daniel-partington / hydromodelbuilder Goto Github PK

https://pypi.python.org/pypi/tqdm

Copy/pasted the Campaspe_transient_PEST.py contents into a notebook to kick off documentation writing.

flopyInterface is currently ~2744 lines long, and as a consequence is hard to read through.

I suggest the following changes:

1. Separate the classes defined in the file into separate class files.
2. There seems to be analysis/visualization helper methods defined within ModflowModel().
   Extract these out into a separate file and import them in as necessary
3. Remove commented out code
4. If certain methods only relate to a specific model implementation (e.g. SFR) then these could be split out into a separate class that inherits from a base class (or we could use composition)

Any other suggestions @mjasher?

@ConnectedSystems

When reproject_layer from reproject.py is called from GDALInterface, the new reprojected shapefile is not being written to copy_dest. This works when the module is run standalone but not when called. Need to sort out so that there is a record of transforms and to prevent recalling the reprojection every time GWModelBuilder class object is being used, i.e. every time a build script is run for a groundwater model.

Note that the function reproject_layer is still returning a perfectly usable GDAL object that works as expected in mapping the shapefile to the model grid, it is just not keeping the record of the reprojection as is desired.

This save_array() function is pretty much identical to the one found in pilotpoints

@ConnectedSystems , in text_writer.py I think we need to convert any list objects to str in the "join" calls. In PESTInterface, the file writing was done using objects of type float, int as well as str.

Expanding on issue #8 there are multiple places where identical or near identical methods are being defined which handle file/data reading and writing.

e.g.
These are identical to each other

It would be great if we could consolidate these.

Hunt down places where file reading/writing occurs, for example here, and replace with consolidated functions

Currently it is a class which, based on its usage in flopyInterface, will get recreated every time Calculate_Rn_from_SFR_with_simple_model() gets called.

Based on this behaviour it would be better for Radon_EC_Simple to get refactored into a single function

Certain methods that appear in ModelProperties.py (for e.g., here) and PESTInterface appear to have overlapping intent/functionalities

Could we consolidate these?

As part of the given spec, a step is to "adjust the height of an inactive cell".
In the context where the current raster is the third one (i.e. Raster 3), the "bottom of this inactive cell should move to become half the height between the bottom of Raster 4 and Raster 1".

I cannot see where this happens in the current version of the function, although

proposed_heights = (top_merge_locs - (next_j * (merge_diff / next_j)))

Looks like it's calculating the height adjustments. For full context (see here)

Additionally, the function as it is currently doesn't appear to check if the new heights are valid (i.e. greater than the minimum thickness value - or should that be greater than or equal to?)

"In the case of it being two overlying inactive cells then it should split them up into three equal parts"

I cannot determine where this happens currently.

Ancillary simplification suggestions:
reclassIsolatedCells() is like a median filter with an extra step that reclassifies cells with value of -1 if any neighbouring cell is -1, where -1 is the value that indicates an invalid cell.

Something like the below can be adapted for this purpose, which would drastically reduce the amount of code. Here, it only does the median filtering (replacement of isolated cells with surrounding values) so it needs to be modified to do the invalid neighbour check.

def filter_isolated_cells(arr):
    """
    Adapted from:
    * https://stackoverflow.com/questions/41550979/fill-holes-with-majority-of-surrounding-values-python
    * https://docs.scipy.org/doc/scipy-0.16.1/reference/generated/scipy.ndimage.filters.median_filter.html
    """
    labels, count = label(arr)
    return ndimage.median_filter(labels, 1)
# End filter_isolated_cells()