The purpose of this demo is to illustrate how to extend TECA for a non-trivial use case and how to load and post process data produced. The demo makes use of a number of existing TECA algorithms (teca_cf_reader, teca_l2_norm, teca_binary_segmentation, teca_connected_components, teca_2d_component_area, teca_component_area_filter, teca_topological_spine, teca_jet_stream_sinuosity, teca_table_reduce, teca_table_sort, teca_table_calendar, teca_table_writer) and 2 new customized algoritghms written in Python (teca_topological_spine.py, teca_jet_stream_sinuosity.py). The demo comes with a command line application (teca_jet_stream_sinuosity) that constructs and runs the pipeline. A second script (plot_sinusoty.py) illustrates how to read in the resulting table and access the data. This demo has been tested with TECA version 3.0.0.

We use a binary segmentation of wind speed(30 m/s) high in the atmosphere(200 hPa) followed by a medial axis transform followed by a pruning step to compute a topological spine of the jet stream. In the above figure the segmentation is bound by black lines, wind speed is shown by pseudocolor, and topological spine is shown as a green line in the northern hemisphere and a blue line in the southern hemisphere.

We compute the sinuosity by dividing the length along the topological spine by the length of an arc of parallel latitude between the most westerly and easterly longitude at the mean latitude. Length along the spine is calculated using Vincenty's formula for great circle distance. In the above figure the green line shows the spine and the yellow line shows the distance between westerly and easterly points.

The command line application runs in parallel and produces a table of sinuosity with a row for northern hemisphere and southern hemisphere for each step in the input dataset. A second app can be used to load this table and plot the results as shown in the above figure.

On cori the code can be run using the included run_demo_cori.sh script on a compute node.

mpiexec -np 10 ./teca_jet_stream_sinuosity \
    --input_regex=/home/bloring/work/teca/jet_stream_data/cam5_1_amip_run2'.*\.nc'  \
    --output_file=sinuosity.bin --area_threshold 7.e6

python plot_sinuosity.py sinuosity.bin sinuosity.png