Fitting is the adjustment of the parameters of a model, curve or geometry so that it follows a set of observations as closely as an objective function allows.12

NIST describes fitting as finding the location, orientation and even size of a geometry such that an objective function is minimised, and says that the least-squares problem uses the sum of squares of the residuals.1 A Brown University handout defines the residuals as the differences between the observed values and those given by the fitted curve, and calls the process least squares fitting when the parameters minimise the root-mean-square error.2

SWEET lists Fit as equivalent to Fitting and as a subclass of its parameterisation class, which supports reading the term as parameter estimation within data analysis.3 The SWEET Turtle file carries labels and subclass links but no definition text.3

Retained sources do not establish which fitting methods Earth observation processing chains use, such as in retrievals or calibration, so this page bounds the term as a general analysis step.

Footnotes

  1. NIST, “On the Enduring Appeal of Least-squares Fitting in Computational Coordinate Metrology”. https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=907559 ↩ ↩2

  2. Brown University, “Curve fitting: least squares methods”. https://www.dam.brown.edu/people/alcyew/handouts/leastsq.pdf ↩ ↩2

  3. ESIP SWEET, “reprDataServiceAnalysis.ttl”. https://raw.githubusercontent.com/ESIPFed/sweet/master/src/reprDataServiceAnalysis.ttl ↩ ↩2