Bathymetry-embedded DEM for the Murray-Darling Basin version 2

Basin-wide Digital Elevation Models (DEMs) with embedded bathymetry for selected main rivers Murray-Darling Basin at 5 metre (GDA2020 Lambert Conformal Conic) and 1 second (WGS 1984) resolution. This collection is an enhancement of the Seamless Composite High Resolution Digital Elevation Model (DEM) for the Murray Darling Basin, whereby available channel bathymetry has been incised into the input DEM.This collection was produced for use in basin- wide flood extent and depth modelling which requires an accurate representation of channel bathymetry in the MDB's trunk rivers.Lineage: The base DEM is the High resolution Digital Elevation Model (DEM) for the Murray Darling Basin (https://data.csiro.au/collection/csiro:64134)BATHYMETRYBathymetry point data were received for large sections of the Murray River (Hume to Wellington), sections of the Darling Anabranches, Edward River, and for waterholes on the Darling and Barwon river. Additionally, bathymetry embedded LIDAR for the Murrumbidgee and Darling River LIDAR flown in 2015, when the riverbed was largely exposed due to dry conditions, was also utilised. See Collaborating Organisations and metadata document for complete list of bathymetry data sourcesSome small sections were clearly derived from gridded bathymetry having dense regular spacing of 10 to 15m. Most of the points were a continuous line of points following either a zig-zag or square wave track, with the remainder being transects perpendicular to the riverbanks at regular intervals ranging from a few hundred metres to many kilometres apart. Coverage for the Murray downstream of Lake Hume was continuous in some form (grid, track or transect) save for gaps between Narrung and Mildura. Various techniques were developed to process these data to form a consistent bathymetry ready to be embedded into the DEM. Grid derived points were interpolated directly to a 5m grid conforming to the DEM using Triangular Irregular Network (TIN) in ArcGIS. The remaining point configurations were unsuitable for direct interpolation to a DEM conforming raster in their received form, and so underwent a data point densification process to make them suitable. A method was devised to form gridded bathymetry from observed data points through data point densification. This was achieved by first creating a dense regular array of points consisting of 30 to 40 files of closely spaced (~10- ~20m) points across the width of the channel and following the course of the channel. Input bathymetry point data was transferred to its nearest (within 5m to 20m search radius) array point. Intervening array points within a given file that had not inherited a close neighbouring bathymetry datapoint value, then had a value linearly interpolated from its next upstream and downstream value. Bathymetry data points confer their bathymetry value to nearby array points. Remaining array points then have an interpolated value calculated based on the next upstream and downstream conferred data in their file. This dense array of data was then interpolated to the DEM conforming 5m raster using TIN. The rasterised bathymetry data were then inserted into the DEM replacing the non-ground channel values with interpolated bathymetry values. For the Murrumbidgee and Darling Rivers, existing LIDAR already had a good representation of river bathymetry, due to bathymetry embedding having already been implemented by a third party or the bed being exposed when LIDAR was flown. In these cases the LIDAR DEM values occurring within the channel were clipped out, resampled and reprojected to match the base DEM and then inserted into the base DEM. For both the rasterised interpolated point data and the clipped LIDAR inserts, bathymetry was embedded into the base DEM by taking the minimum elevation occurring in the overlying cells.This dataset was resampled to ≈ 30 m and used as an input to create the two-monthly maximum water depth spatial timeseries for the MDB version 2024 (https://doi.org/10.25919/1t0t-y110) and monthly maximum water depth spatial timeseries for the MDB (https://doi.org/10.25919/zffy-a921).针对墨累-达令盆地（Murray-Darling Basin，简称MDB）选定的主要河流，本数据集提供了嵌入测深（bathymetry）数据的全流域数字高程模型（Digital Elevation Model, DEM），分辨率分别为5米（采用GDA2020兰伯特正形圆锥投影）与1秒（采用WGS 1984坐标系）。 本数据集是对墨累-达令盆地无缝合成高分辨率数字高程模型的优化升级，将已获取的河道测深数据嵌入至原始输入DEM中。 本数据集专为全流域洪水范围与水深模拟而生成，该模拟需精准表征墨累-达令盆地主干河道的测深地形。 数据谱系：基础DEM采用《墨累-达令盆地高分辨率数字高程模型》（https://data.csiro.au/collection/csiro:64134）。 测深数据 研究团队获取了墨累河大片河段（休姆大坝至惠灵顿段）、达令分流河道部分河段、爱德华河河段以及达令河与巴旺河沿岸水塘的测深点数据。此外，2015年针对马兰比吉河与达令河航拍的激光雷达（Light Detection and Ranging, LIDAR）数据也被纳入使用，该次航拍时河道因干旱大面积裸露，可直接获取河床测深地形。完整测深数据源列表请参阅合作机构名单与元数据文档。 部分小型河段的测深数据明显源自网格格式的测深资料，其测点间距密集且规则，为10至15米。绝大多数测点以锯齿状或方波轨迹形成连续线状分布，其余测点则为垂直于河岸的断面测点，断面间距从数百米至数公里不等。休姆大坝下游的墨累河段均以网格、轨迹或断面形式实现了连续覆盖，仅在纳朗与米尔迪拉之间存在数据空白。团队开发了多种处理技术，以生成统一规范的测深数据，用于嵌入DEM。 对于网格衍生的测点数据，采用ArcGIS软件中的不规则三角网（Triangulated Irregular Network, TIN）工具，直接插值至与DEM匹配的5米网格中。其余测点格式在原始状态下无法直接插值至与DEM匹配的栅格中，因此需先进行测点加密处理，使其适配后续插值流程。 研究团队设计了一种从实测测点数据生成网格测深数据的测点加密方法：首先沿河道走向、横跨河道宽度创建密集规则点阵列，该阵列包含30至40个测点文件，每个文件内的测点间距约为10至20米，分布紧密。 将输入的测深点数据匹配至其最近邻的阵列点，搜索半径设置为5米至20米。对于单个文件中未获取到邻近测深点值的中间阵列点，则通过其上游与下游相邻测点进行线性插值，获取对应高程值。 测深点数据将自身的高程值赋予邻近的阵列点。剩余未赋值的阵列点则根据其所在文件内上游与下游已赋值的测点，计算得到插值高程值。随后，采用不规则三角网将该密集测点阵列插值至与DEM匹配的5米栅格中。最后，将栅格化后的测深数据嵌入至基础DEM中，用插值得到的测深高程值替换河道区域的非地面高程值。 针对马兰比吉河与达令河，由于第三方已完成河道测深数据嵌入，或是航拍激光雷达时河床裸露，现有激光雷达DEM已能较好表征河道测深地形。此类场景下，先裁剪出河道区域内的激光雷达DEM高程值，再对其进行重采样与投影转换，使其与基础DEM匹配，最终将处理后的数据嵌入至基础DEM中。 无论是栅格化加密后的测点插值数据，还是裁剪后的激光雷达插入数据，均通过取叠加单元格内的最小高程值，将测深数据嵌入至基础DEM中。 本数据集已重采样至约30米分辨率，作为输入数据用于生成墨累-达令盆地2024版两月最大水深空间时间序列（https://doi.org/10.25919/1t0t-y110）与逐月最大水深空间时间序列（https://doi.org/10.25919/zffy-a921）。