomalies) and Burned Area products.
http://modis-fire.umd.edu/
•MODIS File Specifications: Detailed file descriptions of all MODIS land products.
ftp://modular.nascom.nasa.gov/pub/LatestFilespecs/collection5
•MODIS Land Team Validation: Information concerning the validation status of all MODIS
land products.
http://landval.gsfc.nasa.gov/
•MODIS LDOPE Tools: A collection of programs, written by members of the Land Data Op-
erational Product Evaluation (LDOPE) group, to assist in the analysis and quality assessment
of MODIS Land (MODLAND) products.
https://lpdaac.usgs.gov/tools/ldope tools
•MODIS Reprojection Tool (MRT), Release 4.1: Software for reprojecting tiled MODIS
Level 3 products into many different projections.
https://lpdaac.usgs.gov/tools/modis reprojection tool
•MODLAND Tile Calculator: Online tool for performing forward and inverse mapping of
MODIS sinusoidal tiles.
http://landweb.nascom.nasa.gov/cgi-bin/developer/tilemap.cgi
•LANCE FIRMS Web Fire Mapper: The Fire Information for Resource Management Sys-
tem (FIRMS) Web Fire Mapper generates custom maps of active fires detected by the Terra
and Aqua MODIS instruments. Users can also active fire locations in ESRI shape file and
ARC/INFO formats.
http://firms.modaps.eosdis.nasa.gov/firemap/
•LANCE Rapid Response System: Access to near-real time Terra and Aqua MODIS re-
flectance, fire, vegetation index, and land surface temperature imagery. Includes a multi-year
archive.
http://earthdata.nasa.gov/data/near-real-time-data/rapid-response
•Reverb at the LP-DAAC: The primary distribution site for most of the MODIS land prod-
ucts. Formerly the Warehouse Inventory Search Tool (WIST), and before that the EOS Data
Gateway (EDG).
https://reverb.echo.nasa.gov/
•MODIS Land Product Quality Assessment: Product quality-assessment (QA) related in-
formation, including a very complete archive of known land-product issues with descriptions
and examples.
http://landweb.nascom.nasa.gov/cgi-bin/QA WWW/newPage.cgi
•NASA Direct Readout Laboratory: Free information and software to acquire, process, and
analyze MODIS Direct Broadcast data.
http://directreadout.sci.gsfc.nasa.gov/
63
•SSEC Terra and Aqua Orbit Tracks: Orbit tracks for various polar orbiting satellites, in-
cluding Terra and Aqua, from the University of Wisconsin-Madison Space Science and Engi-
neering Center (SSEC).
http://www.ssec.wisc.edu/datacenter/
64
