ClimateTools.jl is a climate-analysis toolkit built on top of YAXArrays.jl for gridded observations, reanalyses, and climate-model simulations.

The package focuses on the typical climate-services workflow:

1. Read CF-style gridded datasets as YAXArray or Cube objects.
2. Subset them in time and space.
3. Regrid simulations and observations onto a common domain.
4. Bias-correct simulations with quantile mapping, extreme-value correction, or time-variability correction.
5. Compute indicators, aggregations, and diagnostics.

ClimateTools uses YAXArrays xmap and mapCube patterns for whole-dimension transforms and reductions.

• Active development
• Julia 1.10+
• Data model: YAXArray / Cube
• YAXArrays compatibility: 0.7

using Pkg
Pkg.add("ClimateTools")

For local development:

using Pkg
Pkg.develop(path="/path/to/ClimateTools")

using ClimateTools
using YAXArrays

obs = Cube(open_dataset("obs.nc"))
ref = Cube(open_dataset("ref.nc"))
fut = Cube(open_dataset("fut.nc"))

qq = qqmap(obs, ref, fut; method="additive", detrend=true)
ann = annualmax(qq)

That small example already illustrates the main ClimateTools workflow: open data, bias-correct, then compute a climate summary.

For multi-member workflows, you can open a list of NetCDF or Zarr members directly as an ensemble Dataset using open_ensemble_dataset, which lazily opens each file, normalizes member axis order when needed, and stacks the files along a new realization axis.

using ClimateTools

members = open_ensemble_dataset(
    ["member1.nc", "member2.nc", "member3.nc"];
    realization_values=["r1", "r2", "r3"],
)

stats = ensemble_mean_std_max_min(members)

For quantile mapping, qqmap is a seasonally varying method: it removes leap days, groups samples by day of year using a moving +/- window, and applies an additive or multiplicative correction that changes through the annual cycle. Use qqmap_bulk only when one full-sample correction is acceptable and seasonal variation in the bias is not important.

• Dataset opening and subsetting with YAXArrays and DimensionalData selectors
• Regridding on regular, curvilinear, and rotated-pole grids
• Bias correction with qqmap, qqmap_bulk, biascorrect_extremes, and tvc
• Aggregations such as daymean, daysum, annualmax, annualmean, and annualsum
• A large catalog of temperature, precipitation, threshold, and spell-duration indices
• Ensemble summaries and time-series diagnostics
• Thermodynamic helper functions and return-level estimation

The stable documentation is organized around workflows first and reference second.

• Quick start: open data, inspect dimensions, compute a first result
• Data and subsetting: time ranges, polygons, coordinate conventions
• Interpolation and regridding: regular, rotated, and curvilinear workflows
• Bias correction: method selection, assumptions, and validation
• Building climate scenarios: observations plus simulations, interpolation, correction, diagnostics, and derived indicators
• Indices and aggregations: grouped by climate-analysis task
• API overview: exported functions by category

One of the primary use cases for ClimateTools is climate-scenario construction. A typical scenario workflow is:

1. Choose a reference observation or reanalysis dataset.
2. Open one or more model simulations for a historical and future period.
3. Align calendars, time windows, and grids.
4. Interpolate observations and simulations to a common grid if necessary.
5. Bias-correct the model series using a method suited to the variable and objective.
6. Validate the corrected fields and compute indicators, maps, or summary tables.

The documentation includes a dedicated guide for this workflow.

See the built documentation at:

• Stable: https://juliaclimate.github.io/ClimateTools.jl/stable
• Dev: https://juliaclimate.github.io/ClimateTools.jl/dev