Merge branch 'develop' into feature/tc_model_kwargs

.github/workflows/ci.yml

@@ -26,7 +26,7 @@ jobs:
     steps:
       -
         name: Checkout Repo
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
       -
         # Store the current date to use it as cache key for the environment
         name: Get current date
@@ -64,7 +64,7 @@ jobs:
       -
         name: Upload Coverage Reports
         if: always()
-        uses: actions/upload-artifact@v3
+        uses: actions/upload-artifact@v4
         with:
           name: coverage-report-unittests-py${{ matrix.python-version }}
           path: coverage/

AUTHORS.md

@@ -31,3 +31,4 @@
 * Leonie Villiger
 * Kam Lam Yeung
 * Sarah Hülsen
+* Timo Schmid

CHANGELOG.md

@@ -12,8 +12,53 @@ Code freeze date: YYYY-MM-DD
 
 ### Added
 
+### Changed
+
+### Fixed
+
+- Fix `util.coordinates.latlon_bounds` for cases where the specified buffer is very large so that the bounds cover more than the full longitudinal range `[-180, 180]` [#839](https://github.com/CLIMADA-project/climada_python/pull/839)
+- Fix `climada.hazard.trop_cyclone` for TC tracks crossing the antimeridian [#839](https://github.com/CLIMADA-project/climada_python/pull/839)
+
+### Deprecated
+
+### Removed
+
+## 4.1.0
+
+Release date: 2024-02-14
+
+### Dependency Changes
+
+Added:
+
+- `pyproj` >=3.5
+- `numexpr` >=2.9
+
+Updated:
+
+- `contextily` >=1.3 → >=1.5
+- `dask` >=2023 → >=2024
+- `numba` >=0.57 → >=0.59
+- `pandas` >=2.1 &rarr; >=2.1,<2.2
+- `pint` >=0.22 &rarr; >=0.23
+- `scikit-learn` >=1.3 &rarr; >=1.4
+- `scipy` >=1.11 &rarr; >=1.12
+- `sparse` >=0.14 &rarr; >=0.15
+- `xarray` >=2023.8 &rarr; >=2024.1
+- `overpy` =0.6 &rarr; =0.7
+- `peewee` =3.16.3 &rarr; =3.17.1
+
+Removed:
+
+- `proj` (in favor of `pyproj`)
+
+### Added
+
 - Convenience method `api_client.Client.get_dataset_file`, combining `get_dataset_info` and `download_dataset`, returning a single file objet. [#821](https://github.com/CLIMADA-project/climada_python/pull/821)
 - Read and Write methods to and from csv files for the `DiscRates` class. [#818](ttps://github.com/CLIMADA-project/climada_python/pull/818)
+- Add `CalcDeltaClimate` to unsequa module to allow uncertainty and sensitivity analysis of impact change calculations [#844](https://github.com/CLIMADA-project/climada_python/pull/844)
+- Add function `safe_divide` in util which handles division by zero and NaN values in the numerator or denominator [#844](https://github.com/CLIMADA-project/climada_python/pull/844)
+- Add reset_frequency option for the impact.select() function. [#847](https://github.com/CLIMADA-project/climada_python/pull/847)
 
 ### Changed
 
@@ -24,6 +69,7 @@ Code freeze date: YYYY-MM-DD
 - Recommend using Mamba instead of Conda for installing CLIMADA [#809](https://github.com/CLIMADA-project/climada_python/pull/809)
 - `Hazard.from_xarray_raster` now allows arbitrary values as 'event' coordinates [#837](https://github.com/CLIMADA-project/climada_python/pull/837)
 - `climada.test.get_test_file` now compares the version of the requested test dataset with the version of climada itself and selects the most appropriate dataset. In this way a test file can be updated without the need of changing the code of the unittest. [#822](https://github.com/CLIMADA-project/climada_python/pull/822)
+- Explicitly require `pyproj` instead of `proj` (the latter is now implicitly required) [#845](https://github.com/CLIMADA-project/climada_python/pull/845)
 
 ### Fixed
 
@@ -32,10 +78,8 @@ Code freeze date: YYYY-MM-DD
 - `climada.util.yearsets.sample_from_poisson`: fix a bug ([#819](https://github.com/CLIMADA-project/climada_python/issues/819)) and inconsistency that occurs when lambda events per year (`lam`) are set to 1. [[#823](https://github.com/CLIMADA-project/climada_python/pull/823)]
 - In the TropCyclone class in the Holland model 2008 and 2010 implementation, a doublecounting of translational velocity is removed [#833](https://github.com/CLIMADA-project/climada_python/pull/833)
 - `climada.util.test.test_finance` and `climada.test.test_engine` updated to recent input data from worldbank [#841](https://github.com/CLIMADA-project/climada_python/pull/841)
-
-### Deprecated
-
-### Removed
+- Set `nodefaults` in Conda environment specs because `defaults` are not compatible with conda-forge [#845](https://github.com/CLIMADA-project/climada_python/pull/845)
+- Avoid redundant calls to `np.unique` in `Impact.impact_at_reg` [#848](https://github.com/CLIMADA-project/climada_python/pull/848)
 
 ## 4.0.1
 

climada/_version.py

@@ -1 +1 @@
-__version__ = '4.0.2-dev'
+__version__ = '4.1.1-dev'

climada/engine/impact.py

@@ -451,12 +451,12 @@ def impact_at_reg(self, agg_regions=None):
         at_reg_event = np.hstack(
             [
                 self.imp_mat[:, np.where(agg_regions == reg)[0]].sum(1)
-                for reg in np.unique(agg_reg_unique)
+                for reg in agg_reg_unique
             ]
         )
 
         at_reg_event = pd.DataFrame(
-            at_reg_event, columns=np.unique(agg_reg_unique), index=self.event_id
+            at_reg_event, columns=agg_reg_unique, index=self.event_id
         )
 
         return at_reg_event
@@ -1475,9 +1475,14 @@ def _cen_return_imp(imp, freq, imp_th, return_periods):
 
         return imp_fit
 
-    def select(self,
-               event_ids=None, event_names=None, dates=None,
-               coord_exp=None):
+    def select(
+        self,
+        event_ids=None,
+        event_names=None,
+        dates=None,
+        coord_exp=None,
+        reset_frequency=False
+    ):
         """
         Select a subset of events and/or exposure points from the impact.
         If multiple input variables are not None, it returns all the impacts
@@ -1509,6 +1514,9 @@ def select(self,
         coord_exp : np.array, optional
             Selection of exposures coordinates [lat, lon] (in degrees)
             The default is None.
+        reset_frequency : bool, optional
+            Change frequency of events proportional to difference between first and last
+            year (old and new). Assumes annual frequency values. Default: False.
 
         Raises
         ------
@@ -1580,6 +1588,19 @@ def select(self,
             LOGGER.info("The total value cannot be re-computed for a "
                         "subset of exposures and is set to None.")
 
+        # reset frequency if date span has changed (optional):
+        if reset_frequency:
+            if self.frequency_unit not in ['1/year', 'annual', '1/y', '1/a']:
+                LOGGER.warning("Resetting the frequency is based on the calendar year of given"
+                    " dates but the frequency unit here is %s. Consider setting the frequency"
+                    " manually for the selection or changing the frequency unit to %s.",
+                    self.frequency_unit, DEF_FREQ_UNIT)
+            year_span_old = np.abs(dt.datetime.fromordinal(self.date.max()).year -
+                                   dt.datetime.fromordinal(self.date.min()).year) + 1
+            year_span_new = np.abs(dt.datetime.fromordinal(imp.date.max()).year -
+                                   dt.datetime.fromordinal(imp.date.min()).year) + 1
+            imp.frequency = imp.frequency * year_span_old / year_span_new
+
         # cast frequency vector into 2d array for sparse matrix multiplication
         freq_mat = imp.frequency.reshape(len(imp.frequency), 1)
         # .A1 reduce 1d matrix to 1d array

climada/engine/test/test_impact.py

@@ -27,6 +27,7 @@
 import h5py
 from pyproj import CRS
 from rasterio.crs import CRS as rCRS
+import datetime as dt
 
 from climada.entity.entity_def import Entity
 from climada.hazard.base import Hazard
@@ -67,6 +68,24 @@ def dummy_impact():
         haz_type="TC",
     )
 
+def dummy_impact_yearly():
+    """Return an impact containing events in multiple years"""
+    imp = dummy_impact()
+
+    years = np.arange(2010,2010+len(imp.date))
+
+    # Edit the date and frequency
+    imp.date = np.array([dt.date(year,1,1).toordinal() for year in years])
+    imp.frequency_unit = "1/year"
+    imp.frequency = np.ones(len(years))/len(years)
+
+    # Calculate the correct expected annual impact
+    freq_mat = imp.frequency.reshape(len(imp.frequency), 1)
+    imp.eai_exp = imp.imp_mat.multiply(freq_mat).sum(axis=0).A1
+    imp.aai_agg = imp.eai_exp.sum()
+
+    return imp
+
 
 class TestImpact(unittest.TestCase):
     """"Test initialization and more"""
@@ -868,6 +887,22 @@ def test_select_imp_map_fail(self):
         with self.assertRaises(ValueError):
             imp.select(event_ids=[0], event_names=[1, 'two'], dates=(0, 2))
 
+    def test_select_reset_frequency(self):
+        """Test that reset_frequency option works correctly"""
+
+        imp = dummy_impact_yearly() # 6 events, 1 per year
+
+        # select first 4 events
+        n_yr = 4
+        sel_imp = imp.select(dates=(imp.date[0],imp.date[n_yr-1]), reset_frequency=True)
+
+        # check frequency-related attributes
+        np.testing.assert_array_equal(sel_imp.frequency, [1/n_yr]*n_yr)
+        self.assertEqual(sel_imp.aai_agg,imp.at_event[0:n_yr].sum()/n_yr)
+        np.testing.assert_array_equal(sel_imp.eai_exp,
+                                      imp.imp_mat[0:n_yr,:].todense().sum(axis=0).A1/n_yr)
+
+
 class TestConvertExp(unittest.TestCase):
     def test__build_exp(self):
         """Test that an impact set can be converted to an exposure"""

climada/engine/unsequa/__init__.py

@@ -22,3 +22,4 @@
 from .calc_base import *
 from .calc_impact import *
 from .calc_cost_benefit import *
+from .calc_delta_climate import *

climada/engine/unsequa/calc_base.py

@@ -85,7 +85,6 @@ def __init__(self):
         """
         Empty constructor to be overwritten by subclasses
         """
-        pass
 
     def check_distr(self):
         """
@@ -118,7 +117,6 @@ def check_distr(self):
                 distr_dict[input_param_name] = input_param_func
         return True
 
-
     @property
     def input_vars(self):
         """
@@ -179,7 +177,6 @@ def est_comp_time(self, n_samples, time_one_run, processes=None):
                 "been assigned to exp before defining input_var, ..."
                 "\n If computation cannot be reduced, consider using"
                 " a surrogate model https://www.uqlab.com/", time_one_run)
-
         total_time = n_samples * time_one_run / processes
         LOGGER.info("\n\nEstimated computaion time: %s\n",
                     dt.timedelta(seconds=total_time))
@@ -323,21 +320,23 @@ def sensitivity(self, unc_output, sensitivity_method = 'sobol',
 
         Parameters
         ----------
-        unc_output : climada.engine.uncertainty.unc_output.UncOutput
+        unc_output : climada.engine.unsequa.UncOutput
             Uncertainty data object in which to store the sensitivity indices
         sensitivity_method : str, optional
-            Sensitivity analysis method from SALib.analyse. Possible choices: 'fast', 'rbd_fact',
-            'morris', 'sobol', 'delta', 'ff'. Note that in Salib, sampling methods and sensitivity
-            analysis methods should be used in specific pairs:
+            sensitivity analysis method from SALib.analyse
+            Possible choices:
+                'fast', 'rbd_fact', 'morris', 'sobol', 'delta', 'ff'
+            The default is 'sobol'.
+            Note that in Salib, sampling methods and sensitivity analysis
+            methods should be used in specific pairs.
             https://salib.readthedocs.io/en/latest/api.html
-            Default: 'sobol'
         sensitivity_kwargs: dict, optional
             Keyword arguments of the chosen SALib analyse method.
             The default is to use SALib's default arguments.
 
         Returns
         -------
-        sens_output : climada.engine.uncertainty.unc_output.UncOutput()
+        sens_output : climada.engine.unsequa.UncOutput
             Uncertainty data object with all the sensitivity indices,
             and all the uncertainty data copied over from unc_output.
 
@@ -379,6 +378,7 @@ def sensitivity(self, unc_output, sensitivity_method = 'sobol',
 
         return sens_output
 
+
 def _multiprocess_chunksize(samples_df, processes):
     """Divides the samples into chunks for multiprocesses computing
 
@@ -405,6 +405,7 @@ def _multiprocess_chunksize(samples_df, processes):
         samples_df.shape[0] / processes
         ).astype(int)
 
+
 def _transpose_chunked_data(metrics):
     """Transposes the output metrics lists from one list per
     chunk of samples to one list per output metric
@@ -434,6 +435,7 @@ def _transpose_chunked_data(metrics):
         for x in zip(*metrics)
         ]
 
+
 def _sample_parallel_iterator(samples, chunksize, **kwargs):
     """
     Make iterator over chunks of samples

climada/engine/unsequa/calc_cost_benefit.py

@@ -68,7 +68,9 @@ class CalcCostBenefit(Calc):
         ('haz_input_var', 'ent_input_var', 'haz_fut_input_var', 'ent_fut_input_var')
     _metric_names : tuple(str)
         Names of the cost benefit output metrics
-        ('tot_climate_risk', 'benefit', 'cost_ben_ratio', 'imp_meas_present', 'imp_meas_future')
+        ('tot_climate_risk', 'benefit', 'cost_ben_ratio',
+         'imp_meas_present', 'imp_meas_future')
+
     """
 
     _input_var_names = (