Port remaining horizontal tracer advection schemes

pitch-d3a815 · in MCH

Status: Thinking — at the foot of the hill Shaping — halfway up the hill Ready — at the top of the hill In progress — halfway down the hill Done — at the bottom, over the hill Shelved — off the hill Somebody is building this in the cycle it was bet into — an overrun is re-bet, not extended.
Priority: Medium
Appetite: 2.0 person-weeks A budget in person-weeks, not an estimate: what this is worth, not how long it takes.
Tags: icon4py, tracer-advection, miura, ffsl, weno, port
Owner: muellch
Assignees: muellch
Reviewers: jcanton, nfarabullini
Review waived: —
Parent: MCH
Blocked by: —
PRs: C2SM/icon4py#1396, C2SM/icon4py#1470
Cycle: 38 A bet is for one cycle. Work that overruns is re-bet, not extended by default.
Start date: 14.09.2026
End date: —
Scheduled: 2026-09-14 → 2026-10-09 · overruns cycle 38 by 2.6 weeks
Blocks: —
Progress: 0/2 items

Problem

The higher order horizontal tracer advection schemes are missing from icon4py. Tracer advection currently supports first order Miura only (Strassmann), and although the stencils for the higher order schemes are already ported (Burgdorfer) they have not been glued together into working schemes. jcanton and claude made a start on this.

Appetite

One week for an MVP, two for the extensive version.

Solution

Port the remaining horizontal schemes from Fortran, starting with higher order Miura. FFSL is the highest hanging fruit and comes last. Compare against serialized data, and introduce the idealized experiments from the papers on ICON tracer advection, which Strassmann had partly done already.

Rabbit holes

Combining all the Miura schemes into one may have to wait: it could need halo exchanges in icon4py to be generalised first.

No-gos

Multi-tracer issues: multi-tracer efficiency, the configuration at setup time, and their like.

Progress

#1396 Miura Weno scheme (jcanton), open since 27.07.2026
#1470 port of the weno scheme and idealized experiments (muellch), open since 14.09.2026

Both branch from the same 2024 base as dastrm's #580, which is where the convergence work under pitch-370005 also comes from.

## Problem

The higher order horizontal tracer advection schemes are missing from icon4py.
Tracer advection currently supports first order Miura only (Strassmann), and
although the stencils for the higher order schemes are already ported
(Burgdorfer) they have not been glued together into working schemes. jcanton and
claude made a start on this.

## Appetite

One week for an MVP, two for the extensive version.

## Solution

Port the remaining horizontal schemes from Fortran, starting with higher order
Miura. FFSL is the highest hanging fruit and comes last. Compare against
serialized data, and introduce the idealized experiments from the papers on ICON
tracer advection, which Strassmann had partly done already.

## Rabbit holes

Combining all the Miura schemes into one may have to wait: it could need halo
exchanges in icon4py to be generalised first.

## No-gos

Multi-tracer issues: multi-tracer efficiency, the configuration at setup time,
and their like.

## Progress

- [ ] #1396 Miura Weno scheme (jcanton), open since 27.07.2026
- [ ] #1470 port of the weno scheme and idealized experiments (muellch), open
      since 14.09.2026

Both branch from the same 2024 base as dastrm's #580, which is where the
convergence work under pitch-370005 also comes from.