Yield Stability of Winter Wheat (Triticum sp.) Cultivars and Lines

The purpose of this study was to compare responsiveness to environment as well as the stability of newly developed promising lines with traditional commercial cultivars. Subject research was conducted in Ankara between 1983-1989 on common and durum wheat cultivars and lines. Regression coefficient (b) was used as the criterion of genotypes' responsiveness to environment whereas deviation from regression (S²d) and coefficient of determination (r²) were used as stability parameters. Each experiment year was regarded as an environment and yield average of each year was used as the environmental index.
In consequence of the research, it was determined that new varieties with high yield, adaptable to diverse environments could be developed by means of hybridizations between parents having different genetic characteristics which were provided from diverse ecological regions.     

呼伦贝尔市农业结构调整最优方案

根据不同区域气候特点和产量水平,运用系统工程理论中的线性规划方法对该市的作物种植结构进行了调整,在种植面积基本不变的情况下,净产值增加了15.3%和17.3%;并在不同的气候区域设计了不同作物种植的最佳组合方案。这对增加农业种植效益和农民收入,有效改善生态环境,保持农业的可持续发展等方面都具有显著的经济、社会和生态效益。     

北京八达岭人工油松林地表枯死可燃物负荷量研究

地表枯死可燃物负荷量多少是森林火灾发生的基础。调查八达岭林场不同地形条件下油松林地表枯死可燃物和林分因子,取得42块样地资料;用线性相关分析法对各因子与地表枯死可燃物负荷量进行分析,平均树高、坡度和海拔与对应负荷量有显著线性关系;应用多元线性回归中的逐步回归剔除法,建立了估测模型,除平均胸径外,平均树高、密度、郁闭度、海拔、坡度被引入,经检验回归显著,该估测模型可用于其它油松林地表可燃物负荷量估测。     

基于多方位视觉的果实形状特征的提取研究

针对传统瓜果分级中捕捉单幅果实图像来分析果实特征时信息较少的缺陷,提出了一种基于多方位视觉系统的果实形状特征提取算法。同步采集3个方向的果实图像信息,经OHTA空间上的图像分割和改进的BLOB算法去噪声等处理后,分别给出果实的多方位的图像特征信息。运用统计学原理,设计了线性分类器,对3个方向的图像特征信息进行了互补性融合,得出分级判别准则。利用开发的智能瓜果精选分级试验样机,验证了该算法的有效性, 试验结果证明识别成功率达到97％,达到了实用的要求。     

脐橙糖度光谱图像检测技术研究

对采集到的不同波长的光谱图像灰度分布进行洛伦茨分布(LD)、高斯分布(GD)、指数分布(ED)函数拟合,通过比较发现洛伦茨分布为最优灰度分布拟合函数.将脐橙的糖度与洛伦茨分布函数拟合所得参数分别进行多元线性回归,建立最佳单波长、最佳双波长组合、最佳三波长组合和最佳四波长组合的校正方程.结果表明:利用光谱图像技术无损检测脐橙糖度是可行性的.     

Presence–absence surveys of prey and their use in predicting leopard (Panthera pardus) densities: a case study from Armenia

It is important to predict how many individuals of a predator species can survive in a given area on the basis of prey sufficiency and to compare predictive estimates with actual numbers to understand whether or not key threats are related to prey availability. Rugged terrain and low detection probabilities do not allow for the use of traditional prey count techniques in mountain areas. We used presence–absence occupancy modeling and camera‐trapping to estimate the abundance and densities of prey species and regression analysis to predict leopard (Panthera pardus) densities from estimated prey biomass in the mountains of the Nuvadi area, Meghri Ridge, southern Armenia. The prey densities were 12.94 ± 2.18 individuals km^?2 for the bezoar goat (Capra aegagrus), 6.88 ± 1.56 for the wild boar (Sus scrofa) and 0.44 ± 0.20 for the roe deer (Capreolus capreolus). The detection probability of the prey was a strong function of the activity patterns, and was highest in diurnal bezoar goats (0.59 ± 0.09). Based on robust regression, the estimated total ungulate prey biomass (720.37 ± 142.72 kg km^?2) can support a leopard density of 7. 18 ± 3.06 individuals 100 km^?2. The actual leopard density is only 0.34 individuals 100 km^?2 (i.e. one subadult male recorded over the 296.9 km²), estimated from tracking and camera‐trapping. The most plausible explanation for this discrepancy between predicted and actual leopard density is that poaching and disturbance caused by livestock breeding, plant gathering, deforestation and human‐induced wild fires are affecting the leopard population in Armenia.     

基于 GIS 的中国草地合理载畜量空间化

天然草地合理载畜量是反映草地合理利用强度的基本单位。为提高草地合理载畜量的划分精度,根据“极大似然逼近原理”,创建了异质因子控制下的“线性回归+区域化值残差修正”数据空间化方法。以全国草地资源调查汇总数据中县域理论载畜量和1∶400万中国草地资源图为基础数据,采用分省控制构建了理论载畜量与植被类型-生产力异质植被斑块的线性回归模型,其判定系数由全国模型下的0.867提高到0.937,拟合优势度较高,所获得的中国公里格网草地合理载畜量分布图空间化精度也较高。中国公里格网草地合理载畜量分布图不仅表明了不同类草地因生产力的不同有不同的放牧要求,也体现了不同环境条件下草地载畜能力的差别,有效反映了单位面积内草地植被载畜-生产力水平,为生态治理和草地资源保育工作中合理制定草地利用强度提供了理论参考。     

基于广义混合效应模型的云冷杉林天然更新计数方法研究

天然更新是森林恢复较好的办法,对未来林分结构和生物多样性具有深远的影响。更新模型能够模拟天然更新的现实和未来状况,为森林经营者提供准确的森林计划。以吉林省汪清林业局2013年设置的12块云冷杉针阔混交林为例,选择泊松和负二项分布形式,考虑样地间的随机效应,构建基于林分因子的云冷杉针阔混交林天然更新模型。结果表明：各个树种对林分因子的反应不一,白桦更新株数与林分每公顷株数和平均直径均呈负相关;红松和水曲柳更新株数不受各个林分因子的影响,更新株数是随机的;冷杉更新株数与林分每公顷断面积呈负相关;色木槭和云杉更新株数均与林分平均直径呈负相关。在考虑样地的随机截距效应后,模型的模拟效果显著提高。在构建天然更新模型时,林分密度是十分重要的因子,如果要人工促进天然更新,就要科学合理的采取经营措施,以确保合理的林分密度。另外样地间的差异是不容忽略的因素,需要利用混合效应模型方法来降低预测误差。     

线性红利界下带干扰风险模型的破产概率

考虑到保险公司在实际经营中收益所具有的不确定性和分红策略，建立一类具有线性红利界和带随机扰动的双复合Poisson风险模型，利用鞅方法给出模型关于破产概率的一个定理及上界．     

E－凸集的几个基本性质

利用比较与分析的方法，由E-凸集的定义出发，研究了E-凸集的若干基本性质．所得结果丰富并深化了凸分析的基本内容，拓宽了广义凸性在优化理论及其应用中的研究范围．