Cropland Warming Exposure

A Google Colab notebook that quantifies how much of the world's cropland is already exposed to elevated growing-season temperatures, and how that exposure has evolved across three decades. Analysis runs at three geographic scales — global, continental, and country level — for wheat, maize, and rice.

Overview

The notebook combines four open datasets to produce crop-specific, spatially explicit warming exposure metrics:

• GISTEMP v4 (NASA) — observed surface temperature anomalies since 1880
• CropGrids v1.08 — global harvested area at 0.05° resolution for wheat, maize, and rice
• GGCMI Phase 3 — crop-specific planting and maturity calendars at 0.5° resolution, used to gate warming stripe series to the relevant growing season
• Natural Earth 50m — country boundaries with UN M49 continent attribution, used to build country and continental masks

How to run

1. Open cropland_warming_final_1_.ipynb in Google Colab
2. Run Runtime → Run all
3. Approve the Google Drive mount popup when it appears
4. First run downloads ~850 MB of data to My Drive/cropland_warming/data/ and takes roughly 20–30 minutes
5. Subsequent runs skip all downloads and complete in ~10–15 minutes

All figures and CSV exports are saved to My Drive/cropland_warming/outputs/.

Repository structure

cropland_warming_final_1_.ipynb   Main notebook — open in Google Colab
README.md                         This file

Configuration

Key parameters are defined in cells 0.4 and 0.5 and can be changed without touching anything else:

Parameter	Cell	Default	Description
COUNTRY_FIGURES	0.5	['China', 'India', 'United States of America', 'Brazil']	Countries for individual figures. Set to None for all ~200 countries
CROP_HA_CUTOFF	2.4	100.0	Minimum harvested area (ha) per 2° cell to be included in analysis
EXCEEDANCE_THRESHOLD	5.5	1.0	Warming threshold (°C) for the first-exceedance map
SMOOTH_WINDOW	5.5	5	Years of centred running mean used in first-exceedance map

Methodology

Geographic scales

Country and continental masks are built by rasterising Natural Earth 50m polygons onto the GISTEMP 2° grid. Continental groupings follow UN M49 (Africa, Asia, Europe, North America, South America, Oceania). Individual country figures are produced for the countries listed in COUNTRY_FIGURES; set it to None to generate figures for all countries with cropland.

Area weighting

CropGrids harvested area fractions are coarsened from 0.05° to the GISTEMP 2° grid by block-mean averaging. Cells with less than CROP_HA_CUTOFF hectares of harvested area are excluded. All temperature series are weighted by harvested area × cos(latitude).

Climate spirals

Spiral radius represents the crop-area-weighted monthly temperature anomaly relative to the 1951–1980 baseline. Each year forms one loop; colour encodes year from blue (early) to red (recent). No GGCMI seasonal gating is applied — all 12 months contribute. Three panels per figure: wheat, maize, rice.

Warming stripes

Each stripe shows one year's crop-area-weighted growing-season temperature anomaly, gated to the GGCMI Phase 3 calendar for that crop. Winter wheat and spring wheat are kept as separate rows. Rice uses season 1 only (the dominant global harvesting season). Four rows per figure: winter wheat, spring wheat, maize, rice.

Exposure maps

The 30-year mean anomaly (1995–2024 vs 1951–1980 baseline) is computed per grid cell and assigned to warming tiers (≥1°C, ≥1.5°C, ≥2°C, ≥3°C, ≥4°C). Decadal maps show the same tier scheme for 1995–2004, 2005–2014, and 2015–2024 separately. The first-exceedance map shows the earliest year a 5-year centred running mean crossed 1°C at each cropland cell.

Continental amplification

OLS regression of continental crop-weighted annual anomaly against land-surface mean anomaly over 1960–2024. The slope β quantifies how much faster cropland in each continent warms relative to the land surface mean. The decadal scatter shows how this relationship has evolved across 1995–2004, 2005–2014, and 2015–2024.

Output files

File	Description
fig1_spirals_global.png	Global climate spirals — wheat, maize, rice
fig1_spirals_[continent].png	Continental spirals — one file per continent
fig1_spirals_[country].png	Country spirals — one file per country in COUNTRY_FIGURES
fig2_stripes_global.png	Global warming stripes — winter wheat, spring wheat, maize, rice
fig2_stripes_[continent].png	Continental stripes — one file per continent
fig2_stripes_[country].png	Country stripes — one file per country in COUNTRY_FIGURES
fig3_exposure_bars.png	Exposed cropland area (Mha) by threshold and crop
fig3b_exposure_map.png	Decadal warming tier maps — three side-by-side panels
fig3c_paris_maps.png	Side-by-side Paris 1.5°C vs 2.0°C exposure maps
fig3d_first_exceedance.png	First year sustained 1°C exceedance per cropland cell
fig4_amplification_bars.png	Amplification factor β per continent and crop
fig4b_amplification_scatter.png	Continental vs land anomaly scatter by decade
exposure_table.csv	Exposed cropland area (Mha) per threshold and crop
amplification_table.csv	β and R² per continent and crop

Datasets and citations

Dataset	Citation
GISTEMPv4	GISTEMP Team (2025). GISS Surface Temperature Analysis v4. NASA GISS. https://data.giss.nasa.gov/gistemp/
GISTEMPv4	Lenssen, N. et al. (2024). A GISTEMPv4 observational uncertainty ensemble. J. Geophys. Res. Atmos., 129, e2023JD040179
CropGrids v1.08	Tang, H. et al. (2024). CROPGRIDS: a global geo-referenced dataset of 173 crops. Scientific Data, 11, 413. doi:10.1038/s41597-024-03247-7
GGCMI Phase 3	Jägermeyr, J. et al. (2021). Climate impacts on global agriculture emerge earlier in new generation of climate and crop models. Nature Food, 2, 873–885. doi:10.5281/zenodo.5062513
Natural Earth 50m	Made with Natural Earth. Free vector and raster map data @ naturalearthdata.com

Requirements

The notebook installs all dependencies automatically via pip in cell 0.1. The key libraries are xarray, cartopy, geopandas, netCDF4, scipy, and tqdm. No local installation is required beyond a Google account with Drive access.