利用植被近地面反射波谱季节特征对大针茅草原不同利用强度的植物群落的鉴别研究

测试并分析了大针茅草原不同退化程度（不退化，中（轻）度退化，重度退化）群落的近地面反射波谱的季节变化特征，方差分析与Ｄｕｃａｎ多重检验结果表明：由于不同退化程度群落的优势生活型植物及其数量特征的差异，它们的反射波谱特征彼此之间存在显著差异，在８月初和９月初，波谱特征的差异最为显著，而在５月底至６月底，差异不甚明显；利用两种统计方法－ＰＣＡ和模式识别技术对上述群落不同月份的波谱特征进行了鉴别，发现后者对不同退化群落的统计可分性高于ＰＣＡ，包含蓝光、绿光和近红外反射率的鉴别函数对不同退化程度的群落波谱特征识别精度最高，在８月初和９月初，鉴别效果最佳，鉴别错误概率在１％～７％之间，在５月底和６月底，对重度退化与非重度退化群落的错误概率也较小（约为４％），但对不退化与中（轻）度退化群落的错误概率可达１３％左右．

Distinguishing among Different Grazing Degradation Communities of Stipa grandis Steppe from Measurement of Near ground Multispectral Reflectance of Vegetation

The seasonal change of multispectral reflectance of vegetation under different grazing influence (non degradation,mederately (or lightly) degradation and heavily degradation) were measured on a representation grazing gradient of Stipa grandis steppe from the end of May to the end of September,1996,with a portablet radiometer,which was spectrally configure to match bands 1,2,3,4 of the Landsat 4 Thematic Mapper (TM).The result of analysis of variance and Duncan multiple comparision showed that there were remarkable difference of spectral reflectance,which were highest at the bengining of August and September but lowest at the end of May and June among the three cover types,because of the remarkable difference in dominant plant life types and their quantity.Two statistical procedures PCA and pattern recognition technique were used that would allow different degradation types to be distinguished by use of measurements of grassland multispectral reflectance,were found that the latter was more suitable than PCA for achieving this goal.The discrimination funation including the reflectance of blue,green and infrared bands had highest statistical separability in almost monthes,which were hightest at the bengining of August and September (the probability of error were between 1% and 7%) for all of three cover types and higher that distingshed heavily degradation community from non heavily degradation types (the probability of error were about 4%) at the end of May and June.