Code for generating global monthly diurnal temperature cycle parameters from ERA5-Land (2 m air temperature)

This software package implements the workflow used in the accompanying manuscript to derive monthly climatological diurnal temperature cycle parameters from ERA5-Land 2 m air temperature and to reproduce the validation experiments. Main components Climatology construction: computes multi-year monthly hourly climatology T(s,m,h) from the ERA5-Land hourly archive (1990–2024), where h∈{0,…,23} denotes the UTC hour-of-day. Parameter estimation: fits a dual-harmonic representation (24 h and 12 h components) to the detrended monthly climatological diurnal anomaly ΔT(s,m,h) to obtain parameter fields (Tmean, A1, A2, ϕ1, ϕ2) and fit diagnostics (RMSE、R²). Quality control: generates QC fields derived from the fitted curve (e.g., fitted diurnal temperature range and timing indices) and a binary quality flag based on joint criteria; thresholds are documented in the manuscript Supplement. Reconstruction and validation: reproduces four reconstruction modes under different information constraints (Baseline, Scaled, one-point, two-point) and computes reconstruction metrics against reference hourly series for internal validation (ERA5-Land) and external validation (independent station observations). Inputs and outputs Inputs: ERA5-Land hourly 2 m air temperature obtained from the official source; optional station observations for external validation. This package does not redistribute third-party input data. Outputs: a NetCDF parameter dataset consistent with the Zenodo dataset record (DOI: https://doi.org/10.5281/zenodo.18846788]), and optional intermediate products (climatology checkpoints, QC fields, and validation summaries), depending on the selected workflow stage. Reproducibility All code is written in Python 3.9+ and uses standard scientific libraries. Results should be fully reproducible by following the notebooks in order. License This software is distributed under the MIT License. The parameter dataset itself is licensed under CC BY 4.0 (see Zenodo record).