Introduce rasterized plotting

.gitignore

@@ -1,2 +1,5 @@
 #Ignore all png image files
 *.png
+#Ignore conda location and installation
+conda_loc
+conda

colormaps/gsl_wind_speed.yaml

@@ -0,0 +1,23 @@
+# Colors used for wind speed as defined by NOAA-GSL's pygraf utility
+plot_under: False
+colors:
+ - '#fef8fe'
+ - '#f8d3f9'
+ - '#f1a5f3'
+ - '#e074f0'
+ - '#0045ff'
+ - '#0099ff'
+ - '#00ceff'
+ - '#00e8ff'
+ - '#00ffe6'
+ - '#67d300'
+ - '#7ffa06'
+ - '#b4ff36'
+ - '#eaff12'
+ - '#ffe500'
+ - '#ffc808'
+ - '#ff8608'
+ - '#ff3300'
+ - '#ff0039'
+ - '#f704fc'
+

custom_functions.py

@@ -4,6 +4,7 @@
 """
 import logging
 import uxarray as ux
+import numpy as np
 
 logger = logging.getLogger(__name__)
 
@@ -62,10 +63,38 @@ def vert_min(field: ux.UxDataArray, dim: str = "nVertLevels") -> ux.UxDataArray:
     return vertmin
 
 
+def sum_of_magnitudes(field1: ux.UxDataArray, field2: ux.UxDataArray) -> ux.UxDataArray:
+    """
+    Take two vectors (usually wind vectors) and return the sum of the magnitudes
+    """
+
+    return np.sqrt(np.square(field1) + np.square(field2))
+
+def max_all_times(field: ux.UxDataArray, dim: str = "Time") -> ux.UxDataArray:
+    """
+    Return the maximum value across all input times for a given point.
+    """
+    # Compute differences along Time
+    result = field.max(dim=dim, keep_attrs=True)
+
+    return result
+
+def min_all_times(field: ux.UxDataArray, dim: str = "Time") -> ux.UxDataArray:
+    """
+    Return the minimum value across all input times for a given point.
+    """
+    # Compute differences along Time
+    result = field.min(dim=dim, keep_attrs=True)
+
+    return result
+
 DERIVED_FUNCTIONS = {
     "diff_prev_timestep": diff_prev_timestep,
     "sum_fields": sum_fields,
     "vert_max": vert_max,
     "vert_min": vert_min,
+    "sum_of_magnitudes": sum_of_magnitudes,
+    "max_all_times": max_all_times,
+    "min_all_times": min_all_times,
 }
 

default_options.yaml

@@ -109,16 +109,18 @@ plot:
   # {fnme} = Name of file (minus extension) being read for plotted data
   # {date} = The date of plotted data, in %Y-%m-%d format
   # {time} = The time of plotted data, in %H:%M:%S format
+  # {maxval} = The maximum value in the plotted data
+  # {minval} = The minimum value in the plotted data
 
   filename: '{var}_{lev}.png'
   format: null
   title:
     text: 'Plot of {varln}, level {lev} for MPAS forecast, {date} {time}'
     fontsize: 8
   exists: rename
-  dpi: 300
-  figheight: 4
-  figwidth: 8
+  dpi: 200
+  figheight: 3
+  figwidth: 6
 
   # colormap:
   # Color scheme to use for output plots. Options can either be standard Matplotlib colormaps (reference
@@ -130,11 +132,23 @@ plot:
   #
   colormap: "viridis"
 
+  # pixel_ratio:
+  # This controls the quantity of pixels to sample in the rasterization process; higher numbers result
+  # in higher quality plots, though at a cost of plotting speed.
+  pixel_ratio: 1
+
+  # polycollection:
+  # NOT RECOMMENDED
+  # This is the legacy plotting method that converts the unstructured grid to a set of polygons.
+  # This can be orders of magnitude slower than the default raster method and so is not
+  # recommended for large domains; if you need more detail in your plot, it's recommended to
+  # increase the "pixel_ratio" setting.
   # periodic_bdy:
-  # For periodic domains (including global), the plot routines will omit the boundary cells by default. To plot
+  # For periodic domains (including global), the polycollection routines will omit the boundary cells by default. To plot
   # all data, including boundaries, set this option to True, but note that it will slow down plotting substantially.
   #
-  periodic_bdy: False
+  polycollection: False
+  periodic_bdy: True # Set to false once UXarray issue is fixed! https://github.com/UXARRAY/uxarray/issues/1507
 
   # vmin, vmax:
   # By default the color range will be scaled to the max/min values of the plotted data. To use a custom range,

environment.yml

@@ -5,10 +5,10 @@ channels:
   - ufs-community
   - defaults
 dependencies:
-  - numpy=1.26*
+  - numpy
   - matplotlib
   - netcdf4
-  - xarray=2025.9.0
   - cartopy
   - uwtools=2.9*
-  - uxarray=2025.05*
+  - xarray=2026.04
+  - uxarray=2026.04.1

plot_functions.py

@@ -7,6 +7,7 @@
 import os
 import traceback
 
+import numpy as np
 import cartopy.crs as ccrs
 
 logger = logging.getLogger(__name__)
@@ -47,6 +48,10 @@ def set_patterns_and_outfile(valid, var, lev, filepath, field, ftime, plotdict):
     #filename minus extension
     fnme=os.path.splitext(filename)[0]
 
+    # max and min values for plotted field
+    maxval=float(field.max().compute())
+    minval=float(field.min().compute())
+
     pattern_dict = {
         "var": var,
         "lev": lev,
@@ -56,7 +61,9 @@ def set_patterns_and_outfile(valid, var, lev, filepath, field, ftime, plotdict):
         "fnme": fnme,
         "proj": plotdict["projection"]["projection"],
         "date": "no_Time_dimension",
-        "time": "no_Time_dimension"
+        "time": "no_Time_dimension",
+        "maxval": f"{maxval:.2f}",
+        "minval": f"{minval:.2f}",
     }
     if field.attrs.get("units"):
         pattern_dict.update({
@@ -220,3 +227,67 @@ def set_map_projection(confproj) -> ccrs.Projection:
 
     raise ValueError(f"Invalid projection {proj} specified; valid options are:\n{valid}")
 
+
+def get_data_extent_raster(raster, lon_bounds=(-180, 180), lat_bounds=(-90, 90)):
+    """
+    Computes data extent from image raster for automatic zooming to data domain
+
+    Parameters
+    ----------
+    raster : np.ndarray
+        2D raster array with NaNs outside valid region
+    lon_bounds : tuple(float, float)
+        Longitude range corresponding to full raster width
+    lat_bounds : tuple(float, float)
+        Latitude range corresponding to full raster height
+
+    Returns
+    -------
+    extent : list [lon_min, lon_max, lat_min, lat_max]
+    """
+    valid = ~np.isnan(raster)
+    if not np.any(valid):
+        # no data at all
+        return lon_bounds + lat_bounds
+
+    # pixel indices of valid data
+    ys, xs = np.where(valid)
+
+    # convert indices to lon/lat using proportional scaling
+    nrows, ncols = raster.shape
+    lon_min, lon_max = lon_bounds
+    lat_min, lat_max = lat_bounds
+
+    x_min = lon_min + (xs.min() / ncols) * (lon_max - lon_min)
+    x_max = lon_min + (xs.max() / ncols) * (lon_max - lon_min)
+    y_min = lat_max - (ys.max() / nrows) * (lat_max - lat_min)
+    y_max = lat_max - (ys.min() / nrows) * (lat_max - lat_min)
+
+    pad_fraction=0.05
+    dx = (x_max - x_min) * pad_fraction
+    dy = (y_max - y_min) * pad_fraction
+    # y dimension is flipped for some reason
+    return [x_min - dx, x_max + dx, -y_max - dy, -y_min + dy]
+
+
+def get_data_extent(uxda, pad_fraction=0.05):
+    """Return (lon_min, lon_max, lat_min, lat_max) in degrees, with buffer."""
+    try:
+        if "n_face" in uxda.dims:
+            lons = getattr(uxda.uxgrid, "node_lon", None)
+            lats = getattr(uxda.uxgrid, "node_lat", None)
+        else:
+            lons = uxda.lon
+            lats = uxda.lat
+
+        lon_min = np.nanmin(lons)
+        lon_max = np.nanmax(lons)
+        lat_min = np.nanmin(lats)
+        lat_max = np.nanmax(lats)
+
+        dx = (lon_max - lon_min) * pad_fraction
+        dy = (lat_max - lat_min) * pad_fraction
+
+        return [lon_min - dx, lon_max + dx, lat_min - dy, lat_max + dy]
+    except Exception as e:
+        raise RuntimeError(f"Could not determine lat/lon bounds: {e}")