This page was adapted from a document prepared by David Benz when first processing MODIS Data for the original EDEN FP6 Project. The original with figures can be downloaded here.

This is a set of images produced by temporal Fourier analysis of global MODIS data. The imagery summarises some key environmental indicators, incorporating seasonal dynamics, for the EDENext study area.

The EDENext Partners can download the latest version of the Fouriered MODIS v5 Dataset including the smooth-fill correction from the EDENext Data Archive.  This product is currently available to EDENext members only.

MODIS is a sensor on board two NASA satellites, providing near-daily coverage of the entire Earth. Three MODIS version 4 data products were used in the production of these EDEN images. The MOD11A2 product contains an 8-day average of land surface temperature at 1-kilometre resolution. MCD43B4 imagery comprises 16-day composites of reflectance for seven wavelength bands, also at 1-km resolution. Reflectance values have been adjusted to remove the distortion caused by view angle and land surface texture. Another data set, provided by researchers at the University of Montana, contains 7 years%u2019 worth of global evapotranspiration estimates at 8-day intervals and .05-degree resolution. These were developed by combining information from weather stations and MODIS vegetation products. A paper by Mu et al., 2007 describes the algorithm used.

Temperature and reflectance layers were extracted from MOD11A2 and MCD43B4 files for the years 2001 through 2008. Reflectance bands were combined to create two vegetation indices, the Normalized Difference Vegetation Index (NDVI) and the Enhanced Vegetation Index (EVI). A set of nine measures, or channels, was then processed by a temporal Fourier algorithm: middle infra-red reflectance, daytime land surface temperature, nighttime land surface temperature, NDVI, EVI, evapotranspiration, latent heat flux/transfer, potential evapotranspiration, and potential latent heat flux/transfer.

A stepwise system of thresholds and interpolations screened erroneous values and bridged gaps in the time series. The smoothed series was sampled at 5-day intervals and transformed into a set of sine curves describing annual, bi-annual, and tri-annual fluctuations. For each of these curves, the Fourier algorithm generated images expressing the amplitude, phase, and variance. Other output recorded the mean, minimum, and maximum of the time series, and error measured during the Fourier transform. For a detailed descripton of the Fourier algorithm and its output, please see the article by Scharlemann et al., 2008.

For temperature and reflectance channels, sea pixels were masked with a MODIS land/sea layer and the images were projected from sinusoidal to geographic. In all channels, the EDEN study region was subset from global images. Idrisi rasters were converted to geotiff format in order to give data users more flexibility.


File names


The eg at the start of each file name indicates that the image covers the EDEN region and is in geographic projection. The next pair of characters stands for the time period processed, either 15 for 2001-2005 or 18 for 2001-2008.

The next two characters identify the channel:
03 - middle infra-red
07 - daytime land surface temperature
08 - nighttime land surface temperature
14 - NDVI
15 - EVI
35 - evapotranspiration
36 - latent heat flux/transfer
37 - potential evapotranspiration
38 - potential latent heat flux/transfer

The last two characters of each file name denote the output from Fourier processing:
a0 - mean
mn - minimum
mx - maximum
a1 - amplitude of annual cycle
a2 - amplitude of bi-annual cycle
a3 - amplitude of tri-annual cycle
p1 - phase of annual cycle
p2 - phase of bi-annual cycle
p3 - phase of tri-annual cycle
d1 - variance in annual cycle
d2 - variance in bi-annual cycle
d3 - variance in tri-annual cycle
da - combined variance in annual, bi-annual, and tri-annual cycles
vr - variance in raw data
e1 - error of input files
e2 - error before Fourier analysis
e3 - error after Fourier analysis

egv490lw is the MODIS version 4 land/water layer, which contains the following values:
1 - land
2 - coast
3 - inland water
4 - ephemeral water
eg_mask is a binary image where sea pixels are 0 and pixels included in the Fourier output are 1.



Channels 03, 07, 08, 14, 15, land/water, and mask:
File format: Idrisi binary raster (.rst) and geotiff
Byte size: unsigned 8-bit, signed 16-bit
Unzipped file size: 78.3 MB (u8), 156 MB (s16)
Image dimensions: 10018 columns x 8200 rows
Resolution: .008333 x .008333 degrees
Projection: geographic (latitude/longitude)
Extent: 6.1666 N to 74.5 N, 20 W to 63.48333 E
Background value: 0

Channels 35, 36, 37, 38:
File format: Idrisi binary raster (.rst) and geotiff
Byte size: signed 16-bit
Unzipped file size: 4.35 MB
Image dimensions: 1670 columns x 1367 rows
Resolution: .05 x .05 degrees
Projection: geographic (latitude/longitude)
Extent: 6.15 N to 74.5 N, 20 W to 63.5 E
Background value: 0


MODIS home page: http://modis.gsfc.nasa.gov/

Mu, Q., F. A. Heinsch, M. Zhao, and S. W. Running, 2007. Development of a global evapotranspiration algorithm based on MODIS and global meteorology data. Remote Sensing of Environment 111(4): 519-536.

Scharlemann, J. P. W., D. Benz, S. I. Hay, B. V. Purse, A. J. Tatem, G. R. W. Wint, and David J. Rogers, 2008. Global data for ecology and epidemiology: a novel algorithm for temporal fourier processing MODIS data. PLoS ONE 3(1): e1408.