Conversation
ET0 fix: both full workflows now divide the daily ET0 by its interval in hours (period_et = 24) before writing //Kurven/ET0, mirroring the rain conversion. The kernel reads the ET0 curve as a mm/h rate, so unconverted daily values were integrated 24x too high. RAM: run_one() now thins each run to its single optimisation row immediately (get_simulation_results_optim(lean = TRUE) + add_overflow_events_and_waterbalance) and returns it; run_scenarios() collects the one-row tibbles and the analyse chunk just binds them. This removes the get_simulation_results_optim_parallel() pass that loaded every run's full time series into memory at once. New 'lean' arg on get_simulation_results_optim() reads only the fields the optimisation summary needs (element rates + both water balances), nulling states/meta/ connected-area rates; its intro message is gated behind debug.
Apply the same memory fix as Wien/BadAussee: run_one() now thins each run to its optimisation row immediately (get_simulation_results_optim(lean = TRUE) + add_overflow_events_and_waterbalance) and returns it; run_scenarios() collects the rows and the analyse chunk binds them, replacing the get_simulation_results_optim_parallel() pass that held every run's full time series at once. No ET0 unit change here: the Eisenstadt workflows use the base.h5 constant ET0 placeholder (0.2 mm/h) and the template rain (already mm/h), so there is no mm/d daily series to convert.
Summary page linking the per-site brute-force outputs (workflow html, result tables, CSVs, interactive plots). Added to .Rbuildignore so R CMD check does not build/execute it; render it manually into the assembled results directory. Result-file links match each workflow's paths$modelname (Wien, BadAussee, Eisenstadt_2005).
Since the workflows now thin per run inside run_one(), reading happens inside the future_lapply batch. A result file that exists but cannot be opened/read (engine crashed mid-write for a scenario, or a transient lock) previously threw H5File.open() 'unable to open file' and aborted the whole render at the run_model chunk. Wrap the open+read in an inner function (own on.exit for handle cleanup) plus tryCatch: such a file is now treated like a missing one -- warn, name the scenario, return NULL -- so add_overflow_events_and_waterbalance() emits an NA row and the batch completes.
R CMD check failed with a codoc WARNING because the .Rd still documented the
old signature (no 'lean'). Regenerate the usage block and add the \item{lean}
documentation to match R/get_simulation_results_optim.R (devtools::document()
equivalent), clearing the only WARNING (the 3 NOTEs are pre-existing).
to do: adapt Eisenstadt to the same parameter
Consistency: - workflow_eisenstadt-2005.Rmd and workflow_eisenstadt-2005_neu.Rmd now unconditionally set //Massnahmenelemente/Mulde_Rigole/Parameter_Evapotranspiration/LAI_LeafAreaIndex = 3.9 (Hoernschemeyer grass value, Water 2023, 15, 2840, Tab. 6, plant type 5). Wien already sweeps LAI over c(3.9, 8.5); Bad Aussee uses the same sweep grid as Wien; Eisenstadt was the outlier still running on the base.h5 default 8.5. New feature: - plot_cost_vs_overflow_volume() -- new exported ggplot helper mirroring plot_wb_tradeoff_overflows() for cost-aware optimisation. x = cost_total (EUR), y = overflow volume in m3 (computed from sum_overflows [mm] * mulde_area [m2] / 1000), points coloured discretely by n_overflows using the same 0..x / ">x" palette and top legend. Plotly tooltip carries the cost breakdown (cost_excavation / _profiling / _filter / _storage / _total) plus the varying param_grid entries. - Each of the four case-study workflow vignettes (workflow_wien, workflow_badaussee, workflow_eisenstadt-2005, workflow_eisenstadt-2005_neu) now renders the plot as simulation_results_optimisation_<site>_cost-vs-overflow-volume.html right after the existing water-balance render, matching its PDF / saveWidget pattern exactly. - vignettes/index.Rmd gains a new "Kosten vs. Überlaufvolumen" section under "Interaktive Visualisierungen" that links to the three top-level sites (Eisenstadt 2005, Wien, Bad Aussee). https://claude.ai/code/session_014QrjF51tg7cMVmsgjmfPNG
Builds on the cost-vs-overflow-volume plot (PR #15) with a new cost-by-overflow-count boxplot (three best-selection variants), richer tooltips and i18n parameter labels, and fixes the Eisenstadt cost pipeline the PR review flagged. Bug fix (PR #15 review blocker): - vignettes/workflow_eisenstadt-2005.Rmd now pipes the joined optimisation results through kwb.raindrop::compute_costs(), like the Wien and Bad Aussee workflows already did. Without it the vignette's plot_cost_vs_overflow_volume() call aborted with "missing column(s): cost_excavation, ..." on Windows, so the cost PDF/HTML was never produced and the exported CSV lacked the cost columns. New plot -- plot_cost_overflow_boxplot() (exported): - Boxplot of total construction cost (EUR, y) per number of overflow events (x). Counts 0..x each get their own box (x = max_n_overflows, as in the sibling plots); higher counts collapse into a single ">x" catch-all box (furthest right, red), keeping the axis readable for the long-tailed 15-year runs (Wien / Bad Aussee reach several hundred overflow events). The ">x" box highlights the scenario with the fewest overflow events above x (closest to valid), best_by breaking ties. - Individual scenarios are overlaid as jittered points whose size scales with a chosen variable (size_by): the overflow volume (m3, default) or the element evapotranspiration share (%). The size scale is calibrated to the valid region (0..x) and capped, with a minimum size, so the many-overflow outliers of the ">x" box do not shrink the valid-region points to invisible dots. - One best scenario per box is highlighted with a black-outlined diamond in that box's group colour (so its tooltip inherits the group colour) and the best of all boxes are joined by a frontier line (mark_best / connect_best). best_by picks the objective, cost as tie-breaker -- "min_cost" (cheapest), "min_overflow" (smallest overflow volume) or "max_evapotranspiration" (highest evapotranspiration) -- so the three variants trace three different frontier lines; label_best annotates the marker ("NN m3 / NN %" or "NN %"). - Each case-study vignette loops the three variants into *_cost-by-overflows-boxplot-{cheapest,min-overflow,max-evap}.html, all linked from vignettes/index.Rmd, which now groups the cost plots under one "Kosten" heading with sub-points (toc_depth 4). Tooltip enrichment (shared helpers in R/cost_tooltip.R): - The plotly tooltip of plot_cost_vs_overflow_volume() now carries the element water balance (evapotranspiration / infiltration / overflow, all in %) in addition to the cost breakdown, and names the chosen storage type on its own bold line, bilingually ("Sickerbox / Infiltration box" or "Schotterrigol / Gravel trench"). - The "varying parameters" block is translated via the new exported default_param_labels() helper -- a hovered point shows e.g. "Muldenflaeche [m2]=125" (de) / "Swale area [m2]=125" (en) instead of the raw "mulde_area=125"; override with param_labels =. - cost_tooltip_labels() / cost_tooltip_text() / build_varying_param_html() are shared so both cost plots emit byte-identical tooltips. Code hygiene: - All non-ASCII characters in R code (string literals in the plot functions and in the vignette code chunks) are unicode-escaped (\uXXXX); the few non-ASCII code comments were rewritten in plain ASCII ("Hoernschemeyer", "2xN"). Markdown prose keeps UTF-8. New exports: plot_cost_overflow_boxplot(), default_param_labels().
Both plot_cost_vs_overflow_volume() and plot_cost_overflow_boxplot() now append the share of scenarios meeting the validity criterion (n_overflows <= x) to the plot title, e.g. "Kosten vs. Ueberlaufvolumen (39 % mit <= 5 Ueberlaeufen)" / "Cost vs. overflow volume (39 % with <= 5 overflows)". It goes in the title, not a ggplot subtitle: ggplotly drops subtitles, so the share would be lost in the interactive HTML the vignettes export. The scatter title drops its old "(Anzahl Ueberlaeufe <= x)" parenthetical, which the share now supersedes. vignettes/index.Rmd now spells out the per-site validity threshold that drives this share: Eisenstadt <= 1 (1-year simulation), Wien / Bad Aussee <= 5 (15-year rain/ET series).
…flow-plot Cost-vs-overflow-volume plot + Eisenstadt LAI 3.9 consistency
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.