B. Elfving et A. Kiviste, CONSTRUCTION OF SITE INDEX EQUATIONS FOR PINUS-SYLVESTRIS L. USING PERMANENT PLOT DATA IN SWEDEN, Forest ecology and management, 98(2), 1997, pp. 125-134
Three general methods for construction of site index equations (the gu
ide curve method, the parameter prediction method and the difference e
quation method) and 13 growth functions were tested on data from perma
nent plots in 156 Scots pine (Pinus sylvestris) stands. Most stands we
re of cultivated origin with known year of establishment. Data consist
ed of measured top height and total age at the start and at the end of
observation periods of different lengths. Functions and construction
methods were compared with respect to residual variation (root mean sq
uare error (RMSE), meter) from original data. The guide curve method g
ave RMSE similar to 0.56 while the difference equation method with the
anamorphic formulation gave RMSE similar to 0.46, with appropriate fu
nctions. The difference equation method gave a higher RMSE with polymo
rphic than with anamorphic formulation. The special formulation of the
Hossfeld function for the difference equation method proposed by Cies
zeuski and Bella [Cieszewski, C.J., Bella, I.E., 1989. Polymorphic hei
ght and site index curves for lodgepole pine in Alberta. Can. J. Res.
19, 1151-1160] gave a simple and straightforward, polymorphic solution
with RMSE = 0.40. The parameter prediction method, in combination wit
h the deviation method proposed by Tveite [Tveite, B., 1969. A method
for construction of site-index curves. Meddelelser fra Det Norske skog
forsoksvesen 27, 134-159] and a four-parameter growth model, gave the
lowest RMSE, but this solution had patterns of overfitting. Parameter
prediction with the Mitscherlich function gave an equation with almost
the same accuracy as the polymorphic difference equation with the Hos
sfeld function. These functions deviate much at higher ages, but too f
ew observations in old stands in primary data prevented their separati
on. Published data from yield plots in old, natural stands supported t
he Hossfeld function. But, indications of a changed growth pattern in
later decades still leave open the question of height growth pattern i
n old pine stands of cultivated origin. The presented Hossfeld equatio
n can only be recommended for use in 10-80-year-old pine stands of cul
tivated origin. (C) 1997 Elsevier Science B.V.