Merge branch 'develop' into pr_3888

Author: dlebauer

.github/workflows/docker-build-pr.yml

@@ -0,0 +1,76 @@
+name: Docker build (fork PRs)
+
+# A fast check for Docker build errors introduced by the current PR.
+# Implemented separately from the full-stack build in docker.yml because 
+# pull requests opened from forks get a read-only GITHUB_TOKEN and cannot push
+# to ghcr.io/pecanproject, so the normal Docker GHA workflow (docker.yml) skips
+# them (see the gate on its `rversion` job). This workflow gives those PRs a
+# real "does the image stack still build" signal without pushing anything.
+#
+# The images build on top of each other (depends -> base -> models -> sipnet).
+# The normal workflow runs each as a separate job and pulls the parent back from
+# ghcr, which needs a registry we can't push to from a fork. Here we build the
+# whole chain in a single job with plain `docker build`, so each image stays in
+# the runner's local image store and the next `FROM pecan/<parent>:latest`
+# resolves locally. No registry, no push, no secrets.
+#
+# Scope is the core linear chain only (depends -> base -> models -> sipnet). It
+# fits the default runner disk (~19 GB free; the resident stack is ~5-6 GB). The
+# other model binaries, baseplus (docs/executor/api) and extras images are not
+# built here to keep the job within disk and time budget.
+
+on:
+  pull_request:
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  build-stack:
+    # Only fork PRs. Same-repo PRs are already covered by docker.yml, which can
+    # push pr-* tags to ghcr because they run with a writable token.
+    if: github.event.pull_request.head.repo.full_name != github.repository
+    runs-on: ubuntu-latest
+    env:
+      R_VERSION: "4.4"
+    steps:
+      - uses: actions/checkout@v6
+
+      - name: Disk space before build
+        run: df -h /
+
+      # NOTE: do not add docker/setup-buildx-action here. Plain `docker build`
+      # uses the default docker-engine builder, which resolves each
+      # `FROM pecan/<parent>:latest` from the local image store built by the
+      # previous step. The docker-container buildx driver would not see those
+      # local images and every FROM after depends would fail.
+      - name: Build depends -> base -> models -> sipnet (no push)
+        run: |
+          set -euxo pipefail
+
+          # depends: FROM rocker/tidyverse:${R_VERSION}
+          docker build -f docker/depends/Dockerfile \
+            --build-arg R_VERSION="${R_VERSION}" \
+            -t pecan/depends:latest \
+            docker/depends
+
+          # base: FROM pecan/depends:latest (resolved from the local store)
+          docker build -f docker/base/Dockerfile \
+            -t pecan/base:latest \
+            .
+
+          # models: FROM pecan/base:latest
+          docker build -f docker/models/Dockerfile \
+            -t pecan/models:latest \
+            docker/models
+
+          # sipnet: FROM pecan/models:latest
+          docker build -f models/sipnet/Dockerfile \
+            --build-arg MODEL_VERSION=git \
+            -t pecan/model-sipnet-git:latest \
+            models/sipnet
+
+      - name: Disk space after build
+        if: always()
+        run: df -h /

.github/workflows/docker.yml

@@ -42,6 +42,15 @@ jobs:
   # As an ugly workaround, we assign it to a job output instead.
   # ----------------------------------------------------------------------
   rversion:
+    # Every build job below needs this job, so gating it here skips the whole
+    # push-based Docker stack for pull requests opened from forks. Forks get a
+    # read-only GITHUB_TOKEN and cannot push to ghcr.io/pecanproject, so those
+    # jobs always failed at the push step. Fork PRs are instead built (without
+    # pushing) by docker-build-pr.yml. Non-fork events (push, merge_group,
+    # schedule, workflow_dispatch) and same-repo PRs still run the full stack.
+    if: >-
+      github.event_name != 'pull_request' ||
+      github.event.pull_request.head.repo.full_name == github.repository
     runs-on: ubuntu-latest
     steps:
       - id: mon

CHANGELOG.md

@@ -9,6 +9,7 @@ For more information about this file see also [Keep a Changelog](http://keepacha
 ## Unreleased
 
 ### Added
+- Added `make_scorecard.sh` and documentation to `inst/ilamb/` in PEcAn.benchmark for generating and serving the ILAMB HTML scorecard.
 - New function `PEcAn.utils::netcdf2df()` flattens all dims and vars of a netCDF into a dataframe,
     with units attached as an attribute.
 - New package `PEcAn.RothC` runs the RothC soil carbon model.
@@ -31,6 +32,7 @@ For more information about this file see also [Keep a Changelog](http://keepacha
 - Added `PEcAn.data.land::event_parquet_to_json` for generating PEcAn `event.json` files from well-formatted event parquet files, with support for ensembles of events.
 
 ### Fixed
+- Docker GHA workflow no longer fails on pull requests opened from forks (#3618).
 - Removed unused `grid2netcdf()` from `PEcAn.data.remote` and fixed R CMD check reference notes for `download.LandTrendr.AGB()` (#2758).
 - Fixed broken pecanproject.github.io, pecan.gitbooks.io, and other outdated documentation links across book_source, tutorials, models, modules, web, and shiny files (#3710).
 - Added note to DEV-INTRO.md documenting Traefik workaround for Apple Silicon (ARM64) Macs: use `traefik:v2.11` with `platform: linux/arm64` to fix 404 errors (#3910)

docker/depends/pecan_package_dependencies.csv

@@ -495,6 +495,7 @@
 "rcrossref","*","base/db","Suggests",FALSE
 "readr","*","models/ldndc","Imports",FALSE
 "readr","*","modules/assim.sequential","Suggests",FALSE
+"readr","*","modules/data.land","Suggests",FALSE
 "REddyProc","*","modules/data.atmosphere","Imports",FALSE
 "redland","*","modules/data.land","Suggests",FALSE
 "reshape","*","modules/data.remote","Suggests",FALSE
@@ -700,6 +701,7 @@
 "units","*","modules/data.atmosphere","Imports",FALSE
 "units",">= 0.8.7","base/utils","Imports",FALSE
 "urltools","*","base/remote","Imports",FALSE
+"usethis","*","modules/data.land","Suggests",FALSE
 "utils","*","base/all","Imports",FALSE
 "utils","*","base/logger","Imports",FALSE
 "utils","*","models/ed","Imports",FALSE

modules/benchmark/inst/ilamb/README.md

@@ -296,3 +296,21 @@ score relative to the other variables, and the `weight` on each dataset block
 sets how much that dataset counts within its variable. Both are relative
 weights, not percentages. See the inline comments in the config for the full
 per-field reference.
+
+## Generating the scorecard
+
+ILAMB produces an interactive HTML scorecard alongside the numeric scores. To
+generate it for a window, run `make_scorecard.sh` against that window's model
+root (built by the pipeline above):
+
+```bash
+export ILAMB_ROOT=/path/to/ILAMB_ROOT
+./make_scorecard.sh 2012_2014
+./make_scorecard.sh 2015_2023
+```
+
+Each run writes a self-contained static site (`index.html` plus assets) to a
+build directory, which can be served directly by copying it under a web
+server's document root. The scorecard uses the clean model roots, so it lists
+PEcAn alongside the individual CMIP6 and TRENDY models and the ensemble means,
+rather than the 100 individual PEcAn members used for the spread analysis.

modules/benchmark/inst/ilamb/make_scorecard.sh

@@ -0,0 +1,60 @@
+#!/bin/bash
+#
+# Generate the ILAMB HTML scorecard for a PEcAn benchmarking window.
+#
+# Runs ilamb-run against a window's model root (PEcAn, the individual CMIP6
+# and TRENDY models, and the ensemble means) to produce the interactive HTML
+# scorecard ILAMB writes to the build directory. This is the public-facing
+# comparison view; it uses the clean model roots (no individual PEcAn members),
+# so the scorecard lists PEcAn alongside the models rather than 100 member rows.
+#
+# Prerequisites:
+#   - ILAMB installed and on PATH (ilamb-run).
+#   - ILAMB_ROOT set, with the benchmark datasets under $ILAMB_ROOT/DATA.
+#   - The window model root already built (see build_window_ensembles.py),
+#     e.g. ilamb_models_2012_2014 / ilamb_models_2015_2023.
+#
+# Usage:
+#   ./make_scorecard.sh <window> [model_root] [build_dir]
+#
+#   window      Required. A label for the run, e.g. 2012_2014 or 2015_2023.
+#   model_root  Optional. Directory of model entities to score.
+#               Default: ilamb_models_<window>.
+#   build_dir   Optional. Output directory for the HTML scorecard.
+#               Default: ilamb_scorecard_<window>.
+#
+# Example:
+#   export ILAMB_ROOT=/path/to/ILAMB_ROOT
+#   ./make_scorecard.sh 2012_2014
+#   ./make_scorecard.sh 2015_2023
+#
+# The generated build directory (index.html plus assets) is a self-contained
+# static site that can be served directly, for example by copying it under a
+# web server's document root.
+
+set -euo pipefail
+
+WINDOW="${1:?usage: make_scorecard.sh <window> [model_root] [build_dir]}"
+MODEL_ROOT="${2:-ilamb_models_${WINDOW}}"
+BUILD_DIR="${3:-ilamb_scorecard_${WINDOW}}"
+CONFIG="$(dirname "$0")/pecan_ilamb.cfg"
+
+if [ -z "${ILAMB_ROOT:-}" ]; then
+  echo "ERROR: ILAMB_ROOT is not set." >&2
+  exit 1
+fi
+if [ ! -d "$MODEL_ROOT" ]; then
+  echo "ERROR: model root '$MODEL_ROOT' not found. Build it first." >&2
+  exit 1
+fi
+
+echo "Generating scorecard for window '$WINDOW'"
+echo "  model root: $MODEL_ROOT"
+echo "  build dir:  $BUILD_DIR"
+
+ilamb-run --config "$CONFIG" \
+  --model_root "$MODEL_ROOT" \
+  --build_dir "$BUILD_DIR" \
+  --regions global
+
+echo "Done. Open $BUILD_DIR/index.html, or serve $BUILD_DIR as a static site."

modules/data.land/DESCRIPTION

@@ -71,13 +71,15 @@ Suggests:
     PEcAn.settings,
     PEcAn.visualization,
     raster,
+    readr,
     redland,
     reticulate,
     rjags,
     sirt,
     sp,
     testthat (>= 3.1.0),
     traits,
+    usethis,
     withr
 Remotes:
     github::ropensci/traits

modules/data.land/NAMESPACE

@@ -46,8 +46,8 @@ export(get_veg_module)
 export(ic_process)
 export(id_resolveable)
 export(load_veg)
-export(look_up_ca_compost_amendment)
 export(look_up_ca_n_rate)
+export(look_up_ca_organic_amendment)
 export(look_up_fertilizer_components)
 export(matchInventoryRings)
 export(match_pft)

modules/data.land/NEWS.md

@@ -12,11 +12,12 @@
 * Datasets
   *  `landiq_crop_mapping_codes` dataset mapping LandIQ crop classification codes to human-readable crop names.
   *  `bism_kc_by_crop` dataset containing BISm crop coefficient schedules and stage timing references for use in ET estimation, including columns that map to LandIQ class and subclass.
-  *  `ca_n_application_rate` dataset with recommended N application rates (g N/m2) for 33 California crops from CDFA-FREP and UC ANR sources.
-  *  `ca_compost_amendment` dataset with C:N ratios, carbon, nitrogen, and PAN (g/m2) for 32 organic amendment materials.
+  *  `ca_n_application_rate` dataset with recommended N application rates (g N/m2) for 40 California crops from CDFA-FREP, UC ANR, Rosenstock 2013, Brown 2020 NBMP (Almonds age stages) and Lazicki 2016 (Potato) sources.
+  *  `ca_organic_amendment_properties` dataset with C:N ratios, nitrogen, and PAN for 32 organic amendment materials, plus a `material_class` column mapping each material to the CalRecycle taxonomy (14 CCR section 17852).
+  *  `ca_organic_amendment_app_rate` dataset with row-crop and orchard application rate envelopes (64 rows = 32 materials x 2 crop structures) that joins to `ca_organic_amendment_properties` on `material`.
 * Functions
   *  `look_up_ca_n_rate()` for looking up crop-specific N application rates by name (exact match first, partial match suggestions on miss).
-  *  `look_up_ca_compost_amendment()` for looking up organic amendment properties by material name.
+  *  `look_up_ca_organic_amendment()` for looking up organic amendment properties by material name.
   *  `to_co2e()` for converting SOC change, CH4, and N2O to CO2-equivalent emissions using IPCC Global Warming Potential values.
 
 ## Changed