Seasonal ice storage of desalinated water for district cooling and agricultural irrigation

This is the authors' abstract. We don't add key points for this paper.

Code, input data and results for a study of seasonal ice and desalinated water storage, sized and costed for Riyadh and then evaluated worldwide. This archive accompanies the paper Seasonal ice and desalinated water storage, a solution for cooling, agricultural irrigation and seasonal energy storage. It contains everything needed to reproduce the study, and the numbers behind every figure panel and every table as plain CSV, so the results can be read without running anything. The arrangement studied is a lined earth pit of five million cubic metres and 30 m depth, filled with desalinated seawater in winter, frozen with winter electricity, melted for district cooling through the summer, and drained to irrigated fields. The model fixes the pit geometry and its heat gain, dispatches the ice plant and the pit hour by hour against a cooling-degree-hour load proxy, and then costs the scheme on three attribution scenarios. The same calculation is repeated at every land cell of a global 0.5° grid, of which 45,595 of 61,844 return a result. Weather is the ERA5-HRES hourly reanalysis over 2015 to 2024. Contents: Model and scripts, 19 Python files covering the pit geometry and dispatch, the cost model, the global grid, the comparison against gas and against solar with batteries, the figure generation, and the verification scripts that check every number in the text against the results files Input data, the ERA5-HRES hourly and daily extracts for Riyadh and the global grid definition Results, the model output as JSON and the computed global field as a compressed array Figures, all 28 images plus an editable vector deck Figure data, 12 CSV files, including the full 8,760-hour dispatch of the climatological year and the global result with one row per land cell Table data, 10 CSV files, one for every table in the paper and the Supplementary Information Reproducing the results needs Python 3.10 with numpy, matplotlib, python-docx and python-pptx. Running model.py, then alternatives.py, then global_map.py regenerates every result; verify_numbers.py then checks the manuscript against them. The archive has been tested from a clean copy and reproduces the published results exactly. Two hourly reanalysis arrays on the global grid are not redistributed here, because each is 8.5 GB and the data belong to the European Centre for Medium-Range Weather Forecasts. The script that rebuilds them from the Copernicus Climate Data Store is included, as is the field computed from them, so every global figure and every map value can be reproduced without them. The Python code is released under the MIT licence. The data, figures and manuscript sources are released under Creative Commons Attribution 4.0. The ERA5 extracts remain subject to the Copernicus licence.