What would you like to suggest?
To efficiently query RAPID output with standard Python-based netCDF libraries (e.g. netCDF4, xarray), it is necessary to transpose the simulated discharge (“Qout”) variable such that the first dimension corresponds to the “rivid” variable and the second dimension corresponds to the “time” variable. In our current workflow for visualizing RAPID simulations, we read the RAPID-generated output file, transpose the Qout variable, write the transposed variable to a new file, and delete the original file. This postprocessing step entails a significant amount of computational overhead. We propose to modify RAPID so that it provides an option to write the Qout variable with the ordered dimensions [“rivid”, “time”]. This would eliminate the costly postprocessing step and improve the efficiency of the existing workflow.
Additionally, the "time" variable in the output file is currently assigned an "unlimited" dimension. The lack of a fixed dimension for the time variable makes querying the data less efficient. We currently assign the time dimension a fixed size when rewriting the Qout file, which significantly reduces querying time. We propose to modify RAPID to allow a fixed dimension to be assigned to the time variable.
We hope these adjustments might benefit existing and future RAPID-based workflows written in row-major languages.
Let us ensure open source development is sustainable!
What would you like to suggest?
To efficiently query RAPID output with standard Python-based netCDF libraries (e.g. netCDF4, xarray), it is necessary to transpose the simulated discharge (“Qout”) variable such that the first dimension corresponds to the “rivid” variable and the second dimension corresponds to the “time” variable. In our current workflow for visualizing RAPID simulations, we read the RAPID-generated output file, transpose the Qout variable, write the transposed variable to a new file, and delete the original file. This postprocessing step entails a significant amount of computational overhead. We propose to modify RAPID so that it provides an option to write the Qout variable with the ordered dimensions [“rivid”, “time”]. This would eliminate the costly postprocessing step and improve the efficiency of the existing workflow.
Additionally, the "time" variable in the output file is currently assigned an "unlimited" dimension. The lack of a fixed dimension for the time variable makes querying the data less efficient. We currently assign the time dimension a fixed size when rewriting the Qout file, which significantly reduces querying time. We propose to modify RAPID to allow a fixed dimension to be assigned to the time variable.
We hope these adjustments might benefit existing and future RAPID-based workflows written in row-major languages.
Let us ensure open source development is sustainable!
I have looked at previous GitHub issues and this enhancement has not yet been proposed.
I have already submitted one unique GitHub issue with my RAPID User Information.
I have read the last paragraph of Section 6.3 in David et al., 2016.