jiayuasu commented on PR #2896:
URL: https://github.com/apache/sedona/pull/2896#issuecomment-4366828217
Pushed `16222d28b8`.
| Review point | Action |
|---|---|
| Section 3 prose claims multi-gigabyte support but the code drops `x`/`y`
and uses scalar subqueries | The `selectExpr` after the `raster` data source
now keeps `x` and `y` (tile-index columns). |
| Two-raster `RS_MapAlgebra` will fail when either scene has more than one
tile | The ΔNDVI step is rewritten as a join: `(scene='scene_after')` joined to
`(scene='scene_before')` on `(a.x = b.x AND a.y = b.y)`, with the row-wise
two-raster `RS_MapAlgebra` over the joined tiles. Same SQL works for
single-tile (1 row) and multi-tile (N rows) inputs. |
| `RS_ZonalStats` cross-join generalizes badly to multi-tile inputs |
Section 6 prose now calls out the assumption: with a single-tile delta the
cross-join collapses to one row per parcel; for tiled inputs, swap `'mean'` for
`'sum'` + `'count'` per `(parcel, tile)` and aggregate `SUM(sum) / SUM(count)`
per parcel — same idiom, one extra `GROUP BY`. |
| `mpatches` imported but unused; viz claims parcel boundaries that aren't
drawn | The ΔNDVI panel now draws the 4×4 parcel grid using
`mpatches.Rectangle` and labels each cell `P{ix}{iy}` so the top-greening
parcel (P10) is visually identifiable in the figure. |
Re-verified end-to-end through the local mirror of
`docker/test-notebooks.sh` after every edit:
```
PASS 02-vegetation-change 17s elapsed
```
Output unchanged: top-greening parcel `P10`, COG round-trip 65×65
REAL_64BITS, ranking table identical to the pre-fix run.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
