NDVI eMODIS
We are currently providing two different sources of NDVI: one from AQUA MODIS (eMODIS) and the other from Suomi NPP VIIRS (eVIIRS). There are two primary differences between them.
First, the averaging period being used to calculate anomalies (both absolute and percent of normal) are different. The eMODIS history utilizes a 15-year period from 2003 – 2017; the eVIIRS history uses a 10-year period from 2012 – 2021. So, in addition to the length of the historical period being different, the years being compared are also quite different.
Second, we changed the temporal smoothing algorithm for eVIIRS which treats the last period in the series differently than for eMODIS. The objective of the changes was to improve the quality of the end point smoothing for near real-time data.
Dekadal Period
Product Documentation
eMODIS AQUA Normalized Difference Vegetation Index (NDVI)
(Document updated 07/28/2017)
Background
The U.S. Geological Survey (USGS) Earth Resources Observation and Science (EROS) Center distributes satellite-derived vegetation products generated from the Collection 6 Moderate Resolution Imaging Spectroradiometer (MODIS) instrument flown aboard the Aqua satellite. These products, known as "eMODIS," respond to operational land monitoring applications requiring near-real time Normalized Difference Vegetation Index (NDVI) data for comparison against historical records. Real-time and historical NDVI products are composited in 10-day (dekadal) intervals on a Geographic-mapping grid.
Data
We use eMODIS 10-day maximum-value composite NDVI images at 250m spatial resolution to monitor vegetation conditions throughout major food insecure areas of the world. NDVI, a measure of the density of chlorophyll contained in vegetative cover, is defined as (NIR - RED) / (NIR + RED), where NIR is the near-infrared reflectance and RED is the visible-red reflectance. We generate these vegetation products from MODIS L1B Aqua surface reflectance, corrected for molecular scattering, ozone absorption, and aerosols using MODIS Science Team algorithms.
The NDVI and Anomaly maps are the product of a temporally smoothed 250m NDVI data set. We apply a time series smoother algorithm, developed by Swets et al. (1999), to smooth NDVI composites for the years 2002 to present. The technique uses a weighted least squares linear regression approach to "correct" observations that are of poor quality due to clouds or other atmospheric contamination (Figure 1). We derived a 10-year median NDVI on a pixel-by-pixel basis for each of 36 composite periods per year. As current-year composites become available, they are added to the time series and smoothed, resulting in a smoothed composite comparable to the historical median for a given 10-day period.
While temporal smoothing can be effective at improving time series data, it can be problematic to implement in real time for areas of persistent cloud cover. Therefore, we've implemented a process that steps back in the time series and replaces data after three composite periods, allowing cloud-free observations to have been obtained. Prior to the final corrected data being available, interim graphics for all products include masks using cloud flags from the original input data.
Figure 1. This time series NDVI plot, for a single 250m pixel, shows unsmoothed data in blue and temporally smoothed NDVI in red.
Dramatic reductions in unsmoothed NDVI data represent clouds and/or other atmospheric contamination. The smoothing algorithm effectively corrects these erroneous NDVI values based on characteristics
of the valid NDVI curve.
Image Products
- Temporally Smoothed NDVI - Smoothed 10-day NDVI composite.
- Median Anomaly - Anomalies represent a subtraction of the median NDVI values (2003-2017) for a 10-day period from current-year values for the same period, rendering an image where negative values portray less vigorous vegetation than average, and positive values represent areas that are more vigorous in the current year.
- Previous Year Difference - The difference image is a subtraction of the current year NDVI values from those of the previous year.
- Percent of Median - The percent of normal uses the 2003-2017 median to compare the current composite relative to average conditions and expresses that anomaly as a percent.
Spatial Parameters for eMODIS Data
eMODIS NDVI data are stretched (mapped) linearly (to byte values) as follows:
[-1.0, 1.0] -> [0, 200] - Invalid Values: 201 - 255
NDVI = (value - 100) / 100; example: [(150 - 100) / 100 = 0.5 NDVI]
Anomaly / Difference Classification:
The absolute difference and anomaly image is stretched from - 0.3 to 0.3 NDVI. The area of relatively no difference is approximately -0.05 - 0.05.
Percent of Median:
Expressed as percent, the percent of median data include values between 95 and 105 indicating average conditions. Values below 95 represent below average vegetation conditions, while those above 105 represent above average conditions.
All data are in GeoTIFF format with embedded color tables.
COORDINATE SYSTEM DESCRIPTION:
Geographic
Units: DD (decimal degrees)
Spheroid: WGS84
Pixel size:
x-direction: 0.002 (deg)
y-direction: 0.002 (deg)
References:
Swets, D.L., Reed, B.C., Rowland, J.D. & Marko, S.E. (1999). A weighted least-squares approach to temporal NDVI smoothing. In: 1999 ASPRS Annual Conference: From Image to Information, Portland, Oregon, May 17-21. Proceedings: Bethesda, Maryland, American Society for Photogrammetry and Remote Sensing.
USGS Advanced Research Computing, USGS Denali Supercomputer: U.S. Geological Survey, https://doi.org/10.5066/P9PSW367