What's next for Christchurch surface water modelling?
❄
ABSTRACT
| Motivations are prompting reconsideration of the major pillars of model schematisation, triggered by lessons learned from the results, opportunities from new technologies and motivations to do it better or cheaper. Some change decisions are almost made, and others are early in consideration. This presentation summarises the major themes of potential change, motivations, options considered and what are seen as likely future choices. Flexible mesh: The triangular mesh has worked well, however its cost, learnings about computational performance and the availability of new alternatives is prompting reconsideration of the mesh style. Simple uniform grid and quadtree type variable grid are in the early phases of consideration. Model integration: A desired outcome from Christchurch modelling since the inception of the citywide modelling programme, work on integration is progressing, with the primary initiative being to share timeseries data at the common 1D boundaries between the models. A secondary change will realign model boundaries strictly with the 2D catchment delineation and ignore 1D catchments where they conflict. This will increase the number of 1D shared boundary points, but will result in much improved 2D flood results in key areas. Data centric modelling: This refers to the opportunity to develop the ability to build and update models more quickly and cost effectively by having the right forms and quality of primary data records, coupled with data conversion processes to significantly reduce model build and maintenance costs. Reduction in M11: A number of drivers including model build simplicity, inlet connectivity and computational stability are suggesting a reduction in the extents of M11 in the open channel modelling. Small urban channels would be converted to Mike Urban (and the future Mike 1D equivalent), and larger river channels converted to M21. Depth tolerance: There are legacy issues in parts of the models with water level discontinuities at the river / floodplain interfaces. Initial work suggests a simple approach to reduce this tolerance may be effective. Further work is planned to confirm the success of this approach. Other options have also been identified. Radar rainfall: Rainfall radar is being developed for modelling use with much improved data processing developed for Auckland Council to provide a robust and useful Quantitative Precipitation Estimate product from radar with demonstrated levels of confidence. Performance of the improved system on the Port Hills is a particular interest. Impervious coverage: New impervious coverage data is being developed at high resolution (ie: 0.075m) to supersede the older impervious coverage derived from satellite imaging with a 10m resolution with data blurring sometimes evident even at that coarse resolution LiDAR cross sections: Work has recently been commissioned to reprocess existing raw LIDAR data points with new objectives to derive drainage channel geometry, infer confidence in the derived geometry and learn where such data might be suitable for modelling and where other survey techniques might be needed. Feedback from the audience at the modelling symposium will be sought during Q+A and after the presentation. |