Current issue: 58(4)
This paper reports on tests made for the study of alternative methods in forest survey. Data were acquired by measurements in five areas in Finland and in Mexico, varying in size from 20 to 900 ha. The principal characteristics used in the analysis was the entire volume. By the combination of neighbouring plots, the variation could be studied for different plot sizes and survey strips. Variable (relascope) plots could be compared.
A starting point for the comparison of different sampling methods, calculations were made of the coefficients of variation for each plot type; total and within the strata. The amount of decrease of variation with an increasing plot size could be established. Comparisons have been made of the following sampling methods: simple random, stratified random, simple systematic, and stratified systematic sampling.
On comparisons of the standard error of sample mean it was found that in both stratified sampling and different types of systematic sampling there is, with increasing size and diminishing interval of sample plots, an increase in the relative improvement of the result against simple random sampling. Only in exceptional cases did systematic surveys give results which were less precise than those derived by other methods.
In discussion of some methods for determination of the precision of systematic sampling, possibilities of theoretical determination of the degree of precision was considered. An empirical study was made of the behaviour of some equations based on the sample itself. The larger the plot size and the shorter the plot interval, the more the equations overestimated in general the variance of sample mean.
As none of the equations studied gave reliable results, regression equations were calculated for the relative standard error on the basis of the data measured. The independent variables were plot size, plot or strip interval, area of survey unit and mean volume. The results arrived at are based mainly on the complete measurement of one area only. To enable extension of the scope of application, more material is needed with a complete enumeration of trees.
The PDF includes a summary in Finnish.
Highest degree of precision in determining the areas of different strata in forest survey is achieved when the areas are measured from a map. However, in practice the stratum-areas usually need to be determined on the basis of samples taken in the field or from aerial photographs. The goal of the present investigation was to determine the precision in stratum-area estimation on the application of different sampling methods.
Three sampling methods were used: 1. sampling with random plots, 2. uniform systematic plot sampling, and 3. sampling with equidistant lines.
The dependence of the standard error of stratum-areas in systematic line and plot sampling was examined by regression analysis. The models for regression equations were derived from random sampling formulae. It appears that the characteristics of these formulae were applicable as variables in the regression equations for systematic samples. Also, some characteristics of the distribution of the stratum was found, which seem to influence the error in sampling with equidistant lines.
The results as regards uniform systematic plot sampling indicate that the use of random sampling formulae leads to considerable over-estimation of the standard error. Nonetheless, unless relatively short intervals between sample plots are used in the forest survey made on the ground, it is of advantage to study the division of the area into strata by measuring the distribution of the survey lines in these strata.
The results can be used in two ways: for estimation of the precision in a survey already made, or to predetermine the sample size in a survey to be made. The results may be applicable to areas ranging from 100 to 1,000 ha in size, as well as to larger areas.
Growing demand for wood products, combined with efforts to conserve natural forests, have supported a steady increase in the global extent of planted forests. Here, a two-phase sampling strategy for large-scale assessment of the total area and the total wood volume of fast-growing forest tree crops within agricultural land is presented. The first phase is performed using tessellation stratified sampling on high-resolution remotely sensed imagery and is sufficient for estimating the total area of plantations by means of a Monte Carlo integration estimator. The second phase is performed using stratified sampling of the plantations selected in the first phase and is aimed at estimating total wood volume by means of an approximation of the first-phase Horvitz-Thompson estimator. Vegetation indices from Sentinel-2 are exploited as freely available auxiliary information in a linear regression estimator to improve the design-based precision of the estimator based on the sole sample data. Estimators of the totals and of the design-based variances of total estimators are presented. A simulation study is developed in order to check the design-based performance of the two alternative estimators under several artificial distributions supposed for poplar plantations (random, clustered, spatially trended). An application in Northern Italy is also reported. The regression estimator turns out to be invariably better than that based on the sole sample information. Possible integrations of the proposed sampling scheme with conventional national forest inventories adopting tessellation stratified sampling in the first phase are discussed.