angle_count: Select trees proportional to size using...

In AWF-GAUG/fisim: Forest Inventory Simulator (fisim) is a package for simulating response and samplign designs common in forest inventories.

Description

Select trees proportional to size using Bitterlich/angle-count method

Usage

angle_count(tree_pop, sample_loc, baf = 1, k = NULL)

Arguments

tree_pop	A tree population object (tree_pop)
sample_loc	A sample location object (sample_loc) as provided by the xy_sample function.
baf	Basal area factor. Determines together with the trunk diameter at 1.3 m height up to which distance trees are included in the sample. Typical values for temperate forest conditions range between 1 and 4. The number of trees included at a sample location multiplied with the baf gives a direct estimate of basal area in square meter per hectare.
k	Helper variable needed in the nn2 function for the kd-tree search of candidate trees that are possibly included into the sample. Maximum the numer of rows in tree_loc. See details.

Details

The selection of trees according to the Bitterlich/ange-count method was implemented in several steps. First the critical distance up to which a tree is selected from a specific sample location is calculated using tree_dbh and baf. The larger tree_dbh and baf, the larger the critical distance of inclusion. In the second step the RANN package is used to identify candidate trees for inlcusion into the sample and their distances to the sample locations. The RANN package is a wrapper for the Approximate Near Neighbor (ANN) C++ library allowing for fast nearest-neighbor searches using kd-trees. Here, radius search is used, meaning that the k-nearest neighbors within a specific radius are searched for. As a radius the maximum critial distance is used. In combination with k, a larger than necessary sample of trees is selected at each sample location. If not specified by the user, k is estimated by multiplying the average tree density with the area of a circle with the maximum critical distance as radius. Tree density is inferred using the density.ppp function from the spatstat package. Note that the estimation costs some extra computation time and it is thus recommended to provide a large enough k when the function is used in simulations. In the third and last step, the set of candidate trees is querried for trees, where the actual distance to the sample location is smaller than the individual critical distance.

Value

An integer matrix where rows represent sample locations and columns represent indices of the trees in tree_loc that are selected at the individual sample locations. Zeroes are used to indicate no neighbours and to ensure a rectangular data format. See the output of the nn2 function.

AWF-GAUG/fisim documentation built on May 28, 2019, 11:02 a.m.