Author
For further information log on website :
http://www.nrcresearchpress.com/doi/abs/10.1139/x03-033#/doi/abs/10.1139/x03-055
Various sampling designs were evaluated using data on stand density, stocking, mortality, and top kill, as influenced by the western spruce budworm (Choristoneura occidentalis Freeman) in 17 stands in New Mexico and Colorado. Efficiency improved as plot size decreased from 0.04 to 0.02 ha for all variables and sampling designs, except for 0.01-ha plots, which required extremely large sample sizes and were subject to bias. Cluster designs were much more efficient than simple random sampling designs, allowing twice the reduction in sample size than was gained by relaxing the allowable error from 10 to 15%. Clusters of two plots were as precise as clusters of three plots. Of the four variables evaluated, density required the largest sample sizes, followed by stocking, percent mortality (for stands where mortality exceeded 10%), and top kill. Few plots were necessary to ascertain that mortality was less than 10%. On average, 10 pairs of 0.02-ha plots would estimate density, stocking, and mortality within a 10% allowable error. A field check of density and stocking variables is recommended, and additional samples are suggested in stands with large percent standard errors associated with those variables.
For further information log on website :
http://www.nrcresearchpress.com/doi/abs/10.1139/x03-033#/doi/abs/10.1139/x03-055
No comments:
Post a Comment