Visualization

ClimateTools now includes a small plotting API for the most common climate-analysis views:

geomap for geographic fields
geomapfacet for multi-panel geographic comparisons over time or ensemble dimensions
timeseriesplot for point or reduced time series
statsplot for quick distribution and comparison plots

These methods are available when GeoMakie.jl and a Makie backend are loaded.

Setup

using ClimateTools
using GeoMakie
using CairoMakie

CairoMakie.activate!()

Geographic Maps

For regular longitude-latitude grids, geomap can infer the spatial axes automatically.

fig = geomap(cube; dim=:time, index=1)

For a more publication-oriented layout, geomap now exposes explicit map controls for projection, framing, and colorbars:

fig = geomap(cube;
    dim=:time,
    index=1,
    colorbar=true,
    colorbar_label="Temperature (K)",
    colorbar_position=:right,
    coastline_width=1.5,
    dest="+proj=eqearth +lon_0=0",
    frame=false)

The most useful map keywords are:

source, dest, lon_0 for explicit projection control
limits, fit_limits, limit_padding for controlling the displayed longitude-latitude extent
colorbar, colorbar_label, colorbar_position for layout and legend-like scaling
coastline_color, coastline_width for cartographic emphasis
frame to suppress axis decorations in cleaner publication layouts
axis_kwargs, surface_kwargs, colorbar_kwargs for lower-level Makie customization

By default, geomap and geomapfacet fit the visible map extent to the minimum and maximum longitude and latitude of the plotted data. This works for both ordinary lon-lat grids and rotated-pole inputs once they have been transformed back to geographic coordinates.

If you want to override the default view window explicitly:

fig = geomap(cube;
    dim=:time,
    index=1,
    limits=((-75.0, -55.0), (43.0, 63.0)))

If you want GeoMakie to manage the limits instead, disable the automatic fit:

fig = geomap(cube; dim=:time, index=1, fit_limits=false)

If your cube still has additional non-spatial dimensions, use selectors to reduce them first or specify which dimension should be sliced:

fig = geomap(cube; selectors=(member=1,), dim=:time, index=10)

For rotated-pole or curvilinear datasets, pass the parent dataset and variable name so ClimateTools can recover the geographic coordinates from grid_mapping, lon, and lat metadata:

ds = open_dataset("rotated_model.nc")
fig = geomap(ds, :tas; dim=:time, index=1)

The default projection is selected from the coordinate extent. Global fields use an equal-earth style projection, while regional fields default to geographic longitude-latitude coordinates.

Faceted Maps

Use geomapfacet when you want several aligned panels with a shared projection and shared color scale.

fig = geomapfacet(cube;
    facetdim=:time,
    ncols=3,
    colorbar=true,
    title="Daily fields")

If you want all panels to use one common geographic window, even when the facet payloads do not share the same spatial extent, enable shared_spatial_limits:

fig = geomapfacet(cube;
    facetdim=:time,
    ncols=3,
    shared_spatial_limits=true)

This is separate from limits=...:

shared_spatial_limits=true computes one unioned longitude-latitude window from all facet payloads
limits=((lonmin, lonmax), (latmin, latmax)) enforces an explicit user-defined window

Projection sharing is now explicit as well:

shared_projection=true keeps one common destination projection across all panels
shared_projection=false allows dest to vary per panel
shared_dest is accepted as an alias for the same boolean behavior

With a shared projection, pass one dest string for all panels:

fig = geomapfacet(cube;
    facetdim=:time,
    indices=1:2,
    shared_projection=true,
    dest="+proj=eqearth +lon_0=0")

If you want panel-specific projections, disable projection sharing and pass either a collection or a function:

fig = geomapfacet(cube;
    facetdim=:time,
    indices=1:2,
    shared_projection=false,
    dest=["+proj=eqearth +lon_0=0", "+proj=longlat +datum=WGS84"])

For ensemble data, facet over the ensemble dimension and fix the time slice with selectors:

fig = geomapfacet(cube;
    facetdim=:member,
    selectors=(time=1,),
    ncols=2)

geomapfacet is especially useful when comparing:

sequential dates from the same simulation
ensemble members at a fixed date
raw and corrected outputs after regridding or bias correction

The default behavior uses a shared color range across panels. This keeps visual comparisons honest, especially for multi-member diagnostics.

Time-Series Inspection

Extract a single grid-cell series with selectors:

fig = timeseriesplot(cube; selectors=(longitude=10, latitude=12))

Or collapse the remaining spatial dimensions with a reducer such as mean:

fig = timeseriesplot(cube; reducer=mean, ylabel="Regional mean")

When a cube contains an ensemble-like dimension such as member, scenario, or run, timeseriesplot can render all members directly:

fig = timeseriesplot(cube;
    selectors=(longitude=10, latitude=12),
    mode=:lines)

Or collapse the ensemble into a central line plus spread ribbon:

fig = timeseriesplot(cube;
    selectors=(longitude=10, latitude=12),
    mode=:mean_ribbon,
    spread=:minmax)

If you already summarized a cube with ensemble_stats, you can visualize the resulting :stats dimension directly:

stats = ensemble_stats(cube; dim="member")
fig = timeseriesplot(stats; selectors=(longitude=10, latitude=12), mode=:stats)

You can also compare several already extracted series directly:

fig = timeseriesplot((obs=obs_series, sim=sim_series, corrected=bc_series))

Statistical Summaries

statsplot is intended for fast diagnostics rather than publication styling.

fig = statsplot((raw=raw_values, corrected=corrected_values); kind=:hist)
fig = statsplot((raw=raw_values, corrected=corrected_values); kind=:boxplot)
fig = statsplot((obs=obs_values, sim=sim_values); kind=:scatter)

For a single vector or cube slice, histogram and boxplot views are supported directly.

For ensemble cubes, you can group distributions by member or scenario:

fig = statsplot(cube; groupdim=:member, kind=:boxplot)

When to Use These Helpers

These functions are meant for:

workflow diagnostics after regridding or bias correction
comparing raw and corrected simulations
checking temporal behavior at a grid cell or after regional averaging
quickly inspecting rotated-grid outputs without first hand-building coordinate transforms

For highly customized figure layouts, annotations, or publication graphics, it is still reasonable to extract arrays from ClimateTools outputs and build Makie figures manually.

What to Plot During Scenario Workflows

Useful visual checks include:

the reference observation field
the raw model field on the same date or period
the regridded model field
the corrected field
the difference between corrected and raw simulations

For bias-correction work, plotting a few representative time series is often as informative as map slices.