DEM Elevation Correction and Point Sample Report
កែតម្រូវ DEM មួយ ឬច្រើន ធៀបនឹង DEM យោងដែលអ្នកជ្រើស ហើយចេញរបាយការណ៍គំរូចំណុច។
- រកឃើញនៅឯណា
- Processing ▸ Toolbox ▸ KGA Toolbox ▸ KGA Irrigation Tools ▸ DEM Elevation Correction and Point Sample Report
- លេខសម្គាល់ algorithm
- kga:demelevationcorrectionandpointsamplereport
- ប្រភេទឧបករណ៍
- Batch algorithm
ទិដ្ឋភាពទូទៅ
Corrects one or more DEMs against a Reference DEM you choose — your LiDAR, or whatever survey you trust — optionally merges them all into one seamless raster, and builds a Statistics tab and a Visualization tab summarising correction quality and post-correction agreement.
This is the tool for a project whose elevation data arrived from several sources at different dates and different accuracies: a LiDAR block, a drone survey, an older contour-derived grid. Each has its own vertical bias, and stitching them untreated gives steps at every seam.
The Reference DEM is never modified — it is the elevation truth every other DEM is corrected toward. The other DEMs are auto-sorted from finest to coarsest resolution and chained: the finest is corrected directly against the Reference, the next-finest against that already-corrected result, and so on.
The run is resumable via a checkpoint.json in the output folder.
ប៉ារ៉ាម៉ែត្រ
| ឈ្មោះ | ប្រភេទ | តម្លៃលំនាំដើម | ការពិពណ៌នា |
|---|---|---|---|
| Reference DEM (the elevation truth -- never modified) | Raster layer | — | |
| Reference DEM NoData override (blank = use its own NoData) | String, optional | empty | |
| DEMs to correct against the reference (1 or more) | Multiple raster layers | — | Auto-sorted finest to coarsest and corrected in a chain. |
| NoData overrides (comma-separated, matching DEM order) | String, optional | empty | One value per DEM, in the order you selected them. Blank uses each raster's own NoData metadata. |
| Correction method (all stages) | Enum | Auto - Compare All (Best CV) | Auto - Compare All (Best CV), Global Offset, Polynomial Trend (degree 1), Polynomial Trend (degree 2), IDW (distance-decay). Auto tries them all and keeps whichever cross-validates best. |
| IDW power | Number | 2.0 | IDW method only. |
| IDW neighbors (k) | Integer | 12 | IDW method only. |
| IDW full-strength distance (m) | Number | 3000.0 | Beyond this the correction fades out. |
| Cross-validation folds | Integer | 5 | How the method comparison is scored. |
| Calibration subsample stride | Integer | 4 | Every Nth pixel is used to fit the correction. Raise it for a faster, coarser fit. |
| Merge finest DEM + all corrected DEMs | Boolean | True | |
| Merge extent | Enum | Reference DEM extent | Or Coarsest DEM extent (full). |
| Merge resolution (m) | Number | 2.5 | |
| Report sample stride (for Statistics/Visualization tabs) | Integer (min 1) | 20 | |
| Report max sample points | Integer (min 100) | 200000 | |
| Force re-run everything (ignore checkpoint.json) | Boolean | False | |
| Output folder | Folder destination | — |
របៀបប្រើ
- Load the reference DEM and every DEM to correct.
- Open KGA Irrigation Tools > DEM Elevation Correction and Point Sample Report.
- Set the Reference DEM — the one you trust. It is never written to.
- Select the DEMs to correct. Their order does not matter; they are sorted by resolution.
- Leave the method on Auto for the first run and let the cross-validation pick. Look at what it chose in the Statistics tab before pinning a method.
- Press Run.
- Open the Statistics and Visualization tabs in the Processing Results panel — that is where the answer is, not in the log.
- If the run is interrupted, run it again with the same settings to resume.
លទ្ធផល
លទ្ធផលដែលប្រកាស
| ឈ្មោះ | ប្រភេទ | ការពិពណ៌នា |
|---|---|---|
| Merged DEM (if merge was enabled) | File | |
| Run summary | String | |
| Statistics | HTML | Appears as a tab in the Processing Results panel. |
| Visualization | HTML | Likewise. |
អ្វីដែលអ្នកទទួលបាន
In the output folder:
corrected_{name}.tif— one corrected DEM per input.MERGED_FINAL_DEM.tif— the seamless merge, when merging is enabled.statistics_report.htmlandvisualization_report.html— the two report tabs, also on disk.correction_report.json— the machine-readable run record.checkpoint.json— resume state. Safe to delete when finished.
កំណត់សម្គាល់ និងដែនកំណត់
- The chain matters. Each DEM is corrected against the already-corrected one above it, so an error early in the chain propagates down. Check the Statistics tab stage by stage, not just at the end.
- Correction does not create accuracy. It removes systematic bias relative to the reference; it cannot fix a DEM that is simply wrong in detail.
- The reports are sampled, not exhaustive. Report sample stride and Report max sample points bound the cost; the statistics describe that sample.
- Resumable via
checkpoint.json. Tick Force re-run to ignore it. - NoData overrides must match the order you selected the DEMs in. Getting the order wrong silently corrects against the wrong nodata mask.
- For a per-point table across several DEMs, use DEM Point Sample Export to Excel, which is the same sampling in spreadsheet form.