. . . . . . "Duration x return-period scaling of precipitation extremes over Germany in the Destination Earth Climate DT" . . . . "Different data and different statistics from the original; both are\ndeliberate, and together they are what make agreement informative rather than circular.\n\n1. Model: a single coupled global Earth-system model (IFS-NEMO, one realization) in place\n of a four-member ensemble of CMIP5 GCMs downscaled with COSMO-CLM.\n2. Resolution: about 5 km (HEALPix level 10, ~6.3 km) in place of 2.8 km. Deep convection\n is therefore NOT explicitly resolved here, whereas the original's added value came\n precisely from resolving it.\n3. Domain: all of Germany (8,697 cells) in place of the original's convection-permitting\n nest over Southern and Central Germany (~411,000 km2).\n4. Scenario: SSP3-7.0 in place of RCP 8.5.\n5. Periods: two 20-year windows (1991-2010, 2030-2049) in place of a transient 1971-2100\n simulation with a 1971-2005 reference and 30-year running windows.\n6. Warming: one interval of 0.93 K of global-mean warming between the windows, in place of\n the original's GWL2 and GWL3 (2 K and 3 K). This study therefore cannot test the\n linearity of the response with warming, only the ordering at one interval.\n7. Durations: 1, 3, 6, 12, 24 hours in place of 1, 6, 12, 24, 72 hours. There is no 72-hour\n duration here and no 3-hour duration there; the shortest, most convectively driven\n durations the original emphasises are below the archive's hourly resolution.\n8. Estimator: a regional index-flood GEV fitted by L-moments to annual maxima, in place of\n a partial-series peak-over-threshold selection with an exponential fit in logarithmic\n space following the German DWA-A 531 guideline.\n9. Return periods: up to 20 years in place of up to 100 years. Twenty years of data cannot\n support a 100-year return level even with pooling.\n10. No sub-hourly correction: the original applies DWA-A 531 factors (1.14, 1.07, 1.04,\n 1.03) for event durations up to 4 hours to compensate for hourly sampling. This study\n applies none, so its short-duration intensities are, if anything, conservative — which\n matters because the short durations are where it finds the largest changes.\n11. No observational comparison: the original evaluates against DWD station data and\n KOSTRA-DWD-2010R. This study compares model windows to each other only." . . "Data: Destination Earth Climate DT, generation 2 — model IFS-NEMO,\nrealization 1, hourly `clte` stream, variable `avg_tprate` (hourly-mean precipitation\nrate, kg m-2 s-1), `levtype=sfc`, `resolution=high`, i.e. the native HEALPix level 10\ngrid (nside 1024, about 6.3 km). Two 20-year windows of equal length: 1991-2010\n(activity `baseline`, experiment `hist`) and 2030-2049 (activity `projections`,\nexperiment SSP3-7.0).\n\nRetrieval: ECMWF Polytope server-side polygon feature extraction over Germany, one\nrequest per calendar month (480 in total, about 23 GB). MARS cannot crop a HEALPix field\nwith a lat/lon box, so server-side extraction is what makes 20 years of hourly level-10\ndata tractable. The domain polygon is committed with the code (Eurostat NUTS 2021\nlevel 0, simplified to 128 vertices) so the study area is fixed. 8,697 cells fall inside\nit.\n\nReduction: rates converted to millimetres per hour (x3600); backward rolling\naccumulations over 1, 3, 6, 12 and 24 hours; the annual maximum of each accumulation per\ncell per calendar year, with each year carrying the tail of the previous one so a\n24-hour total ending on 1 January still sees 31 December.\n\nEstimation: a regional index-flood analysis per (window, duration). Each cell's\nannual-maxima series is divided by its own at-site mean, the standardised series are\npooled across all cells, and a single generalised extreme value distribution is fitted to\nthe pool by L-moments (Hosking). Return levels for 2, 5, 10 and 20 years are the pooled\ngrowth factor times each cell's index flood. Pooling is what allows a 20-year return\nlevel from a 20-year window, and is the same device the original uses. Uncertainty comes\nfrom resampling CELLS with replacement (n=400), because spatial correlation, not record\nlength, limits the effective independence of the pooled sample.\n\nComparison: the change per matrix cell is the median over grid cells of the per-cell\nratio of return levels between windows, so each cell's own climatology cancels.\n\nCoordinate systems: the Climate DT is delivered on HEALPix defined on a sphere. The whole\nanalysis is repeated on HEALPix on the WGS84 ellipsoid, converted with healpix-resample's\nconservative (mass-preserving) operator, and both grids are reported.\n\nData attribution (Destination Earth Terms and Conditions v2.0, Art. 3.4): this\ndata is created based on data of the European Union, using the Destination Earth\nPlatform, but has been modified by Anne Fouilloux. The results are a derived\nproduct - aggregation and transformation of Climate DT output - and are not\noriginal DestinE Data. The underlying hourly data is not redistributed.\n\nSoftware: Python; numpy, scipy, xarray, earthkit-data, polytope-client, healpix-geo,\nhealpix-resample. The extreme-value estimator, the coordinate conversion and the verdict\nlogic are unit tested — the GEV against scipy.stats.genextreme, and the verdict against\nmatrices whose answer is known by construction." . "Tested: the ORDERING the claim asserts — that extreme precipitation events of\nshort duration and long return period intensify proportionally more than events of long\nduration and short return period. Operationally, whether the largest change in return\nlevel across a duration x return-period matrix falls at the short-duration,\nlong-return-period corner, and whether that corner exceeds the long-duration,\nshort-return-period corner.\n\nNot tested: the original's reported magnitude of 6-8.5% per degree of global warming.\nThat figure is a maximum across the members of one convection-permitting ensemble over\none domain; elsewhere it can only be re-measured, not tested. It is recorded in the\nOutcome as evidence alongside this study's own per-degree figures, with both definitions\nattached.\n\nAlso not tested: the original's comparison against observations (DWD stations,\nKOSTRA-DWD-2010R), its 72-hour duration, its return periods beyond 20 years, and its\nGWL2/GWL3 warming levels." . . . "Anne Fouilloux" . "2026-09-07T12:44:07.550Z"^^ . . . . . "Duration x return-period scaling of precipitation extremes over Germany in the Destination Earth Climate DT" . . "RSA" . "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" . "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" . . .