Harmonic Time Series Fitting

Fit a harmonic regression model to a per-pixel time series. Decomposes temporal signal into intercept, linear trend, and seasonal harmonics (amplitude + phase). Based on EEFA Book Ch. 18. Model: y(t) = a0 + a1*t + sum_k[a_k*cos(2*pi*k*t/T) + b_k*sin(2*pi*k*t/T)]

ID	harmonic_fit
Domain	geo
Category	analysis
Type	raster-collection-to-raster
Version	1.0.0

view yaml ◦view github ◦

Parameters

n_harmonics: integer · default 2 · Number of harmonic terms. 1 = annual, 2 = + semi-annual, 3 = + tri-annual, etc. More harmonics capture finer temporal detail but require more observations.
fit_trend: boolean · default true · Whether to include a linear trend term. Captures long-term monotonic change (e.g., greening/browning over years).
period: number · default 365.25 · Period of the fundamental harmonic, in the same units as the time axis. Default 365.25 for day-of-year inputs (annual cycle). Use 12 for monthly data, 1 for fractional-year inputs.