Abstract: [Objective] In order to accurately obtain the spatial distribution information of forest trees in forestry surveys, it is important to select the most appropriate spatial pattern analysis method. Therefore, it is necessary to compare the existing various spatial pattern methods and master their adaptability to different application scenarios.[Method] In this paper, the simulated sample plot and the actual survey plot are used as data sources. MATLAB and R language tools are used. The similarities and differences between the five methods of nearest neighbor method, uniform angle index, Voronoi coefficient of variation, Ripley's L-function and pair-correlation function in working principle, usage and evaluation criteria, and the adaptability of each methods to different conditions are compared and analyzed. In order to facilitate comparison, the pattern analysis methods used in this paper are divided into fixed-scale method and variable-scale method according to whether they depend on spatial scale or not.[Result] Broad-leaved forests in the study area have significant spatial distribution patterns,with aggregate distribution and random distribution as common spatial patterns. The fixed-scale methods, which rely on spatial neighbor structure, are suitable for small scale, and are widely used in stand management and micro-structural adjustment. The variable-scale methods are closely related to the spatial scale, and could provide more abundant spatial information, which are suitable for long-term forest monitoring under complex conditions. Uniform angle index and coefficient of variation are consistent in evaluating spatial pattern, which can be verified or replaced by each other in practical application, while the nearest neighbor method may fail to distinguish spatial patterns in some cases. In most cases, the probability density function(pair-correlation function)are easier to explain and analyze than the cumulative distribution function(Ripley's L-function). This paper also found that pair-correlation function was better than Ripley's L-function. The advantages and disadvantages of uniform angle index, pair-correlation function and Ripley's L-function are still controversial. The author thinks that each spatial pattern analysis methods has its applicable premise and conditions, and its advantages and disadvantages are only relatively speaking. Two types of pattern analysis methods are affected by factors, such as sample size, plot size, stand density, etc. The more samples there are, the more accurate the evaluation result will be, and the increasing workload of pattern investigation, analysis and calculation will be followed.[Conclusion] Various spatial pattern analysis methods have their own characteristics and applicable preconditions. In practical applications, appropriate methods should be selected in conjunction with specific sample conditions. In the future, the study of the spatial distribution pattern in forest should not be limited to a simple description of aggregate distribution, uniform distribution or random distribution of the overall data. It should also be applied to optimize and regulate forest structures to maximize the multiple functions of the forest, in order to provide a reference and scientific basis for the accurate analysis of forestry data and the management practices of forestry production.

Key words: forest spatial distribution pattern, fixed-scale method, variable-scale method, adaptability