Computes the long and short term trends of a time series or optionally, the trends of the ratio of the time series and a covariate. The long term trend is estimated from the linear term of a regression on the full time series. The short term trend is computed as the windowed minimum over the time series.

The time series and covariate series are expected to contain a single band each, and the time series is expected to be evenly spaced in time. The output is 4 float bands: the long and short term trends, the t-test of the long term trend against the time series, and the Bruce Hansen test of parameter stability.

ImageCollection.formaTrend(covariates, windowSize)Image
this: timeSeriesImageCollection

Collection from which to extract trends.

covariatesImageCollection, default: null

Cofactors to use in the trend analysis.

windowSizeInteger, default: 6

Short term trend analysis window size, in images.