Estimating the Area of Irregular-Shaped Objects by Dot Counts : Forest Inventories in Europe with Special Reference to Statistical Methods

1991

S. 223-231

6 Abb., 8 Lit. Ang.

Unselbständiges Werk

200086415

Counting dot grids is a well established method of area estimation. It is normally more efficient than randomized sample designs, but the error cannot be determined by statistical methods. For cases where the density function of point distances is known, an exact solution can be found with a formula given by Matern (1960, 1985). For the case of two classes ("0" and "1", "forest" and "nonforest") in this paper some results of a simulation study on irregular objects are presented showing the influence of the objects shape on the standard error percent of the area estimates. To characterize the form of the objects a measure (Formel) based on an "actual perimeter" is calculated directly from the sample data. In a further study the error of dot grid estimates with fixed grid direction is compared to that of estimates with random direction, showing that the former is greater: the difference, however, becomes smaller the more complicated the form and distribution of the objects are.