Potential vegetation distribution: New South Wales - 3arcsecond gridded 1990-centred baseline predictions of the pre-clearing extents of "Keith" Vegetation Classes using kernel regression with GDM-scaled environments for Vascular Plants

This collection contains 3-arcsecond gridded datasets (ESRI binary float format in WGS84) showing the baseline (1990-centred) predicted potential distribution of 102 (class numbers range between 1 and 125) "Keith" Vegetation Classes for New South Wales based on their correlation with baseline ecological environments (c.1990 climates, substrate and landform). The vegetation patterns and classification derive from a map for NSW compiled by David Keith. A kernel regression was used with a geographically even sample of 9,951 locations of training classes for the 102 classes attributed with 21 GDM-scaled environmental predictors for Vascular Plants representing baseline ecological environments. The training class data input to the kernel regression is provided with this package. The GDM-scaled environmental predictors, source biological data and model fit parameters are also provided with the data package. Using the 1990 baseline training class data, and without constraining the prediction to pre-existing map boundaries, the kernel regression predicted the potential distribution of the 102 Vegetation Classes using 1990-centred (30 year average) baseline climates derived from ANUCLIM v6.1 (Xu and Hutchinson 2011) and soil/geology/landform attributes. The kernel regression generates unconstrained probabilities varying in the range from 0 and up to 1 for each of the 102 classes. The data are provided as 3-arcsecond (approximately 90m), ESRI binary float grid format in WGS84. Each class is denoted “UNCON###”, where the number refers to the code originally assigned to that class in the vegetation map. A lookup table linking the vegetation classes to the output codes and descriptive title is provided. The methods are described in "Doerr, VAJ, Williams, KJ, Drielsma, M, Doerr, ED, Davies, MJ, Love, J, Langston, A, Low Choy, S, Manion, G, Cawsey, EM, McGinness, HM, Jovanovic, T, Crawford, D, Austin, M & Ferrier, S 2013, Designing landscapes for biodiversity under climate change: Final report, National Climate Change Adaptation Research Facility, Gold Coast, 260 pp.". A plain English description of the method used (but applied Nationally) can be found in the AdaptNRM Guide “Helping biodiversity adapt to climate change: a community-level modelling approach”, available online at: www.adaptnrm.org. Source of vegetation class data: KEITH, D. A. (2002) A compilation map of native vegetation for New South Wales. NSW Biodiversity Strategy, New South Wales Government. KEITH, D. A. (2004) Ocean shores to desert dunes, Hurstville, Department of Environment and Conservation (NSW).

本数据集集合包含分辨率为3角秒的网格化数据（采用WGS84坐标系下的ESRI二进制浮点格式），展示了新南威尔士州102个基思（Keith）植被类群（类别编号范围为1至125）基于其与基线生态环境（约1990年的气候、基质与地貌）的相关性，所预测得到的以1990年为中心的基线潜在分布。这些植被格局与分类体系源自戴维·基思（David Keith）为新南威尔士州编制的植被图。
本研究采用核回归（kernel regression）方法，针对上述102个植被类群，使用9951个地理分布均匀的训练样点样本，并结合21个经广义相异模型（Generalized Dissimilarity Model, GDM）缩放的维管植物基线生态环境预测因子开展建模。核回归所用的训练类群数据已随本数据包一并提供；经GDM缩放的环境预测因子、源生物数据以及模型拟合参数也同步随数据包附赠。
基于1990年基线训练类群数据，且未将预测结果约束于现有地图边界范围内，研究团队利用由ANUCLIM v6.1（Xu与Hutchinson 2011）生成的以1990年为中心（30年平均值）的基线气候数据，以及土壤、地质与地貌属性，通过核回归预测了102个植被类群的潜在分布。核回归会为每个类群生成取值范围介于0至1之间的无约束概率值。
本数据集以3角秒（约90米）分辨率、WGS84坐标系下的ESRI二进制浮点网格格式提供。每个类群均以"UNCON###"命名，其中###代表该类群在原始植被图中分配的专属代码。本数据包附带了一张将植被类群与输出代码及描述性标题相关联的查找表。
相关研究方法详见文献：Doerr, VAJ, Williams, KJ, Drielsma, M, Doerr, ED, Davies, MJ, Love, J, Langston, A, Low Choy, S, Manion, G, Cawsey, EM, McGinness, HM, Jovanovic, T, Crawford, D, Austin, M & Ferrier, S 2013, 《气候变化下生物多样性景观设计：最终报告》，国家气候变化适应研究设施，黄金海岸，共260页。
关于该方法（已在全国范围内应用）的通俗说明可参阅AdaptNRM指南《帮助生物多样性适应气候变化：社区级建模方法》，在线访问地址为：www.adaptnrm.org。
植被类群数据来源：
1. KEITH, D. A. (2002) 《新南威尔士州原生植被编制地图》，新南威尔士州生物多样性战略，新南威尔士州政府。
2. KEITH, D. A. (2004) 《从海岸到沙漠沙丘》，赫斯特维尔，新南威尔士州环境与保护部。