The project, funded by the German Federal Ministry of Education and Research under grant 01LP1911F, ran from 1 September 2019 to 31 August 2022 at the University of Munich. Its aim was to transform the city‑climate model PALM‑4U into a practical product for municipal use while retaining scientific rigor. The core task of the sub‑project was to integrate the three‑dimensional radiation transfer model TenStream into PALM‑4U, extend it with a building representation, and validate the new scheme against a high‑accuracy Monte‑Carlo benchmark.

To enable realistic 3‑D radiation calculations in urban street canyons, the Monte‑Carlo model MYSTIC was first extended to support triangular grids. This allowed the computation of radiation fluxes and heating rates in arbitrarily complex street geometries. Sensitivity studies were performed to determine the necessary detail of input data such as spectral albedo and bidirectional reflectance distribution functions, and to establish requirements for the PALM radiation module. A key design decision emerged from these experiments: sub‑grid surface effects were omitted in TenStream, and only axis‑aligned buildings were permitted. This simplification facilitated the implementation of building geometry in TenStream while still capturing the dominant 3‑D effects.

The upgraded TenStream model was coupled to PALM‑4U and linked to the PALM land‑surface and vegetation data. A Monte‑Carlo solver, RayLi, was embedded within TenStream to allow direct comparison of results from PALM runs. Experiments on a small Berlin‑based model domain compared the distribution of net solar and thermal fluxes on the ground for several radiation solvers: a conventional 1‑D scheme, the RayLi benchmark, the operational PALM RTM scheme that includes 3‑D surface effects, and the new TenStream implementation. Quantitative results, presented in the project’s tables, showed that TenStream with three streams for direct solar radiation and ten streams for diffuse radiation produced heating rates within a few percent of the RayLi benchmark. For typical large‑eddy simulation (LES) applications, the standard configuration of TenStream was deemed sufficiently accurate, while the full Monte‑Carlo approach remained the most precise but computationally expensive option.

The final phase of the project consolidated the TenStream implementation into the official PALM‑4U release. An automated installation process and comprehensive documentation were provided, and the model was released under the GPL open‑source license, making it freely available to all PALM users. Future development plans focus on improving accuracy and efficiency of the TenStream library, benefiting the broader PALM community.

Collaboration was central to the project’s success. The team worked closely with the partner sub‑project TP5 (grant 01LP1911E), which led the publication effort. Joint tasks included compiling TenStream and its dependencies on the computing systems Cirrus at the HUB and HLRN, coupling TenStream to PALM for complex land‑surface and vegetation scenarios, and documenting the usage of the new solver. The partnership ensured that the technical advances were aligned with user needs and that the results would be disseminated through peer‑reviewed publications under the leadership of TP5. The project’s outcomes, including the integrated TenStream module and validation studies, are already available for use and will be formally published in forthcoming scientific articles.