In program `test_set_max_speed`, read `cont_list_proj` instead of
computing it. Computing it is not possible with a non
longitude-latitude grid.

Note that, as each shapefile should have an associated DBF file with
at least one field, we choose to put the SSH field in `cont_list_proj`
and the speed field in `cont_list`.

Read the same NetCDF files than those used to create `cont_list`. It
does not change anything since the window is going to be created
inside `test_set_max_speed`, so it is an unnecessary complication to
have input NetCDF files on a restricted domain. Furthermore, if we
want to use `test_set_max_speed` on a non longitude-latitude grid,
then it would be a problem to convert x, y corresponding to the grid
of `cont_list` to x, y corresponding to a restricted grid.

Make `cont_list` a polyline and `cont_list_proj` an SSH contour,
instead of vice-versa. The motivation is clarity in `get_1_outerm`: we
only define `cont_list` at the end.

Because we may drop all the contours, if the area is not large enough,
or some of the contours, if `n_cont == n_max_cont-1`, convert to
longitude and latitude only when we are are sure it is useful.

Transmit projection coordinates of contour to `mean_speed`.

Lionel GUEZ authored 8 months ago and GUEZ Lionel committed 8 months ago

Transmit `cont_list_proj` from `get_1_outerm` to `set_max_speed`.

Lionel GUEZ authored 8 months ago and GUEZ Lionel committed 8 months ago

This is equivalent for a return value of `good_contour`. It is a
little faster to test a logical value, maybe.

Lionel GUEZ authored 8 months ago and GUEZ Lionel committed 8 months ago

We do not need it. we can just have a local variable at initialization.

Lionel GUEZ authored 9 months ago and GUEZ Lionel committed 8 months ago

Do not orient nor compute the area of `cont_list(n_cont)`. When the
area is too small, then we have one more copy than before, of
`cont_list(n_cont)` to `out_cont`, but, on the other hand, we have one
less call to `ccw_orient` than before. So this change should not
impact performance. Note that the orientation and area of
`cont_list(n_cont)` were not used after `get_1_outerm`.

Preparing to keep `cont_list` in projection coordinates.

Lionel GUEZ authored 9 months ago and GUEZ Lionel committed 8 months ago

Procedure `good_contour` now takes input in projection coordinates and
returns a contour in projection coordinates. The dummy argument step
of `good_contour` is thus removed.

Lionel GUEZ authored 9 months ago and GUEZ Lionel committed 8 months ago

Promote corner from dummy argument of `input_ssh` to variable of
module `input_ssh_m`. The dummy argument corner of `set_all_extr` is
also removed, since `set_all_extr` now accesses corner from
`input_ssh_m`. Motivation: preparation to access the corner of NetCDF
files from `get_1_outerm`.

Lionel GUEZ authored 9 months ago and GUEZ Lionel committed 8 months ago

This is a waste of computing time but we need to do this because we
are going to remove corner from the dummy arguments of `set_all_extr`.

Lionel GUEZ authored 9 months ago and GUEZ Lionel committed 8 months ago

Do not re-define `coord_proj` if it was read from a shapefile.

We want to be able to re-compute the cost, starting from the output of
a previous run of `cost_function.py`, when we have discarded the input
file of the previous run.

Segments is no longer a test with only isolated segments since commit
e1c18ad8.

Add tests `Cost_function_avg_date`, `Greece_date` and
`Eurec4A_cost_date`.

Add option `--avg_fix` in tests to reproduce previous behaviour.