Contains a selection of potentially useful code snippets that support but are not a critical part of the main workflow.
Filename(s): ERA5_find_download_coordinates_from_shapefile.ipynb, ERA5_find_download_coordinates_from_shapefile.py
The ERA5 download scripts and several others related to downloading and processing parameter fields need to know the spatial extent of the domain in regular lat/lon format. This script loads the user-provided shapefile as specified in the control file, determines the bounding box of the shapefile and rounds these coordinates slightly down (for minimum latitude and longitude) and up (for maximum latitude and longitude). This ensures that downloads and cutouts based on these coordinates are slightly larger than the modelling domain.
This procedure is not integrated in the main workflow scripts because it is not always necessary (if the user has another way to determine the bounding box) and potentially quite slow for larger domains.
Filename(s): ERA5_check_merged_forcing_values.ipynb, ERA5_check_merged_forcing_values.py
After merging ERA5 surface and pressure level data into a single file, performing some rudimentary checks on the data can give peace of mind about the download and merging procedures. This script iterates over each merged file and performs a few sanity checks. It generates a log file listing for each forcing variable how often its value equals NaN or is missing and how often values fall outside user-specified ranges. It equally checks that the time dimension has equidistant and consecutive values (in each individual file, this is not between different files).
Filename(s): ERA5_subset_forcing_file_by_lat_lon.py
Subsets a smaller region from a larger ERA5 file. Useful to avoid lengthy re-downloading.
Filename(s): MERIT_fix_circular_downstream_ids.py
MERIT river reach shapefiles contain a column that for each reach indicates the ID of its downstream connection. In certain cases, this downstream ID is set to the reach' own ID creating a circular river path. This script checks the user's river reach shapefile for such occurrences. If any exist they are replaced with a downstream ID of 0, which mizuRoute indicates to mizuRoute that such reaches are the end of that particular stream section.
Filename(s): MIZUROUTE_split_out_to_statistics.sh
By default, mizuRoute generates simulation files that contain the full spatial domain but are split by calendar years. This script takes a (hard-coded) number of years, selects the two routing variables KWTroutedRunoff and IRFroutedRunoff as well as the segment ID variable reachID and stores the mean runoff per reach as a single output file.
Note that mizuRoute's output frequency can be adjusted by specifying the desired frequency in the mizuRoute control file as follows:
<newFileFrequency> month ! Options: day, month, annual (default), single
Filename(s): SUMMA_concat_split_summa.py
SUMMA's split-domain runs (i.e. those with the -g argument) result in output files that only contain data for the given subset of GRUs. This script concatenates multiple split-domain output files into a single file. Usage: python SUMMA_concat_split_summa.py [path/to/split/outputs/] [input_file_*_pattern.nc] [output_file.nc].
Filename(s): SUMMA_merge_restarts_into_warmState.py
SUMMA's restart files are intended to be used as initial condition files to either pick up a run from a given point or as estimates of the initial states for a new run. Restart files generated from a run with a subset of GRUs (i.e. using the -g argument) will only contain information for the selected subset of GRUs. This file concatenates multiple split-domain restarts into a single file.
Note that this function does not utilize the control_active.txt file and manual changes to the file will be needed. See the file itself for a description.
Filename(s): SUMMA_plot_computational_times.py
SUMMA keeps a high-level overview of computational time spent on a variety of sub-tasks on prints these as a summary to the screen on simulation end. If SUMMA's terminal output is instead saved to file(s), this script can be used to extract the timing information and summarize this info in a boxplot.
Filename(s): SUMMA_split_out_to_mizuRoute_split_in.py, SUMMA_split_out_to_mizuRoute_split_in.sh
SUMMA's split-domain runs (i.e. those with the -g argument) result in output files that only contain data for the given subset of GRUs, but for the full temporal domain for each GRU. mizuRoute can read inputs that are split across time (e.g. year1.nc, year2.nc, etc.) but requires that each file has data for all GRUs in the domain. This file converts SUMMA's some-grus-but-all-time.nc files into mizuRoute's required all-grus-but-some-time.nc files. Such conversion is mostly useful in cases of larger domains, where storing the full timeseries for all GRUs in one file is infeasible.
Note that mizuRoute's ability to read input files from a list is currently (2021-11-01) only available on the feature/mpi-pio branch.
Filename(s): SUMMA_summarize_logs.py
Log files are created for a run as usual when SUMMA is run with the option -g startGRU numGRU. If such a run is part of a parallel simulation where the modelling domain is divided into multiple chunks of GRUs, analyzing these individual log files for successes or graceful failures can be cumbersome. This script summarizes the log files by reading the final few lines in each log file and divides the full run into success/SUMMA error/early termination. It also reports the time needed for successful runs. The summary file is placed at the top of the log folder if no other name for the summary file is provided. Usage: python SUMMA_summarize_logs.py [log_folder] [optional: name_of_summary_file.ext] [optional: extension of log files (default: .txt)].
Note that this function assumes that the provided folder only contains SUMMA log files. Things may break if this is not the case.
Filename(s): SUMMA_timeseries_to_statistics_parallel.py
SUMMA can produce split-domain output files if the model is used with the -g <start> <num> command line option. This script analyzes the resulting files and summarizes the timeseries of a (set of) variable(s) into a statistical value. Currently implemented are finding the maximum and mean value across the entire time series. Outputs are stored in a single file that covers the full spatial extent of the domain.
Note that this requires the Python package multiprocessing, which is not included in the provided environment.yml and requirements.txt files.