Exponentially Weighted Moving Average Change Detection. This algorithm computes a harmonic model for the 'training' portion of the input data and subtracts that from the original results. The residuals are then subjected to Shewhart X-bar charts and an exponentially weighted moving average. Disturbed pixels are indicated when the charts signal a deviation from the given control limits.
The output is a 5 band image containing the bands:
ewma: a 1D array of the EWMA score for each input image. Negative values represent disturbance and positive values represent recovery.
harmonicCoefficients: A 1-D array of the computed harmonic coefficient pairs. The coefficients are ordered as [constant, sin0, cos0, sin1, cos1...]
rmse: the RMSE from the harmonic regression.
rSquared: r-squared value from the harmonic regression.
residuals: 1D array of residuals from the harmonic regression.
See: Brooks, E.B., Wynne, R.H., Thomas, V.A., Blinn, C.E. and Coulston, J.W., 2014. On-the-fly massively multitemporal change detection using statistical quality control charts and Landsat data. IEEE Transactions on Geoscience and Remote Sensing, 52(6), pp.3316-3332.
Collection from which to extract EWMA. This collection is expected to contain 1 image for each year and be sorted temporally.
Threshold for vegetation. Values below this are considered non-vegetation.
Start year of training period, inclusive.
End year of training period, exclusive.
|Integer, default: 2|
Number of harmonic function pairs (sine and cosine) used.
|Float, default: 1.5|
Threshold for initial training xBar limit.
|Integer, default: 20|
Threshold for running xBar limit.
|Float, default: 0.3|
The 'lambda' tuning parameter weighting new years vs the running average.
|Float, default: 3|
EWMA control bounds, in units of standard deviations.
|Boolean, default: true|
Should rounding be performed for EWMA
|Integer, default: 3|
Minimum number of observations needed to consider a change.