A Kalman filter approach to localizing height-age equations

• Published in: Forest science Vol. 37; no. 6
• Main Authors: Walters, D.K. (University of Minnesota, St. Paul, MN), Burkhart, H.E, Reynolds, M.R. Jr, Gregoire, T.G
• Format: Journal Article
• Language: English
• Published: 01.12.1991
• Subjects: AGE; ALTURA; ARBOLES; ARBOLES FORESTALES; ARBRE; ARBRE FORESTIER; CARACTERISTICAS DEL SITIO; EDAD; FACTEUR LIE AU SITE; HAUTEUR; MATEMATICAS; MATHEMATIQUE; MODELE; MODELOS; PINUS TAEDA
• ISSN: 0015-749X; 1938-3738
• DOI: 10.1093/forestscience/37.6.1526

Equations for modeling the height-age pattern of forest trees or stands are typically developed on a regional basis. Due to variation in growth patterns within regions, these equations may be polymorphic and quite complex. As an alternative to increasing the complexity of an equation, this study presents procedures for the evaluation of height-age relationships through the use of simple, regional equations combined with local stand data. The base model in this study was Schumacher's (1939) logarithm of height-reciprocal of age model. An average regional guide curve was developed using data from a wide range of loblolly pine plantations. A Kalman filter procedure was used to localize this regional equation to a specific stand. This procedure employs the general concept of feedback (local stand data combined with a simple regional model) in localizing the simple equation. By using information on a particular stand's height growth development, the original equation is modified to better model that stand. The modification is made by feeding local information into an estimation procedure that combines it with the regional information (equation) and modified parameter estimates are calculated. Three loblolly pine data sets were used to compare the localization strategy with the simple, regional model, as well as a more complex polymorphic equation. The results indicate that the Kalman filter method can result in improved prediction for specific stands