Implementation strategy for data assimilation into AWRAv0.5

[摘要] This report describes work that has been undertaken to assess the suitability of generic data assimilation (DA) packages for inclusion with the Australian Water Resource Assessment (AWRA) system. A second purpose was to propose an implementation plan for DA in AWRA. The purpose of DA is to reduce the uncertainty in dynamic model predictions through the optimal combination of observations and model outputs. In other words, DA allows generation of model predictions of a higher confidence by blending in information from a suitable range of observation data. DA, if implemented correctly, will improve the predicative skill of AWRA.The DA method favoured by the AWRA team is the Ensemble Kalman Filter (EnKF) with the main attraction of the technique being its relative simplicity. Due to the inherent complexity of DA, it is desirable to assess existing toolsets for use with AWRA, which is consistent with the “adopt, adapt, invent” mantra of Water Information Research and Development Alliance (WIRADA).Two DA packages were identified as possibly being suitable for AWRA, the NASA Land Information System (LIS) and Deltares OpenDA. The assessment undertaken was largely qualitative and was based on the team’s practical experience with the toolsets. It was undertaken in two stages: an initial familiarisation, during which a number of qualitative evaluation criteria were developed, and a subsequent case study, in which a more detailed assessment was performed. During the initial familiarisation, the team found the complexity of LIS to be very high and the documentation reasonable for simple application although deficient for specific application and support. In particular, LIS was difficult to install, complex to run, and required a Fortran version of AWRA-L to be developed. Similarly with OpenDA, LIS is a very complex software system that requires from the modeller not only comprehensive knowledge of the physical problem but good understanding of DA theory and methods and contemporary software engineering approaches. After the initial familiarisation, due to the complexity, licensing issues, poor documentation and support, software incompatibilities and the finite resources of the team it was not possible to undertake a suitable case study for LIS, which made the further qualitative assessment even more difficult.The detailed assessment of suitability of OpenDA and LIS for AWRA was undertaken in qualitative terms such as Ideal, Substantial, Moderate, Acceptable and Poor.The qualitative evaluation showed that both packages were equally highly rated as Ideal or Substantial for being generic and functional, having a satisfactory range of DA algorithms and ability for extending the DA libraries. Both packages were equally rated as Poor for the Programming platform criterion as both required adapting the AWRA-L code to their programming languages.Except for software cost and documentation, both packages were rated relatively low for the rest of criteria. In rare cases of successful communication with the tool developers, their help was never in time and/or sufficient. This is probably because the user support service has no priority for the developers of LIS and OpenDA. The user communities are still too immature to provide technical and developer support at the level required by new comers. Both packages are available free, but LIS has strict licensing conditions and the software is not open source. OpenDA is open source, but poorly documented compared to LIS.AWRA-L was not implemented into LIS. Therefore the performance and implementation criteria were evaluated for OpenDA only.Performance was rated as Moderate and OpenDA required a lot of time and team effort and we rated the Implementation cost and Ease of implementation criteria as Poor. Through this careful evaluation of both LIS and OpenDA, we came to the conclusion that neither of these existing technologies were suitable for implementation in AWRA. The recommendation fr

[发布日期] 2012-05-30 [发布机构] CSIRO

[学科分类] 地球科学(综合)