Smooth height/age curves from stem analysis with linear programming

Abstract

Stem analysis data defines a range of possible heights for each age. A smooth stem/age curve is obtained with linear programming (LP) when the sum of the absolute second differences of heights is minimized subject to constraints obtained from the stem analysis. The method is analogous to cubic splines. A LP problem can include additional constraints that are based on the assumption that the crosscut is randomly located within the annual height increment. The method produces smoother height curves than Issa method which is utilizing second order differences of ring counts. It was found using simulated data that the method provides better results than earlier methods for short bolts if height growth is sufficiently regular.