Automated Orthogonal Grid Generation for Regional Ocean Modeling with Schwarz-Chrisotffel Conformal Mappings
You need to specify the path of the file which have the boundary coordinate of target(csv) in ./model and the shape of the grid(like 50x50). After that, run the command:
python main.py --boundary_name [FILENAME] --shape1 [NUM1] --shape2 [NUM2]
Example:
python main.py --boundary_name boundary_bh_0.5d.csv --shape1 50 --shape2 50
The image and .nc file can be found in ./result/res.
Here is an illustration of the Bohai Sea(China):
you also can download other ETOPO1(bedrock) data from Grid Extract. For example, you can choose Bohai Sea(China) region and download its NetCDF file. Put it into ./shp and then run python get_boundary_bh.py to obtain a new .csv file for main.py.
if you want to make a new classification model, you can download the geographic boundary data from GADM which are the natural and enough simple-polygon data. Put .shp and .dbf files we needed in ./shp and run the command:
python train_data_process.py --node_num [NUMBER OF VERTEX]
Example:
python train_data_process.py --node_num 10
after that we can get 10-polygon data in ./shp/data_rough. run the following command and we can get the train data for the base classificaiton model in ./shp/train_data(the number of vertex is 10):
python train_data_get.py
In ./model we have finished the base model named lgb_model_10corenr_9adj.pkl by train_lightgbm.py(need to modify some program statements). In order to make a new training data for powerful classificaiton model, e.g., we want to make a new model for 15-polygon with 10-polygon data. First, run python train_data_process.py --node_num 15 and then run:
python make_new_classification.py --modelname lgb_model_10corenr_9adj.pkl --adj_num 9 --From 10 --To 15
actually,the effect of this command is similar to train_data_get.py. Last but not least, we can obtain a new calssification model through the command below.
python train_lightgbm.py --adj_num 15 --From 10 --To 15