ghislainv / riskmapjnr

📦🐍 Mapping deforestation risk following JNR methodology

👋 Hi, I'm Ghislain 😊

I am an ecologist at Cirad (https://www.cirad.fr), a French research center working with developing countries to tackle environmental and development issues. I am conducting research projects in community ecology 🌱 🌿 🌲 🌳 🍂, conservation biology 🌳 🔥 🐦 🦍 🌍, and applied statistics ✏️ 📏 💻 📊 📉. More info at https://ecology.ghislainv.fr.

riskmapjnr's Issues

don't manage to run the lib on a FCC derived from GEE

details

Hi @ghislainv,

I'm trying to run the riskmapjnr in a module. before building up the interface I would like to create a running notebook using some GEE based map for the forest cover change (namely TMF and hansen&al. This is the current stage of dev: https://github.com/12rambau/riskmapjnr/blob/main/riskmapjnr.ipynb

when I try to run it on SEPAL I get some small deprecation error (not very important at the moment) but I get a strange error when running rmj.makemap:

File ~/.local/lib/python3.8/site-packages/riskmapjnr/dist_edge_threshold.py:226, in dist_edge_threshold(fcc_file, defor_values, dist_file, dist_bins, tab_file_dist, fig_file_dist, figsize, dpi, blk_rows, verbose)
    222 res_df.to_csv(tab_file_dist, sep=",", header=True,
    223               index=False, index_label=False)
    225 # Distance and percentage for 99% threshold
--> 226 index_thresh = np.nonzero(res_df["perc"].values > 99)[0][0]
    227 dist_thresh = res_df.loc[index_thresh, "distance"]
    228 perc_thresh = np.around(res_df.loc[index_thresh, "perc"], 2)

IndexError: index 0 is out of bounds for axis 0 with size 0

Could you simply try to run my notebook from your side and let me know if you get the same error ? (I tried on your favorite AOI: guadeloupe). I think you just need to install sepal-ui.

Side note

I extracted the GADM interface from sepal-ui and created a small lib out of it. As you run many things out of GEE, you might be interested to get administrative geometries on the fly as a geodataframe:

https://pygadm.readthedocs.io/en/stable/

pool is never joined

when run in parralel, Pool's async-methods return objects which are conceptually "explicit futures", you need to call .get() to await and receive the actual result which is exactly what you did. On the other hand, before going any further in your code you also need to close and join the pool. Which is not done here:

riskmapjnr/riskmapjnr/makemap.py

Line 430 in 96d33ee

res = pool.starmap_async(makemap_ws, args).get()

A call to the method in a script will execute next command before all the runners have finished:

results_makemap = rmj.makemap(**kwargs)

# add the results in the textfields 
ws_hat = results_makemap["ws_hat"]
m_hat = results_makemap["m_hat"]
dist_thresh = results_makemap["dist_thresh"]

will return:

UnboundLocalError: local variable 'results_makemap' referenced before assignment

Is there other places where you used this method ? Happy to make a PR

note

It's invisible in notebooks as tornado is making sure every promise is closed before finishing cell computation.

makemap is shutting down the interactive backend when launched in SEPAL

still in the same example (https://github.com/12rambau/riskmapjnr/blob/main/riskmapjnr.ipynb) I get a weird behavior from matplotlib.

In the first cells everything goes as expected but after a call to makemap I get the following warning:

No display found. Using non-interactive Agg backend.

Which of course prevents any display of matplotlib results downstream. I assume it's due to the SEPAL environment but I'm curious where do you use matplotlib in the computation steps ?

Error version 1.2.1 defrate_per_cat

In function makemap_ws defrate_per_cat contains the key defor_values , while in the function does not exist see functions bellow:

makemap_ws :
defrate_per_cat(
fcc_file,
defor_values=1,
riskmap_file=riskmap_file,
time_interval=time_interval[0],
tab_file_defrate=tab_file_defrate,
blk_rows=blk_rows,
verbose=False)

defrate_per_cat(fcc_file, riskmap_file, time_interval,
period="calibration",
tab_file_defrate="defrate_per_cat.csv",
blk_rows=128, verbose=True):

riskmapjnr display issue

I have installed the riskmap package as per the procedure.
However, I am getting following message when setting a virtual environment

(conda-rmj) C:\Users\Asdfhg>riskmapjnr
no display found. Using non-interactive Agg backend

riskmapjnr: Map of deforestation risk following JNR methodology.

https://ecology.ghislainv.fr/riskmapjnr

riskmapjnr version 0.2.

The above message is not allowing the tool to function properly, is there any help that I can get on this particular issue. I am a intermediate user of python.

Thank you

are you considering publishing the releases on pipy ?

It's just out of curiosity to know what process to use for the installation on SEPAL.

AttributeError: module 'numpy' has no attribute 'int'

I am getting an error while running the makemap function of the package. Here are the details:

Model calibration and validation
.. Model 0: window size = 5, slicing method = ei.

AttributeError Traceback (most recent call last)
in <cell line: 1>()
----> 1 results_makemap = rmj.makemap(
2 fcc_file=output_raster_path,
3 time_interval=[3, 3],
4 output_dir=out_dir,
5 clean=False,

4 frames
/usr/local/lib/python3.10/dist-packages/riskmapjnr/makemap.py in makemap(fcc_file, time_interval, output_dir, clean, dist_bins, win_sizes, ncat, methods, csize, no_quantity_error, parallel, ncpu, figsize, dpi, blk_rows, verbose)
411 for i in range(n_ws):
412 s = win_sizes[i]
--> 413 ii, wRMSE_list, ncell, csize_km = makemap_ws(
414 i, s, fcc_file, time_interval,
415 dist_file, dist_v_file,

/usr/local/lib/python3.10/dist-packages/riskmapjnr/makemap.py in makemap_ws(i, win_size, fcc_file, time_interval, dist_file, dist_v_file, dist_edge_thresh, calval_dir, ncat, methods, meth, n_m, csize, nqe, figsize, dpi, blk_rows, verbose)
115 verbose=False)
116 # Validation
--> 117 val = validation(
118 fcc_file=fcc_file,
119 time_interval=time_interval[1],

/usr/local/lib/python3.10/dist-packages/riskmapjnr/validation.py in validation(fcc_file, time_interval, riskmap_file, tab_file_defrate, csize, no_quantity_error, tab_file_pred, fig_file_pred, figsize, dpi, verbose)
111
112 # Make square
--> 113 squareinfo = make_square(fcc_file, csize)
114 nsquare = squareinfo[0]
115 nsquare_x = squareinfo[1]

/usr/local/lib/python3.10/dist-packages/riskmapjnr/misc/miscellaneous.py in make_square(rasterfile, square_size)
145 nsquare = nsquare_x * nsquare_y
146 # Upper-left coordinates of each square
--> 147 x = np.arange(0, ncol, square_size, dtype=np.int).tolist()
148 y = np.arange(0, nrow, square_size, dtype=np.int).tolist()
149 # Size (number of col and row) of each square

/usr/local/lib/python3.10/dist-packages/numpy/init.py in getattr(attr)
317
318 if attr in former_attrs:
--> 319 raise AttributeError(former_attrs[attr])
320
321 if attr == 'testing':

AttributeError: module 'numpy' has no attribute 'int'.
np.int was a deprecated alias for the builtin int. To avoid this error in existing code, use int by itself. Doing this will not modify any behavior and is safe. When replacing np.int, you may wish to use e.g. np.int64 or np.int32 to specify the precision. If you wish to review your current use, check the release note link for additional information.
The aliases was originally deprecated in NumPy 1.20; for more details and guidance see the original release note at:
https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations