| Learning about the status quo and variation trend of forest resources is the premise work offorest resources management. Because of the constraints from manpower,material,financialand other economic,time conditions, sampling methods are usually be used in forest resourcesinventory.Consequently, the designing of sampling plan is an important part of forest resourcesmanagement. Many traditional sampling methods are used in domestic forest resourcesinventory, including simple random sampling, systematic sampling, stratified sampling, clustersampling and so on. The theoretical basis of traditional sampling methods is classical statistics,which is research on the variation of the random variable, however, many forest inventoryfactors, such as forest category, tree species, land type and canopy density, are all related tospatial location, not pure random variables, but regional variables, and have both randomnessand structural attributes. This caused that traditional sampling methods have many defects inthe study of individual tree, community, spatial heterogeneity and spatial autocorrelation, in theproduction practice there are defects such as obvious space relevance and being inadaptable, etc.At the same time, as the forest district social economic conditions change, the change ofsampling frame and non-response sample unit in forest inventory are becoming a prominentphenomenon. Designing a sampling method with strict statistical foundation, high efficiency,low cost and strong adaptability has became an urgent task for forest exploration and designworkers.Spatial Balanced Sampling (SBS), firstly proposed by Stevens of Statistics Department inOregon University, in1997. As a innovative solution which is being used for solving spatialautocorrelation and uncertainty in natural resources survey area, SBS allows the spatial patternof the sample and spatial pattern of study population to have approximate types, andconsidering the influence of the non-response sample unit in the sampling program. GIS-BasedSpatial Balanced Sampling has a great advantage in establishing sampling frame, sampling planvisualization, facilitating the positioning and seeking of sample points.This article introduced the traditional sampling theory and methods, on currently existingfoundation, by the way of definite geo-statistics basis of SBS, building sampling performanceevaluation index,selecting study object in typical area, simulation experiments were carried outwith traditional sampling methods and spatial balanced sampling methods based on GIS. Theresults show that:(1) The geo-statistics analysis in forest resources survey showed that many forestinventory factors have generally spatial auto-correction, and they are not pure random variables, but regional variables. The existing sampling methods of forest resources inventory are basedon classical statistics which study the variation of pure variables. Exploring new samplingmethods based on geo-statistics is becoming an urgent task for forestry workers.(2) Combining of computer, remote sensing data, GIS platform and sampling simulationsoftware makes the traditional ground sampling change into the computer screen samplingmethod, which can fastly and accurately get a variety of information, overcome the manpowerconsumption, material consumption and high time cost defects in traditional sampling methods.(3) Different sampling simulations show that: Spatial Balanced Sampling showed greateradvantage than traditional sampling methods in quantitative analysis of sampling performanceevaluations.(4) The study of Spatial Balanced Sampling methods has an important theoretical andpractical significance in remedying the strong spatial auto-correction and poor adaptabilitydefects, reducing the sampling frame changes and non-response sample unit and solving thecontradiction of limited investigation costs and rapid increasing costs in forest inventory. SBSprovides a new opportunity for solving the forest survey sampling problems under the newsituation. |
PDF Full Text Request