Native Vegetation of the Murray Catchment Management Authority Area. VIS_ID 3808, VIS_ID 3809, VIS_ID 3810, VIS_ID 3811

Native vegetation was delineated into stands using feature recognition software. A hybrid classification method that combined spatial modelling and visual interpretation was used to combine the features and create a vegetation map.\r\n\r\n\r\n\r\nSPOT 5 and Landsat satellite imagery was used in the creation of image objects. The spectral response of individual SPOT 5 scenes varied widely across the catchment so it was not used in the classification of vegetation type. Spatial layers used in the classification included a Digital Elevation Model (DEM), Landsat reflectance data, radiometric data and soil and climate layers, all of which are available for the entire State. Over 340 new full floristic survey sites were commissioned and the results were combined with 900 existing survey site records to create training areas for spatial modelling. Each survey site was assigned a New South Wales Vegetation Classification and Assessment (NSWVCA) vegetation type.\r\n\r\n\r\n\r\nThe relationship between survey sites and spatial layers was explored by using machine learning software and vegetation type was classified using an object-based nearest neighbour approach. The catchment was divided into three discrete spatial models with separate training and validation survey sites. Model performance was assessed on the basis of the number of NSWVCA types mapped correctly in five classes of precision. The percentage of correctly modelled vegetation types ranged between 58% and 68%.\r\n\r\n\r\n\r\nSeveral vegetation community types were not able to be modelled (e.g. chenopods) or were poorly modelled due to lack of sample data. These communities were added or amended based on the visual interpretation of remotely sensed data. The amended map was assessed against a limited subset of independent survey data. The percentage of correctly mapped vegetation types in five classes of precision ranged between 72% and 78%.\r\n\r\n\r\n\r\nThe mapping was presented in a geodatabase, which allows for user-generated updates so that the product can evolve as more field data are collected.\r\n\r\n\r\n\r\nROFF, A., SIVERTSEN, D., AND DENHOLM, B. 2010. The Native Vegetation of the Murray Catchment Management Authority Area, NSW Department of Environment, Climate Change and Water, Sydney, Australia.\r\n\r\n\r\n\r\nVIS_ID 3808 VIS_ID 3809 VIS_ID 3810 VIS_ID 3811

本研究采用特征识别软件对原生植被进行斑块划分，并结合空间建模与目视解译的混合分类方法整合特征信息，生成植被分布图。

本研究利用SPOT 5与Landsat卫星影像构建影像对象。由于研究流域内单景SPOT 5影像的光谱响应差异显著，因此未将其用于植被类型分类。分类所用的空间图层涵盖数字高程模型（Digital Elevation Model, DEM）、Landsat反射率数据、辐射测量数据以及土壤与气候图层，上述图层均可在全州范围内获取。本研究委托开展了340余个全新的完整植物区系调查样点，并将其结果与900份现存调查样点记录相结合，以此构建空间建模的训练区域。每个调查样点均被赋予新南威尔士州植被分类与评估（New South Wales Vegetation Classification and Assessment, NSWVCA）的植被类型标签。

本研究借助机器学习软件探究调查样点与空间图层之间的关联，并采用基于对象的最近邻分类法完成植被类型划分。研究将流域划分为三个独立的空间模型，分别设置独立的训练与验证样点。模型性能以5个精度等级下正确匹配的NSWVCA植被类型数量为依据进行评估，经建模正确识别的植被类型占比介于58%至68%之间。

部分植被群落类型（如藜科植被类群）因样本数据不足，无法完成建模或建模效果欠佳。针对此类群落，本研究通过遥感影像目视解译进行补充或修正。修正后的植被分布图通过有限的独立调查样点子集进行精度验证，5个精度等级下正确匹配的植被类型占比介于72%至78%之间。

本植被分布图以地理数据库（geodatabase）形式发布，支持用户自主更新，以便随着野外调查数据的积累不断完善该成果。

ROFF A、SIVERTSEN D、DENHOLM B，2010年。《墨累流域管理局区域原生植被》，澳大利亚悉尼新南威尔士州环境、气候与水资源部。

VIS_ID 3808 VIS_ID 3809 VIS_ID 3810 VIS_ID 3811