Update dependencies and clear out deprecation warnings.

.github/workflows/tests.yml

@@ -5,7 +5,14 @@ on: [pull_request, push]
 jobs:
   tests:
 
-    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false  # don't cancel other matrix jobs when one fails
+      matrix:
+        python-version: ["3.10", "3.11", "3.12"]
+        os: [ubuntu-latest, macos-latest, windows-latest]
+
+
+    runs-on: ${{ matrix.os }}
 
     steps:
     - name: Checkout
@@ -14,7 +21,7 @@ jobs:
     - name: Set up Python
       uses: actions/setup-python@v5
       with:
-        python-version: "3.x"
+        python-version: ${{ matrix.python-version }}
 
     - name: Install dependencies
       run: |

Changelog.md

@@ -69,4 +69,7 @@ site itself.
 - Added matplotlib dependency
 - Move demos back to the parent directory using nblink
 # Version 0.4.6 - Sept 13, 2024
-- Fix license name in cfg
+- Fix license name in cfg
+# Version 0.4.7 - Sept 14, 2024
+- Updating dependencies  and test targets
+- Fix deprecation warnings raised by numpy

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = solarspatialtools
-version = 0.4.6
+version = 0.4.7
 author = 'Joe Ranalli'
 author_email = [email protected]
 description = A package for spatial analysis of solar energy and solar irradiance
@@ -19,14 +19,14 @@ classifiers =
 
 [options]
 install_requires =
-    numpy<2
-    pandas<=2.2.2
-    tables<=3.9.2
-    pyproj<=3.6.1
-    pvlib<=0.11.0
-    netcdf4<1.7.2
-    scipy<=1.14.0
-    matplotlib<=3.9.2
+    numpy
+    pandas
+    tables
+    pyproj
+    pvlib
+    netcdf4
+    scipy
+    matplotlib
 
 [options.extras_require]
 tests =
@@ -36,8 +36,8 @@ demos =
     jupyter
 
 docs =
-    sphinx
-    sphinx_rtd_theme
-    nbsphinx
-    ipython
-    nbsphinx-link
+    sphinx==7.4.7
+    sphinx_rtd_theme==2.0.0
+    nbsphinx==0.9.5
+    ipython==8.27.0
+    nbsphinx-link==1.3.0

src/solarspatialtools/signalproc.py

@@ -462,7 +462,7 @@ def _residual(dels, freqs, phases, mask):
 
     if method == 'diff':
         if not np.any(np.array(np.shape(tf)) == 1):
-            raise ValueError('tf must be a 1D array for method: fit')
+            raise ValueError('tf must be a 1D array for method: diff')
 
         # # Method 1: unwrap phase and take derivative
         tfv = tf.values.flatten()
@@ -479,7 +479,7 @@ def _residual(dels, freqs, phases, mask):
         try:
             return (curve_fit(_delay_fitter,
                               np.expand_dims(tf.index, axis=1)[ix],
-                              np.unwrap(np.angle(tf.loc[ix]).flatten()))[0],
+                              np.unwrap(np.angle(tf.loc[ix]).flatten()))[0].item(),
                     ix)
         except ValueError:
             from warnings import warn
@@ -653,7 +653,7 @@ def compute_delays(ts_in, ts_out, mode='loop', scaling='coeff'):
 
             extras['peak_corr'][i] = xcorr[peak_lag_index]
             extras['mean_corr'][i] = np.mean(xcorr)
-            extras['zero_corr'][i] = xcorr[zero_lag_ind]
+            extras['zero_corr'][i] = xcorr[zero_lag_ind].item()
 
         # Scale
         extras['peak_corr'] *= corr_scale

src/solarspatialtools/spatial.py

@@ -254,6 +254,9 @@ def magnitude(vec):
     magnitude : numeric
         the magnitude of the vector
     """
+    # respond differently for Pandas Types
+    if isinstance(vec, pd.Series):
+        return np.sqrt(vec.iloc[0]**2 + vec.iloc[1]**2)
     return np.sqrt(vec[0]**2+vec[1]**2)
 
 

src/solarspatialtools/stats.py

@@ -383,7 +383,10 @@ def calc_quantile(timeseries, n_days="30d", quantile=0.9):
                                    p90.index.astype(str))
 
         # Concat this day onto the output object
-        out_df = pd.concat((out_df, p90), axis=0)
+        if len(out_df) == 0:
+            out_df = p90
+        else:
+            out_df = pd.concat((out_df, p90), axis=0)
 
     # localize the timeseries and the column names back to the input
     out_df = out_df.tz_localize(ts.index.tz).reindex(ts.index)

tests/test_cmv.py

@@ -76,7 +76,7 @@ def test_cmv_artificial(theta_deg, velocity, mode):
     signals[4, :] = x
     for i, delay in enumerate(delays):
         signals[i, :] = np.roll(x, int(delay*fs))
-    df = pd.DataFrame(signals.T, index=pd.TimedeltaIndex(t, 's'))
+    df = pd.DataFrame(signals.T, index=pd.to_timedelta(t, 's'))
 
     # Compute the CMV
     cld_spd, cld_dir, dat = cmv.compute_cmv(df, pos_utm,

tests/test_field.py

@@ -130,7 +130,7 @@ def test_compute_delays(delay, delay_method):
     y3 = np.roll(x, int(3*delay * fs))
     y4 = np.roll(x, int(4*delay * fs))
 
-    df = pd.DataFrame(np.array([x,y1,y2,y3,y4]).T, index=pd.TimedeltaIndex(t, 's'), columns=['x1','x2','x3','x4','x5'])
+    df = pd.DataFrame(np.array([x,y1,y2,y3,y4]).T, index=pd.to_timedelta(t, 's'), columns=['x1','x2','x3','x4','x5'])
     ref = 'x1'
 
     delays, coh = field.compute_delays(df, ref, navgs=5, coh_limit=0.6, freq_limit=1, method=delay_method)
@@ -154,7 +154,7 @@ def test_compute_delays_nan(delay_method):
     y3 = np.roll(x, int(3*delay * fs))
     y4 = np.nan * np.zeros_like(y3)
 
-    df = pd.DataFrame(np.array([x,y1,y2,y3,y4]).T, index=pd.TimedeltaIndex(t, 's'), columns=['x1','x2','x3','x4','x5'])
+    df = pd.DataFrame(np.array([x,y1,y2,y3,y4]).T, index=pd.to_timedelta(t, 's'), columns=['x1','x2','x3','x4','x5'])
     ref = 'x1'
 
     delays, coh = field.compute_delays(df, ref, navgs=5, coh_limit=0.6, freq_limit=1, method=delay_method)

tests/test_irradiance.py

@@ -34,7 +34,7 @@ def test_clearsky_index():
     assert_allclose(out, expected, atol=0.001)
 
     # GHI series & CS series
-    times = pd.date_range(start='20180601', periods=2, freq='12H')
+    times = pd.date_range(start='20180601', periods=2, freq='12h')
     ghi_measured = pd.Series([100,  500], index=times)
     ghi_modeled = pd.Series([500, 1000], index=times)
     out = clearsky_index(ghi_measured, ghi_modeled)
@@ -44,7 +44,7 @@ def test_clearsky_index():
     assert out.values == approx(expected.values)
 
     # GHI 2D frame & CS 1D series
-    times = pd.date_range(start='20180601', periods=3, freq='12H')
+    times = pd.date_range(start='20180601', periods=3, freq='12h')
     ghi_measured = pd.DataFrame([[100,  100], [200, 200], [500, 500]], index=times)
     ghi_modeled = pd.Series([500, 800, 1000], index=times)
     out = clearsky_index(ghi_measured, ghi_modeled)

tests/test_signalproc.py

@@ -77,7 +77,7 @@ def test_averaged_psd():
     T = 5.0    # seconds
     t = np.linspace(0, T, int(T*fs), endpoint=False)  # time variable
     x = 0.5*np.sin(2*np.pi*2*t)  # +0.1*np.sin(2*np.pi*10*t)
-    input_tsig = pd.Series(x, index=pd.TimedeltaIndex(t, 's'))
+    input_tsig = pd.Series(x, index=pd.to_timedelta(t, 's'))
 
     # Calculate the averaged PSD using the function
     navgs = 5
@@ -103,8 +103,8 @@ def test_averaged_tf():
     t = np.linspace(0, T, int(T*fs), endpoint=False)  # time variable
     x = 0.5*np.sin(2*np.pi*2*t)  # input signal
     y = 0.5*np.sin(2*np.pi*2*t + np.pi/4)  # output signal with phase shift
-    x_tsig = pd.Series(x, index=pd.TimedeltaIndex(t, 's'))
-    y_tsig = pd.Series(y, index=pd.TimedeltaIndex(t, 's'))
+    x_tsig = pd.Series(x, index=pd.to_timedelta(t, 's'))
+    y_tsig = pd.Series(y, index=pd.to_timedelta(t, 's'))
 
     # Calculate the averaged transfer function using the function
     navgs = 5
@@ -168,8 +168,8 @@ def test_tf_delay(delay):
     noise = np.random.random(len(t))/5  # add broadband noise
     x = 0.5*np.sin(2*np.pi*2*t) + noise  # noisy signal
     y = np.roll(x, int(delay*fs))  # Explicitly delay the signal
-    x_tsig = pd.Series(x, index=pd.TimedeltaIndex(t, 's'))
-    y_tsig = pd.Series(y, index=pd.TimedeltaIndex(t, 's'))
+    x_tsig = pd.Series(x, index=pd.to_timedelta(t, 's'))
+    y_tsig = pd.Series(y, index=pd.to_timedelta(t, 's'))
 
     # Calculate the averaged transfer function using the function
     navgs = 5
@@ -199,8 +199,8 @@ def test_tf_delay_multi(delay):
     y2 = np.roll(x, int(2 * delay * fs))
     y3 = np.roll(x, int(3 * delay * fs))
     y4 = np.roll(x, int(4 * delay * fs))
-    x_tsig = pd.Series(x, index=pd.TimedeltaIndex(t, 's'))
-    ysigs = [pd.Series(y, index=pd.TimedeltaIndex(t, 's')) for y in
+    x_tsig = pd.Series(x, index=pd.to_timedelta(t, 's'))
+    ysigs = [pd.Series(y, index=pd.to_timedelta(t, 's')) for y in
              [y1, y2, y3, y4]]
     ysigs_df = pd.DataFrame(np.array(ysigs).T, columns=[0, 1, 2, 3],
                             index=ysigs[0].index)
@@ -235,8 +235,8 @@ def test_tf_delay_nan():
     y3 = np.roll(x, int(3 * delay * fs))
     y4 = np.nan * np.zeros_like(y3)
 
-    x_tsig = pd.Series(x, index=pd.TimedeltaIndex(t, 's'))
-    ysigs = [pd.Series(y, index=pd.TimedeltaIndex(t, 's')) for y in
+    x_tsig = pd.Series(x, index=pd.to_timedelta(t, 's'))
+    ysigs = [pd.Series(y, index=pd.to_timedelta(t, 's')) for y in
              [y1, y2, y3, y4]]
     ysigs_df = pd.DataFrame(np.array(ysigs).T, columns=[0, 1, 2, 3],
                             index=ysigs[0].index)
@@ -267,8 +267,8 @@ def test_xcorr_delay(delay):
     noise = np.random.random(len(t))/5  # add broadband noise
     x = 0.5*np.sin(2*np.pi*0.01*t) + noise  # noisy signal
     y = np.roll(x, int(delay*fs))  # Explicitly delay the signal
-    x_tsig = pd.Series(x, index=pd.TimedeltaIndex(t, 's'))
-    y_tsig = pd.Series(y, index=pd.TimedeltaIndex(t, 's'))
+    x_tsig = pd.Series(x, index=pd.to_timedelta(t, 's'))
+    y_tsig = pd.Series(y, index=pd.to_timedelta(t, 's'))
 
     # Calculate the averaged transfer function using the function
     delay_result, _ = xcorr_delay(x_tsig, y_tsig, "coeff")
@@ -310,7 +310,7 @@ def test_averaged_psd_multi():
     x4 = 0.5 * np.sin(2 * np.pi * 2 * t) + np.random.random(len(t))
 
     xs = [x1, x2, x3, x4]
-    tsigs = [pd.Series(x, index=pd.TimedeltaIndex(t, 's')) for x in xs]
+    tsigs = [pd.Series(x, index=pd.to_timedelta(t, 's')) for x in xs]
     tsigs_df = pd.DataFrame(np.array(tsigs).T, columns=[0, 1, 2, 3], index=tsigs[0].index)
     navgs = 5
     overlap = 0.5
@@ -342,8 +342,8 @@ def test_averaged_tf_multiout():
     y2 = np.roll(x, int(2*delay * fs))
     y3 = np.roll(x, int(3*delay * fs))
     y4 = np.roll(x, int(4*delay * fs))
-    x_tsig = pd.Series(x, index=pd.TimedeltaIndex(t, 's'))
-    ysigs = [pd.Series(y, index=pd.TimedeltaIndex(t, 's')) for y in [y1, y2, y3, y4]]
+    x_tsig = pd.Series(x, index=pd.to_timedelta(t, 's'))
+    ysigs = [pd.Series(y, index=pd.to_timedelta(t, 's')) for y in [y1, y2, y3, y4]]
     ysigs_df = pd.DataFrame(np.array(ysigs).T, columns=[0, 1, 2, 3], index=ysigs[0].index)
 
     navgs = 5
@@ -378,8 +378,8 @@ def test_averaged_tf_multiin():
     y2 = np.roll(x, int(2*delay * fs))
     y3 = np.roll(x, int(3*delay * fs))
     y4 = np.roll(x, int(4*delay * fs))
-    x_tsig = pd.Series(x, index=pd.TimedeltaIndex(t, 's'))
-    ysigs = [pd.Series(y, index=pd.TimedeltaIndex(t, 's')) for y in [y1, y2, y3, y4]]
+    x_tsig = pd.Series(x, index=pd.to_timedelta(t, 's'))
+    ysigs = [pd.Series(y, index=pd.to_timedelta(t, 's')) for y in [y1, y2, y3, y4]]
     ysigs_df = pd.DataFrame(np.array(ysigs[0]).T, columns=[0], index=ysigs[0].index)
 
     xsigs = [x_tsig, ysigs[0], ysigs[1], ysigs[2]]
@@ -416,8 +416,8 @@ def test_averaged_tf_multiboth():
     y2 = np.roll(x, int(2*delay * fs))
     y3 = np.roll(x, int(3*delay * fs))
     y4 = np.roll(x, int(4*delay * fs))
-    x_tsig = pd.Series(x, index=pd.TimedeltaIndex(t, 's'))
-    ysigs = [pd.Series(y, index=pd.TimedeltaIndex(t, 's')) for y in [y1, y2, y3, y4]]
+    x_tsig = pd.Series(x, index=pd.to_timedelta(t, 's'))
+    ysigs = [pd.Series(y, index=pd.to_timedelta(t, 's')) for y in [y1, y2, y3, y4]]
     ysigs_df = pd.DataFrame(np.array(ysigs).T, columns=[0, 1, 2, 3], index=ysigs[0].index)
 
     xsigs = [x_tsig, ysigs[0], ysigs[1], ysigs[2]]
@@ -464,8 +464,8 @@ def test_compute_delays(delay, compute_delays_modes):
         ys.append(yi)
         delay_ins.append(deli)
 
-    x_tsig = pd.Series(x, index=pd.TimedeltaIndex(t, 's'))
-    ysigs = [pd.Series(y, index=pd.TimedeltaIndex(t, 's')) for y in ys]
+    x_tsig = pd.Series(x, index=pd.to_timedelta(t, 's'))
+    ysigs = [pd.Series(y, index=pd.to_timedelta(t, 's')) for y in ys]
     xsigs = [x_tsig for y in ys]
     xsigs = pd.DataFrame(np.array(xsigs).T, index=x_tsig.index)