diff --git a/mlair/data_handler/abstract_data_handler.py b/mlair/data_handler/abstract_data_handler.py
index 36d6e9ae5394705af4b9fbcfd1d8ff77572642b5..a6f49d2d756200a8a4e5f15c13ecf385ef08dabc 100644
--- a/mlair/data_handler/abstract_data_handler.py
+++ b/mlair/data_handler/abstract_data_handler.py
@@ -3,8 +3,12 @@ __author__ = 'Lukas Leufen'
__date__ = '2020-09-21'
import inspect
+import logging
from typing import Union, Dict
+import dask
+from dask.diagnostics import ProgressBar
+
from mlair.helpers import remove_items
@@ -84,3 +88,13 @@ class AbstractDataHandler:
def _hash_list(self):
return []
+
+ @staticmethod
+ def _force_dask_computation(data):
+ try:
+ with ProgressBar():
+ logging.info(f"DefaultDH: _force_dask_computation")
+ data = dask.compute(data)[0]
+ except:
+ logging.info("can't execute dask.compute")
+ return data
diff --git a/mlair/data_handler/data_handler_wrf_chem.py b/mlair/data_handler/data_handler_wrf_chem.py
index 4aedd310145614ac4276f0895d166d3c1a7d4a42..158a7fb2bb8f5d3674b9b312f47a389dc0c5e90f 100644
--- a/mlair/data_handler/data_handler_wrf_chem.py
+++ b/mlair/data_handler/data_handler_wrf_chem.py
@@ -612,6 +612,7 @@ class DataHandlerSingleGridColumn(DataHandlerSingleStation):
date_format_of_nc_file=None,
as_image_like_data_format=True,
time_zone=None,
+ input_output_sampling4toarstats : tuple = None,
**kwargs):
self.external_coords_file = external_coords_file
self.var_logical_z_coord_selector = self._return_z_coord_select_if_valid(var_logical_z_coord_selector,
@@ -627,7 +628,7 @@ class DataHandlerSingleGridColumn(DataHandlerSingleStation):
self.as_image_like_data_format = as_image_like_data_format
self.time_zone = time_zone
- self.tzwhere = tzwhere.tzwhere()
+ self.input_output_sampling4toarstats = input_output_sampling4toarstats
super().__init__(*args, **kwargs)
@staticmethod
@@ -742,19 +743,23 @@ class DataHandlerSingleGridColumn(DataHandlerSingleStation):
self.loader.physical_y_coord_name: self.loader.get_coordinates()['lat'].tolist()}
meta = pd.DataFrame(_meta, index=station)
- if isinstance(self.sampling, tuple) and len(self.sampling) == 2:
+ if isinstance(self.input_output_sampling4toarstats, tuple) and len(self.input_output_sampling4toarstats) == 2:
if self.var_logical_z_coord_selector != 0:
raise NotImplementedError(
f"Method `apply_toarstats` is not implemented for var_logical_z_coord_selector != 0: "
f"Is {self.var_logical_z_coord_selector}")
- data = self.apply_toarstats(data, meta, self.sampling[1])
+ data = self.apply_toarstats(data, meta, self.input_output_sampling4toarstats[1])
data = self._slice_prep(data, start=start, end=end)
return data, meta
+ @TimeTrackingWrapper
def apply_toarstats(self, data, meta, target_sampling="daily"):
- local_time_zone = self.tzwhere.tzNameAt(latitude=meta[self.loader.physical_y_coord_name],
- longitude=meta[self.loader.physical_x_coord_name])
+ tz_where = tzwhere.tzwhere()
+ local_time_zone = tz_where.tzNameAt(latitude=meta[self.loader.physical_y_coord_name],
+ longitude=meta[self.loader.physical_x_coord_name])
+ # local_time_zone = self.tzwhere.tzNameAt(latitude=meta[self.loader.physical_y_coord_name],
+ # longitude=meta[self.loader.physical_x_coord_name])
hdata = data.squeeze().to_pandas()
hdata.index = self.set_time_zone(hdata.index)
hdata.index = hdata.index.tz_convert(local_time_zone)
@@ -774,7 +779,7 @@ class DataHandlerSingleGridColumn(DataHandlerSingleStation):
for coord in missing_squeezed_coords:
sampling_data.coords[coord] = data.coords[coord]
- return self. sampling_data
+ return self._force_dask_computation(sampling_data)
def set_time_zone(self, time_index):
"""
diff --git a/mlair/data_handler/default_data_handler.py b/mlair/data_handler/default_data_handler.py
index 188283703cb55481c6e21524077663f8ae2a9f17..1b4715a05d7e072add90137b89f7944299f449f5 100644
--- a/mlair/data_handler/default_data_handler.py
+++ b/mlair/data_handler/default_data_handler.py
@@ -7,15 +7,12 @@ import inspect
import gc
import logging
import os
-import pickle
import dill
import shutil
from functools import reduce
from typing import Tuple, Union, List
import multiprocessing
import psutil
-import dask
-from dask.diagnostics import ProgressBar
import numpy as np
import xarray as xr
@@ -111,16 +108,6 @@ class DefaultDataHandler(AbstractDataHandler):
attr_dict[attr] = val
return attr_dict
- @staticmethod
- def _force_dask_computation(data):
- try:
- with ProgressBar():
- logging.info(f"DefaultDH: _force_dask_computation")
- data = dask.compute(data)[0]
- except:
- logging.info("can't execute dask.compute")
- return data
-
def _load(self):
try:
with open(self._save_file, "rb") as f:
diff --git a/mlair/model_modules/model_class.py b/mlair/model_modules/model_class.py
index 8f464b16cbc74a241f70ba1c5bb9f8fe611dba95..fd43674e009631f4fd7f1c5648f500a0eaff2ec1 100644
--- a/mlair/model_modules/model_class.py
+++ b/mlair/model_modules/model_class.py
@@ -518,7 +518,7 @@ class MyLuongAttentionLSTMModel(AbstractModelClass):
self.initial_lr = 0.01
self.clipnorm = 1
- self.n_hidden = 100
+ self.n_hidden = 32
# apply to model
self.set_model()
diff --git a/run_wrf_dh.py b/run_wrf_dh.py
index a406f19346636dadaad22b07ead1daca2df3d539..bd5aa802299b247ee8341868162148a5d6dca034 100644
--- a/run_wrf_dh.py
+++ b/run_wrf_dh.py
@@ -8,7 +8,7 @@ from mlair.workflows import DefaultWorkflow
from mlair.helpers import remove_items
from mlair.configuration.defaults import DEFAULT_PLOT_LIST
-from mlair.model_modules.model_class import MyLSTMModel, MyCNNModel
+from mlair.model_modules.model_class import MyLSTMModel, MyCNNModel, MyLuongAttentionLSTMModel
import os
@@ -42,7 +42,7 @@ def main(parser_args):
# data_handler=DataHandlerSingleStation,
# data_handler=DataHandlerSingleGridColumn,
upsampeling=False,
- epochs=20,
+ epochs=200,
window_lead_time=4,
window_history_size=6,
# ('Germany', (5.98865807458, 47.3024876979, 15.0169958839, 54.983104153))
@@ -90,7 +90,7 @@ def main(parser_args):
'no2': {"method": "standardise"},
'co': {"method": "standardise"},
'PSFC': {"method": "standardise"},
- 'CLDFRA': {"method": "min_max"},
+ 'CLDFRA': {"method": "min_max","min": 0., "max": 1.},
},
variables=['T2', 'o3', 'wdir10ll', 'wspd10ll', 'no', 'no2', 'co', 'PSFC', 'PBLH', 'CLDFRA'],
target_var='o3',
@@ -120,6 +120,8 @@ def main(parser_args):
# train_end='2009-01-02',
# train_start='2009-01-01',
#train_end='2009-01-15',
+ # train_end='2009-02-15',
+ # train_end="2009-01-31",
train_end='2009-02-28',
# val_start='2009-01-02',
@@ -130,26 +132,37 @@ def main(parser_args):
###################################
val_start='2009-03-01',
val_end='2009-03-14',
+ ###################################
+ # val_start='2009-02-15',
+ # val_end='2009-03-02',
+ ###################################
+ # val_start='2009-02-01',
+ # val_end='2009-02-28',
# test_start='2009-01-03',
# test_end='2009-01-04',
###################################
- #test_start='2009-01-22',
- #test_end='2009-01-31',
+ # test_start='2009-01-22',
+ # test_end='2009-01-31',
###################################
test_start='2009-03-15',
test_end='2009-03-31',
+ ###################################
+ # test_start='2009-03-02',
+ # test_end='2009-03-31',
# sampling='hourly',
- sampling=("hourly", "daily"),
+ sampling="daily",
+ input_output_sampling4toarstats=("hourly", "daily"),
interpolation_limit=0,
- # as_image_like_data_format=False,
+ as_image_like_data_format=False,
# model=MyLSTMModel,
+ model=MyLuongAttentionLSTMModel,
use_multiprocessing=True,
# as_image_like_data_format=True,
# model=MyLSTMModel,
- model=MyCNNModel,
+ # model=MyCNNModel,
**parser_args.__dict__)
workflow.run()