FORESTAR：A decision—support system for multi—objective forest management in Northeast China

Past monoculture forestry in China has contributed to countrywide ecological disasters and economic difficulties in forestry regions. China‘s new forestry programs, Natural Forest Conservation Program and Returning Farmlands to Forests Program, provide opportunities for ecosystem management of mountain forests in China. A decision support system, FORE-STAR, has been developed for better managing and protecting natural forests in Changbai Mountain area. It uses GIS-based forest inventory data at a scale of forestry bureau. The first version contains two sub-modules: forest operation and forest res-toration. Under each sub-module, users can compare several decision options and make optimal choices. It can help field for-esters, forest managers, and policy makers make multi-objective and consistent decisions in planning forest management at hierarchical administrative scales.     

从化温泉松材线虫病综合防治示范区建设及其成效

2001～2005年,在严密监测疫情的基础上,对从化温泉松材线虫病综合防治示范区采取检疫封锁、清除病死树、应用引诱剂防治和树干打孔注药等综合防治措施,取得了良好的防治效果,发病松林面积由2001年开始发病时的573.3 hm2减少到2005年底的346.7 hm2,病死树数量由2001年的3 726株降低到2005年的239株,林间松墨天牛种群数量也大幅度下降.初步形成了一套较为完整的松材线虫病防治技术方法.     

弱度抚育对青海云杉胸径与材积生长变化的数学分析

生长量是评估立地条件及经营措施效果好坏的重要指标。正确研究分析林木生长规律以及采取相应的经营管理措施，对改善林分状况，森林环境和提高林木生长量具有重要的意义，研究表明，强度为１０％的抚育措施，可使西营河试验区树木胸径定期平均生长量提高８４．８５％，相当于２２％，的强度抚育措施使哈溪试验区树木胸径定期平均生长量提高６９．２０％，材积定期平均生长量提高６５．３３％，经Ｆ０．０５检验，两试验区胸径生长与     

Simulating coconut growth, development and yield with the InfoCrop-coconut model

Simulation modeling of perennial crops has immense potential for generating information for plantation managers. We report the development of the InfoCrop-coconut model and its application to coconut (Cocos nucifera L.) growing in diverse tropical and subtropical environments. The model is based on the generic crop model InfoCrop that simulates various annual crops in tropical and subtropical regions. The InfoCrop-coconut model was calibrated and validated with data compiled from published studies comprising many physiological, agronomical and nutritional experiments conducted between 1978 and 2005 in diverse geographic locations throughout India. The treatments included various water and nutrient regimes and varieties of coconut. Time to first flowering varied between 4 and 6 years, leaf production varied from 8 to 15 leaves year(-1) and nut yield ranged from 3000 to 27,000 nuts ha(-1) year(-1). The genetic coefficients used for calibration and validation were generated from field experiments conducted during 1995-2005. Model efficiency and validation performance were analyzed statistically. Simulated trends in phenological development, total dry mass and its partitioning, and nut yield agreed closely with observed values, although a 15% error was observed in a few cases. Considering that field measurements have an experimental error of 10-15% and wide variation existed within treatments, the model adequately simulated the effects of management practices and agro-climatic conditions over short periods. For a range of agro-climatic zones, simulated potential yields varied from 26 to 30 Mg ha(-1) year(-1) and potential annual dry mass production varied from 52 to 62 Mg ha(-1), depending on environment. We conclude that InfoCrop-coconut can be used to increase the efficiency of agronomic experiments designed to aid coconut crop management.     

沙化土地信息提取研究

采用LandsatETM 多时相影像,在地类光谱特征分析的基础上,用分层分离的方法,提取出沙化土地信息。以全国沙漠化普查地类划分标准中的植被覆盖度为确定沙化土地程度的依据,通过建立NDVI和实测植被覆盖度的回归模型,确定不同程度沙化土地对应的植被覆盖度与NDVI的对应关系,用NDVI阈值进行划分,得到沙化土地程度图。     

干涉测量土地利用影像分类决策树法森林识别研究

利用ERS－1和ERS－2 SAR串行轨道数据经干涉测量处理生成的干涉测量土地利用影像对森林识别方法进行初步研究。内容包括基于目标识别选择合成干涉测量土地利用影像处理方法、应用斜分类器（OCI）生成分类决策树以及在自主开发的软件中调整分类决策树、对分类结果进行像元级别上的精度检验和误差分析等。结果表明：ERS－1和ERS－2 SAR串行轨道数据经干涉测量处理，利用其强度影像和相干影像可以合成多种干涉测量土地利用影像；其中最小值影像和 标准差影像较之于其它强度影像和变化影像更有利于区分水体和森林；由于OCI生成决策树的算法决定了在分类处理中没有不可分类别的存在，所以在初期选择分类类别时，要尽可能多的覆盖原始影像的数值区间；选择的各类别样本数据要尽可能的“纯”，以减少类别间数值区间的重叠，从而减少误分类情况。     

基于神经网络和颜色特征对木材进行分级的分析

为了解决木材的分级问题，提出了利用神经网络和木材表面颜色特征对木材进行分级的方法。利用竞争神经网络对114种不同树种进行粗分，然后再用BP神经网络算法对此进行验证，又对四种木材进行分类验证。从计算机分类分级仿真效果上看，达到了预期的目的。     

沼泽地改造实验的调查与研究

调查研究结果表明，实验所采取的各项措施是可行有效的，沼泽地生境发生了明显变化，地下水平均下降０．６－１．０ｍ，地温５－２０ｃｍ处提高６－８℃，土壤的理化性质得到明显改善，水、气、热趋于协调。林木生长良好，６年生落叶松树高、胸径生长量分别为０．４６ｍ和０．５０ｃｍ，已达到速生丰产林标准。     

基于分形维木材表面粗糙度的研究

应用分形维对木材表面粗糙度特征进行研究。首先，对Sarkar和Chaudhuri提出的分形维计算方法做了改进：其次，以60个树种为研究对象，提取和分析木材表面的分形维数。得出结论：分形维数能够反映木材表面的粗糙程度，可作为木材表面粗糙度的一种度量？     

图像分割技术在木材表面缺陷识别中的应用

阐述了图像分割技术在图像处理及分析中重要意义,以及主要的图像分割技术。以木材表面缺陷为主要研究对象,利用微分算子边缘检测、最优迭代阈值分割及形态学方法针对具有代表性缺陷死节和虫眼进行分割处理。