feat: climate risk from cmip cordex by jdroenner · Pull Request #18 · geo-engine/BioIS

jdroenner · 2026-07-08T22:55:23Z

No description provided.

coveralls · 2026-07-08T23:04:48Z

coverage: 82.65% (+2.2%) from 80.475% — cmip_cordex into main

ChristianBeilschmidt

Konnte das aus Gründen (Karte) nicht anwenden. Aber hier schon mal meine Kommentare.

ChristianBeilschmidt · 2026-07-09T06:17:56Z

      volumeMounts:
        - name: pgdata
-          mountPath: /var/lib/postgresql/data
+          mountPath: /var/lib/postgresql/18/docker


Warum ist das anders, obwohl sich sonst nichts an der Container-Config geändert hat?

docker-library/postgres#1259

ChristianBeilschmidt · 2026-07-09T06:24:59Z

+
+/// Calculates climate-risk indicators from CORDEX/CMIP5 climate data for a given point and time window.
+#[derive(Debug, Clone)]
+pub struct ClimateCordex5Process;


Vielleicht mehr User-facing benennen. Was macht der User damit?

Cordex sagt ja mehr über die Quelldaten.

ich hab es umbenannt

Ok, am besten auch noch Modul umbenennen.

ChristianBeilschmidt · 2026-07-09T06:41:50Z

+    let wfl_ids =
+        try_join_all(variable_workflows.map(|wfl| register_workflow_handler(configuration, wfl)))
+            .await;
+
+    let wfl_ids = match wfl_ids {
+        Ok(id) => id,
+        Err(e) => {
+            if let Some(error) = error_response(&e) {
+                anyhow::bail!("Failed to register a workflow `{e}`: {error:?}");
+            }
+            anyhow::bail!("Failed to register a workflow `{e}`");
+        }
+    };
+
+    let wfl_id_str = wfl_ids
+        .iter()
+        .map(|workflow_id| workflow_id.id.to_string())
+        .collect::<Vec<_>>();
+
+    for (wf_id, v) in wfl_id_str.iter().zip(var_props.iter()) {


Die Variablen-Namen werden in einem Monat kryptisch sein. Ist nicht so gebräuchlich, also ausschreiben?

ChristianBeilschmidt · 2026-07-09T06:43:56Z

@@ -0,0 +1 @@
+./backend/rust-toolchain.toml


Coverage am besten > 80% lassen.

…Process, add multi-scenario support, default year range

Resolve conflicts from main's land-use/sealed-area PR (#20): - Keep main's create/ form architecture (delete create-new-auto/) - Port StringArray multi-select support into main's schema-info.ts and inputs-visualizer.component.ts - Keep both ClimateRiskProcess and LandUseSealedAreaProcess registered - Port BoundingBox2D into main's parameters/ module - Fix latent mat-checkbox-in-mat-form-field error for boolean inputs - Rename to_api_workflow -> to_api_vector_process - Re-add Deserialize to DataResource (needed by climate outputs)

The merge commit missed unstaged fixes. Include them: - backend: rename to_api_workflow -> to_api_vector_process, add Deserialize to DataResource - frontend: StringArray field type in schema-info, multi-select in inputs-visualizer, precedence case in create.component, fix mat-checkbox-in-mat-form-field error for boolean inputs - unify BoundingBox2D into BoundingBox (contains/around_point added, climate process uses wfs_string instead of field destructuring)

- Replace year_end with year_range (5-30, default 20) plus optional reference_year_begin (default 2020, null to opt out of anomaly) - Backend computes hex colors for occurrence probability and anomaly cells; TableSchemaField.color_field wires companion columns - Replace fragile chunks_exact(2) pairing with structural tuple batches for analysis/reference WFS queries

…ta floor - Emit ExecuteResults under machine-name scenario keys (rcp26/45/85) while keeping display names in table rows - Log a warning instead of silently dropping anomalies when the reference period yields no data - Enforce DATA_START_YEAR=2006 for both analysis and reference start years

- referenceYearBegin input schema is now nullable (Option<Year>) with default 2006 and min_occurs 0 - Frontend resolves nullable integer schemas to a number field, defaults optional-with-default fields to their default, and sends explicit null instead of dropping it so the backend opt-out is reachable

Mark the rawEnsembleData output with a default-disabled metadata role in the process description, have the frontend leave such outputs unchecked by default, and omit rawEnsembleData from execute results unless requested.

resolve_requests now reports whether rawEnsembleData was requested, so the raw-data gating in execute no longer re-scans the output keys directly.

The OGC API Input enum has no null variant, so sending null for unset optional inputs (e.g. region) returned 404. Omit null values when building the request instead, and make the anomaly reference period opt-out express as omission: referenceYearBegin is now absent=None in the backend while the UI still defaults it to 2006 via the schema default.

The region input's enum lives behind an anyOf/oneOf with a null branch and a $ref into $defs, so enumOptions returned no options and the StringEnum default was an empty string, which the zod enum schema rejected and rendered red. Resolve single-value enums through the null branch and $ref, and default StringEnum inputs to their first enum value.

ChristianBeilschmidt

Ich konnte jetzt erst einmal nur Code-only anschauen. Müssen noch mal wegen der Daten sprechen.

ChristianBeilschmidt · 2026-08-07T14:25:01Z

+    /// Name of a row property that carries the cell color (hex) for this column.
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub color_field: Option<String>,
+    /// Name of a row property that carries a ready-to-display label for this column.
+    /// When set, the frontend renders it instead of the raw value.
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub label_field: Option<String>,


Check: Sind das validate table schema Felder (https://datapackage.org/standard/table-schema/)

ChristianBeilschmidt · 2026-08-07T14:26:10Z

    String,
    Number,
    Integer,
+    Percentage,


Es gibt keinen Percentage-Typ bei TableSchema: https://datapackage.org/standard/table-schema/#field-types

ChristianBeilschmidt · 2026-08-07T14:28:31Z

+
+impl Default for YearRange {
+    fn default() -> Self {
+        YearRange(20)


Ich denke nicht, dass das ein guter Default ist.
Wenn das bei dem einen Prozess sinnvoll ist, dann vielleicht den Default dort definieren.

ChristianBeilschmidt · 2026-08-07T14:29:18Z

    }
 }

+/// Length of a climate window in years (e.g., 5-30 years).


Suggested change

/// Length of a climate window in years (e.g., 5-30 years).

/// Length of a time window in years (e.g., 5 years).

Hat ja nicht immer was mit Klima zu tun und man kann kein von bis angeben.

ChristianBeilschmidt · 2026-08-07T14:32:58Z

+    pub region: Option<CordexRegion>,
+}
+
+#[derive(Deserialize, Serialize, Debug, JsonSchema, ToSchema, Clone)]


brauchst du bei den Ausgaben immer auch Deserialize?

ChristianBeilschmidt · 2026-08-07T15:20:16Z

+                </td>
+              }
+            }
+            @case (ColumnType.Percentage) {


Den Typ gibt es ja nicht in TableSchema

ChristianBeilschmidt · 2026-08-07T15:27:48Z

+                  <mat-chip class="cell-content">
+                    <span
+                      class="color-dot"
+                      [style.background-color]="element[column.colorField]"


Irgendwie fühlt es sich falsch an, eine Spalte mit Farbwerten zu haben.
Evtl. finden wir andere Felder für Metadaten oder müssen eine Extension machen (https://datapackage.org/standard/extensions/).

ChristianBeilschmidt · 2026-08-07T15:28:55Z

+          title:
+            typeof innerValue === 'object' &&
+            innerValue !== null &&
+            'name' in innerValue &&
+            typeof innerValue.name === 'string'
+              ? innerValue.name
+              : this.fieldName(key),


Woh, was ist hier passiert? Evtl. eine Funktion mit Doc?

ChristianBeilschmidt · 2026-08-07T15:31:09Z

+
+/// Calculates climate-risk indicators from CORDEX/CMIP5 climate data for a given point and time window.
+#[derive(Debug, Clone)]
+pub struct ClimateCordex5Process;


Ok, am besten auch noch Modul umbenennen.

jdroenner added 7 commits July 2, 2026 22:32

wip1

0e54078

wip2

418127a

cimate v1

c7b2fb7

update ui again

9c24444

add more models and don't fail on null values

258a3b0

fix workflow

6fc39d5

more descriptions

3991f87

jdroenner added 3 commits July 9, 2026 01:16

fix end_years semantics

910a2f5

add tests

38aeefa

more tests

5f30658

jdroenner changed the title ~~feat: climeate rist from cmip cordex~~ feat: climate risk from cmip cordex Jul 8, 2026

jdroenner added 4 commits July 9, 2026 02:11

lint and fmt

665729a

more tests

9322b61

even more tests

873ede3

format

aed5307

ChristianBeilschmidt requested changes Jul 17, 2026

View reviewed changes

jdroenner added 13 commits July 21, 2026 12:10

refactor: replace strum with hand-rolled enums, rename to ClimateRisk…

5c223d4

…Process, add multi-scenario support, default year range

feat: pre-select all values for string array fields in form

1e7f5ca

Merge branch 'main' of github.com:geo-engine/BioIS into cmip_cordex

d5adf89

remove link to toolchain

48a39a2

rework climate risk process

211bb06

move BoundingBox to global parameters

57dc78e

lint & fmt

ce8875d

update selection of variables

a3b02f4

jdroenner added 5 commits August 4, 2026 12:22

Disable raw ensemble data output by default

ac4919d

Mark the rawEnsembleData output with a default-disabled metadata role in the process description, have the frontend leave such outputs unchecked by default, and omit rawEnsembleData from execute results unless requested.

Fold raw ensemble gating into resolve_requests

d90eb28

resolve_requests now reports whether rawEnsembleData was requested, so the raw-data gating in execute no longer re-scans the output keys directly.

colors in climate risk result

fd08c9b

ChristianBeilschmidt requested changes Aug 7, 2026

View reviewed changes

		@@ -0,0 +1 @@
		./backend/rust-toolchain.toml No newline at end of file

	/// Length of a climate window in years (e.g., 5-30 years).
	/// Length of a time window in years (e.g., 5 years).

Uh oh!

Conversation

jdroenner commented Jul 8, 2026

Uh oh!

coveralls commented Jul 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ChristianBeilschmidt left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ChristianBeilschmidt left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

coveralls commented Jul 8, 2026 •

edited

Loading