A monitoring toolkit for banksia woodlands: comparison of different scale methods to measure recovery of vegetation after fire
A monitoring toolkit for banksia woodlands: comparison of different scale methods to measure recovery of vegetation after fire
复制标题
DOI：
10.1002/rse2.88
复制
发表时间：
2018-07
影响因子：
5.5
通讯作者：
Mark C. Brundrett;R. Dongen;B. Huntley;N. Tay;V. Longman
中科院分区：
文献类型：
--
作者：
Mark C. Brundrett;R. Dongen;B. Huntley;N. Tay;V. Longman
在这里，我们比较了遥感和基于地块的方法测量西澳大利亚城市地区班克西亚林地林下和树冠植被覆盖的效率和准确性。方法比较了目测法、从照片、卫星图像和航空照片计算植被盖度的方法。1m2、100m2和625m2样方的观测和图像分别测量了小型植物、林下植物和乔木的盖度。通过航空摄影和卫星图像，可以估计出625平方米和1公顷地块的树木数量、高度和覆盖率。方法的准确性是使用28个月的时间序列进行比较的，时间序列开始于一场强烈的森林大火之前和之后，这场大火移除了所有的树叶覆盖。直接可比较的方法非常一致，结合起来可以非常详细地量化植物回收率。目测野外盖度很耗时，但对衡量个别物种的贡献是必要的。从1m2向下的照片中的目测估计允许测量植物的功能群。在照片中手动选择的绿色像素数量证实，使用算法从地面照片计算的覆盖率是准确的，除非覆盖率很低。我们开发了一种新的算法，用于从照片中计算覆盖率，该算法在低覆盖率下是准确的(Gperc)。利用算法估计冠层覆盖度存在较大误差，需要剔除部分图像。陆地卫星卫星图像使人们能够在1988年以来相对一致的季节变化的背景下确定严重干旱和以前的火灾的影响。从1953年开始的航拍照片显示，在砍伐树木后的60多年里，班克西林地的树木逐渐重新定居。这些方法为监测干扰后植被恢复提供了工具包，为监测班克西亚林地提供了基线数据。这个工具包也应该适用于大多数其他植物群落。
Here, we compare the efficiency and accuracy of remote sensing and plot‐based methods for measuring vegetation cover for the understory and canopy of banksia woodland in an urban area of Western Australia. Methods compared were visual estimation, foliage cover computation from photographs, satellite imagery and aerial photographs. Observations and images from 1 m2, 100 m2 and 625 m2 quadrats measured cover of small plants, understory plants and trees respectively. Aerial photography and satellite imagery allowed the number, height and cover of trees to be estimated in 625 m2 and 1 ha plots. The accuracy of methods was compared using a 28 month time series commencing before and after an intense bushfire that removed all foliage cover. Directly comparable methods were in close agreement and in combination allowed plant recovery to be quantified in great detail. Visual estimation of cover in the field was time‐consuming but necessary to measure the contribution of individual species. Visual estimates from 1 m2 downward photos allowed functional groups of plants to be measured. The number of green pixels selected manually in photographs confirmed that cover calculated from ground‐based photographs using algorithms was accurate, except when cover was very low. We developed a new algorithm for computing cover from photographs that was accurate at low cover (Gperc). Canopy cover estimation by algorithm from upward photographs was subject to more errors, requiring exclusion of some images. Landsat satellite images allowed the impacts of severe drought and previous fires to be identified against a background of relatively consistent seasonal variations since 1988. Aerial photographs from 1953 onwards showed gradual recolonisation by banksia woodland trees over 60 years following tree felling. These methods provide a toolkit for monitoring vegetation recovery after disturbance and baseline data for monitoring banksia woodland. This toolkit should also be suitable for most other plant communities.