• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>ACRS 2011;Asian conference on remote sensing >EVALUATING PERFORMANCE OF MAXENT AND EFFECTS OF SAMPLING STRATEGIES ON MODELING FOREST TREE SPECIES-Schima superba
【24h】

EVALUATING PERFORMANCE OF MAXENT AND EFFECTS OF SAMPLING STRATEGIES ON MODELING FOREST TREE SPECIES-Schima superba

机译：林木树种最大优势的评估性能和抽样策略的影响

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Predictive model for species distribution has been the core of ecological research since late 20th century with the development of statistical techniques and 3S tools, which not only can be applied to biodiversity conservation and management, but also enhance the ability of predicting species habitat distribution. The sample points of Chinese guger-tree (Schima superba, CGT) in the Huisun study area were obtained by GPS, and GIS technique was used to overlay five environmental factors (including terrain factors and vegetation index derived from SPOT-5 satellite images). Besides, we designed different ratios of background to target to evaluate different sampling designs for modeling individual species' distribution. The study applied MAXENT, DT and DA models to predict the suitable habitat of CGT in Huisun. The results showed that the accuracy of DT was only slightly better than that of MAXENT and accuracies of the two models were much better than that of DA. Implementation of model creation and validation was efficient, but it needed a cross-platform operation for modeling and mapping CGTs' suitable habitats. More importantly, MAXENT and DT can greatly reduce the area of field survey to 6% of the entire study area at the first stage, and thus saving both cost and labor. However, increasing background samples was not always beneficial to model accuracy. Especially when the ratio of background to target became too large, the species prediction did not correspond with real distribution, thereby reducing model accuracy. This ratio falling within the range from one to five was good for species distribution modeling, but may not be optimal. Next studies will attempt to incorporate predictor variables with species spectral information extracted from high spatial, spectral resolution remotely sensed data into predictive models so that newly developed models can be applied at a larger spatial scale.
机译：自20世纪末以来，随着统计技术和3S工具的发展，物种分布的预测模型已成为生态学研究的核心，不仅可以应用于生物多样性的保护和管理，而且还可以增强预测物种栖息地分布的能力。利用GPS获取了惠顺研究区古树的采样点，并利用GIS技术覆盖了5个环境因子（包括SPOT-5卫星图像中的地形因子和植被指数）。此外，我们设计了不同的目标背景比率，以评估用于模拟单个物种分布的不同采样设计。该研究运用MAXENT，DT和DA模型来预测惠顺CGT的适宜生境。结果表明，DT的精度仅略高于MAXENT，两个模型的精度均远优于DA。模型创建和验证的实施是有效的，但是它需要跨平台操作来对CGT的合适栖息地进行建模和映射。更重要的是，在第一阶段，MAXENT和DT可以将实地调查的面积大大减少到整个研究区域的6％，从而节省了成本和劳力。但是，增加背景样本并不总是对模型准确性有利。特别是当背景与目标的比例太大时，物种预测与实际分布不符，从而降低了模型的准确性。该比率在1到5的范围内对于物种分布建模是好的，但可能不是最佳的。接下来的研究将尝试将预测变量与从高空间，光谱分辨率遥感数据中提取的物种光谱信息合并到预测模型中，以便可以在更大的空间规模上应用新开发的模型。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号