高频电磁波法是一种新的测量树轮密度的技术方法，具有高效、安全、不破坏样本等优点，但在具体应用时受众多的因素影响，如扫描参数、温度、样本表面光滑程度等。本文设计了重复测量实验、扫描参数设置实验、温度变化实验、打磨预处理实验等对比实验来研究测量值如何受这些因素影响，并以X射线法测量结果为标准，检验实验中高频电磁波法测量结果的准确性和稳定性。结果表明：（1）重复测量数据的稳定性存在样本差异、树轮指标的差异和测量间的差异。表现在样本A01各项树轮数据的稳定性好于样本A02；早晚材宽度数据的稳定性好于最小密度、最大密度数据；重复测量时，测量次数间有的没有显著差异，有的存在显著差异。（2）当扫描次数较少时（1次、3次），原始测量值有波动，自动识别年轮界线情况较差；增加扫描次数（5次、7次），原始测量值的波动变小，自动识别年轮界线能力也相应提高。合适的扫描次数可以确定为5次。（3）当扫描次数相同，不同扫描间距的测量结果相关性很高，可以高达0.98，扫描间距的不同对密度数据的影响不大。0.20mm的扫描间距自动识别年轮界线能力总体上要好于0.10mm的扫描间距，因此可以将扫描间距定为0.20mm。（4） 在不同温度下测量结果之间的相关性很高，但均值和方差呈现出差异。温度的不同并不影响测量出来的树轮密度变化趋势的相似程度，但可能影响树轮密度的均值和方差。（5）打磨预处理的实验表明，前6次打磨厚度逐渐减小的自动打磨可以快速的提高样本表面的光滑程度，树轮密度间的相关性较低但逐渐提高；后10次厚度不变的精细自动打磨很少再提高样本表面质量，此时树轮密度间的相关性较高且稳定；两次手动砂纸打磨能大大提高样本表面光滑程度，但树轮数据和自动打磨的结果相关很低。（6）与X射线法测量结果的对比，高频电磁波法能准确地反映出树轮密度的年内变化，能准确地测量出早晚材宽度数据。树轮密度数据和X射线法测量结果的相关呈现多种形式并且相关值很低，高频电磁波法测量树轮密度的准确性和稳定性很差，主要原因可能在于难以得到光滑、平整的高质量样本表面。

High-Frequency (HF) densitometry is a new technique for measuring tree-ring parameters which has advantages of fast measurement, low cost and no damage to samples. But the measurement is affected by many factors, such as scan settings, temperature, and the surface smoothness of the specimen. In this thesis, we compared HF measurements under different conditions to investigate how measurement values are affected by those factors. We also compared our measurement with X-ray data to evaluate the accuracy and stability of the HF results. The main conclusions are as follows:(1) The stability of duplicate measurements varies among different samples, different tree ring indexes and measuring times. That is, sample A01 is better than sample A02 in all types of tree ring data, earlywood width and latewood width is better than minimum wood density and maximum wood density, and among the duplicate measurements, some have significant differences, some have no significant differences.(2) When scanning a few times (one, three), fluctuations exist in original tree ring density curves, and the ability of automatic tree ring detection is poor. When increasing the scan times, fluctuation becomes smaller, and the tree ring border detection ability is improved. An appropriate scan time could be set as five.(3) When scan times are same, measurement results in different scan distances have very high correlations as high as 0.98. Tree ring density values are little affected by different scan distances. The ability of detecting tree ring border is generally better if scan distances is set as 0.20mm, which could be set as an appropriate scan distance.(4) Measurements under different temperatures have very high correlations, but the Mean and Variance values appear different. Temperature change has no effect on the similarity of tree-ring density change trend detected by HF densitometry, but may has effect on Mean and Variance values.(5) For the test of sanding pretreatment show, the first six times of automatic sanding with sanding thickness decreases could quickly improve the smoothness of specimen surface, and the correlations of tree-ring density curves are from low to higher gradually. The last ten times of fine automatic sanding with same sanding thickness rarely further improve the quality of the specimen surface, and the tree-ring density curves have high and stable correlations. Two times of manual sanding with sandpaper could improve the smoothness of specimen surface hugely, but the tree-ring curves between manual sanding and automatic sanding have low correlations.(6) Compared with X-ray method, the HF densitometry could also catch the intra-annual variability of tree-ring density and hence, get precise measurement of tree-ring width. Correlations of tree-ring density curves between HF and X-ray results are various low. The accuracy and stability of tree-ring density measured by HF densitometry are poor, mainly due to difficult to get a high quality of very smooth and flat specimen surface.