MODELS FOR ESTIMATING THE STAND VOLUME FR OM PHOTOGRAMMETRIC VARIABLES

The present paper deals with regression models to estimate the stand v olume of different tree species in the southwestern part of the Lower Saxony. The data were collected by 2-stage sampling procedures in scot s pine, larch, oak, beech and included 2 additional groups of hardwood s, representing low and high rotation species respectively. The more s pecific research question was whether regression models with independe nt variables being measured on aerial photographs and used as predicto rs reduce the cost of stand inventories. The following variables were measured and incorporated into the equations height; crown cover; crow n diameter; number of crowns. Transformation variables representing in teractions and higher powers of the primary variables were added to th e models. The unadjusted and adjusted R2 as well as MALLOWS'CP index w ere used to quantify the goodness of fit of the models; The study indi cated that subsets which contain 5 predictor variables produced satisf actory estimates, also in the upper and lower sections of the range of the predictor variables. The addition of more predictors did not subs tantially improve the models. Combining single species in species grou ps was justified only for those hardwoods, which could not be clearly distinguished on the aerial photograph, e.g. ash and maple. Dummy-vari ables, representing species were used for the hardwood prediction equa tions (Tab. 1), but species-dependent dummy-variables were nonsignific ant for Nor-way spruce, scots pine and larch. Comparing the accuracy o f the estimates from prediction equations with photogrammetric variabl es as predictors with those based on conventional terrestrial variable s indicated that a four to fivefold increase in the number of photo pl ots is necessary to match the precision obtained by conventional groun d surveying methods. The cost of measuring a single photo plot however is considerably less than the corresponding cost of ground surveying. This cost comparison however, is based on the assumption that recent aerial photos of a suitable scale are available.