Deriving Metrics from NDVI
Methods for Deriving Metrics
Challenges in Deriving Metrics
Data Smoothing - Reducing the "Noise" in NDVI
The reflected light waves that satellite sensors detect coming from vegetation on the Earth's surface can be altered or blocked by a variety of phenomena, including aerosols and clouds in the atmosphere as well as changing illumination patterns and the angle at which the satellite views the ground at any given time. These phenomena introduce "noise" into raw satellite data. To address this problem, raw data are processed using techniques that filter out noise and produce a clearer, more representative data set.
Filtering techniques vary. Compositing—merging maximum NDVI values acquired over (typically) 7-, 8-, 10-, 14-, or 16-day intervals—increases data quality. But residual effects of sub-pixel clouds, prolonged cloudiness, and other negative elements require further processing in the form of data smoothing. Data smoothing facilitates time-series analyses by reducing aberrant, noise-induced peaks and valleys that appear when NDVI values are plotted graphically to reveal vegetation changes over time.
At USGS/EROS, we smooth raw satellite data temporally, using a weighted, least-squares linear regression approach that involves a moving temporal window to calculate a regression line. The window is moved one period at a time, resulting in a family of regression lines associated with each data point. This family of lines is then averaged at each point, and interpolated between points, to provide a continuous, relatively smooth NDVI signal over time. Furthermore, since the phenomena that introduce noise into raw satellite data usually reduce NDVI values, we apply a weighting factor during the smoothing process that favors peak points over sloping or valley points. A final operation assures that all peak NDVI values in the moving window are retained. The resulting relationship between raw and smoothed data is statistically based.