Assessment of White Gum Moist Forest on NSW Crown Forest Estate

An indicative map for White Gum Moist Forest (WGMF) was constructed to resolve long-standing issues surrounding its identification, location and extent within the NSW State Forest estate covered by the eastern Regional Forest Agreements. \r\nThe determination of WGMF was reviewed by the project’s Threatened Ecological Community (TEC) Reference Panel (the Panel), and a set of diagnostic parameters for identifying the WGMF TEC was agreed. \r\nOur mapping process relied upon the occurrence of E.dunnii to diagnose the presence of WGMF. \r\n\r\nWe reviewed existing vegetation maps, predictive models and observation records of E.dunnii to identify State Forests that are known or likely to include stands of the species. We then attempted several different approaches to sampling and mapping E.dunnii using ground based surveys, predictive modelling and aerial photograph interpretation (API). We used API assessment of known E.dunnii stands to identify image patterns and signatures that indicated the presence of E.dunnii. We used our findings to examine un-surveyed areas of relevant state forests via API, and then we mapped any areas which appeared to be dominated or co-dominated by E.dunnii. We also developed a Random Forest presence-absence model and used it to predict the distribution of WGMF across its range.\r\nWe constructed an indicative map of WGMF using the combined results of our API mapping and our predictive model. In total, we mapped approximately 980 hectares of forest likely to be dominated or co-dominated by E.dunnii across 16 State Forests. Two thirds of the mapped area is associated with the northern populations of E.dunnii – the largest areas were in Beaury, Donaldson and Yabbra State Forests. In the southern area, Kangaroo River State Forest includes the largest representation of E.dunnii in State Forest. \r\nOur conclusions from this exercise is that our API interpretation is capable of separating E.dunnii from other related eucalypts but only where it is supported by field reconnaissance. Therefore, further work is required to increase API confidence throughout its range before our maps are suitable for operational applications. Nonetheless, our indicative map is still useful for providing a list of State Forests that include mapped areas of E.dunnii and identifying the areas that have corroborating field based evidence of E.dunnii.\r\nAs our indicative map stands at present, we consider that it overestimates the extent of E.dunnii and its dominance, however, it is unlikely that extensive stands exist outside our mapped areas. We also conclude that existing mapping (including both forest type mapping and OEH (2012) mapping) significantly underestimates the likely true extent).\r\n\r\nIndicative TEC Mapping have been generated from best available composite environmental data layers - standardised to 30 m pixels.

本研究构建了白胶湿润林（White Gum Moist Forest, WGMF）的指示地图，以解决长期以来困扰学界与实践中，该森林类型在东部区域森林协议（Regional Forest Agreements）覆盖的新南威尔士州（New South Wales, NSW）州立森林范围内的识别、分布与范围界定难题。
本项目的濒危生态群落（Threatened Ecological Community, TEC）咨询委员会（下称委员会）对WGMF的判定标准开展评审，并就识别WGMF TEC的一套诊断参数达成共识。
本次制图流程以邓恩桉（E.dunnii）的出现作为WGMF的诊断依据。
我们梳理了现有植被地图、预测模型以及邓恩桉的观测记录，以识别已知或疑似分布有该树种林分的州立森林。随后我们尝试了多种采样与制图方法，包括地面调查、预测建模以及航片解译（Aerial Photograph Interpretation, API）。我们通过对已知邓恩桉林分的航片解译，确定了指示邓恩桉存在的影像特征与标识。我们利用上述发现，通过航片解译对相关州立森林的未调查区域进行排查，并将其中以邓恩桉为优势种或共优势种的区域进行制图。此外，我们还构建了随机森林（Random Forest）存在-不存在模型，用于预测WGMF在其分布范围内的分布情况。
我们结合航片解译制图与预测模型的结果，构建了WGMF的指示地图。总计在16处州立森林中，共绘制了约980公顷疑似以邓恩桉为优势或共优势种的森林区域。其中三分之二的制图区域对应邓恩桉的北部种群，最大的分布区域集中在比尤里（Beaury）、唐纳森（Donaldson）与亚布拉（Yabbra）州立森林。南部区域中，袋鼠河（Kangaroo River）州立森林拥有州立森林中规模最大的邓恩桉分布。
本研究得出的结论为，航片解译能够将邓恩桉与其他近缘桉属树种区分开来，但该效果需依赖野外踏勘的支持。因此，在本地图可投入业务应用之前，仍需开展更多工作以提升其在整个分布范围内的解译置信度。尽管如此，本指示地图仍可用于梳理包含邓恩桉制图区域的州立森林清单，并识别拥有邓恩桉实地佐证证据的区域。
就当前版本的指示地图而言，我们认为其高估了邓恩桉的分布范围及其优势程度，但目前并无证据表明在我们绘制区域之外存在大面积的邓恩桉林分。我们同时发现，现有制图（包括森林类型制图与新南威尔士州环境与遗产办公室（Office of Environment and Heritage, OEH）2012年制图）显著低估了其实际分布的真实范围。
本次生成的濒危生态群落指示地图，基于当前可获取的最佳复合环境数据图层构建，并统一采样为30米像素分辨率。