• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>The photogrammetric journal of Finland >ACCURACY OF GEOCOVER SATELLITE IMAGE MOSAICS FOR TIMBER VOLUME MAPPING IN BOREAL FORESTS IN FINLAND
【24h】

ACCURACY OF GEOCOVER SATELLITE IMAGE MOSAICS FOR TIMBER VOLUME MAPPING IN BOREAL FORESTS IN FINLAND

机译：芬兰北部森林木材体积映射的地覆盖卫星图像卫星精度。

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Medium resolution satellite data have recently become available at low or no cost. GeoCover dataset consisting of Landsat TM/ETM+ images covers most of the earth's land surface, and is available for users as large mosaics and individual images. Here we focused on evaluating the estimation accuracy of timber volume and tree species proportion, using GeoCover mosaic data. Interpretation was carried out using regression analysis. Three field datasets with a total of 1083 sample plots from three different vegetation zones in eastern Finland were used in modeling. Separate stand volume models were developed for three individual GeoCover images as well as for a GeoCover mosaic consisting of 41 images. The results of the individual image interpretation were compared with those of the mosaic interpretation. The relationship between the interpretation result and the location of the sample clusters was also evaluated. In addition to the spectral values, regional average volume statistics of National Forest Inventory were also used as explanatory variables.The RMSE (%) values for the stand volume models based on individual images were between 46% and 49%. For the model based on the GeoCover mosaic the RMSE (%) value was 51%. The result was evaluated by classifying the estimates in groups of 50 m~3/ha. In one of the three areas the accuracy of mosaic interpretation was the same or better than that of individual image interpretation, while in the other two areas the single image interpretation was more accurate. The accuracy of interpretation was highly dependent on the distance from the sample area. For estimating the proportion of different tree species, a seemingly unrelated regression (SUR) method was used. The recognition of different tree species was rather unreliable. The separation between coniferous species (Scots pine and Norway spruce) was especially inaccurate, since the dominant species in the sample data also predominated in the estimation. The stand wise separation between coniferous and deciduous stands could be done with 79% accuracy.
机译：中分辨率卫星数据最近已经可以低成本或免费获得。由Landsat TM / ETM +图像组成的GeoCover数据集覆盖了地球的大部分陆地表面，并可供用户使用大型镶嵌图和单个图像。在这里，我们专注于使用GeoCover镶嵌数据评估木材量和树木物种比例的估计准确性。使用回归分析进行解释。在建模中使用了三个现场数据集，总共有来自芬兰东部三个不同植被带的1083个样地。针对三个单独的GeoCover图像以及一个包含41个图像的GeoCover马赛克开发了独立的展台体积模型。将单个图像解释的结果与镶嵌解释的结果进行了比较。还评估了解释结果与样本簇位置之间的关系。除了光谱值外，还使用国家森林清单的区域平均体积统计数据作为解释变量。基于单个图像的林分体积模型的RMSE（％）值在46％至49％之间。对于基于GeoCover镶嵌的模型，RMSE（％）值为51％。通过将估计值按50 m〜3 / ha的组进行分类来评估结果。在这三个区域之一中，镶嵌图像的解释精度与单个图像解释相同或更好，而在其他两个区域中，单个图像解释的准确性更高。解释的准确性高度取决于与样品区域的距离。为了估计不同树种的比例，使用了一种看似无关的回归（SUR）方法。对不同树种的识别相当不可靠。针叶树种（苏格兰松树和挪威云杉）之间的分离特别不准确，因为样品数据中的优势树种在估计中也占主导地位。针叶林和落叶林之间的林分间隔可以79％的精度完成。

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