An algorithm for least-squares fitting of cubic spline surfaces to functions on a rectilinear mesh over a rectangle

In this paper a method is presented for fitting, in the least-squares sense, a bivariate cubic spline function to values of a dependent variable, specified at points on a rectangular grid in the plane of the independent variables. Products of B-splines are used to represent the bicubic splines. The coefficients in this representation are determined by solving a set of one-dimensional least-squares problems.