Combine Two Terrains¶
Watch this step (1:55), and the merge itself at 3:25.
Your survey tile covers 2.5 km. A flood model usually wants more: somewhere for water to come from and somewhere for it to go. This step adds free global terrain around the survey, deals with the fact that the two measure height from different surfaces, and merges them into one continuous DEM.
Add free global terrain¶
- Open the Inputs panel and find the Terrain section
- Click the globe icon, tooltip Download Global DEM
- The Import terrain from Copernicus GLO-30 panel opens. Give it a Title,
or keep the default
Copernicus GLO-30 DEM - Click Define import area, then click one corner on the map and the opposite corner
- Hydrata shows Review your terrain selection with the selected area, the estimated number of cells at 30 m resolution, and an estimate of how long the import will take. Click Download this terrain
Draw an area of a few kilometres around the tutorial tile. Import areas are capped at 40,000 km2, about 200 km by 200 km; a larger study imports several tiles.
GLO-30 arrives at about 30 m resolution, which is coarse next to 1 m LiDAR. That is exactly what you want here: the survey carries the detail, GLO-30 carries the context.
Two DEMs, two vertical datums¶
The two terrains do not agree about where zero is, and the disagreement is much larger than the flooding you are modelling.
- Copernicus GLO-30 reports orthometric heights on EGM2008, roughly heights above mean sea level. This is Hydrata's canonical vertical frame.
- The tutorial LiDAR reports ellipsoidal heights, straight from GNSS, measured from the WGS 84 ellipsoid.
At Dar es Salaam the geoid sits about 27.6 m below the ellipsoid, so the same ground reads about 27.6 m lower in the LiDAR than in GLO-30. Merge the two without dealing with that and you get a silent 27.6 m cliff at the seam.
The Check vertical datum badge¶
Hydrata measures this for you at import. It compares each new terrain against GLO-30 over stable, flat, bare cells and classifies the result, and it is deliberately quiet when there is nothing to say:
| What Hydrata concluded | What you see on the terrain row |
|---|---|
| Ellipsoidal heights | A Check vertical datum badge, opened, with the evidence and three actions |
| EGM2008, confidently | A quiet Datum: EGM2008 tick |
| EGM2008, but not confidently | The same Check vertical datum badge |
| No conclusion, and nothing contradictory | Nothing. Hydrata does not guess |
| The file's own tag disagrees with the measurement | Datum tag disagrees with measurement, the loudest state there is |
The badge fires on the tutorial tile, and that is correct. The tile really is ellipsoidal, so expect to see it and read it as the check working, not as a problem with the file or with your upload.
The badge names its evidence: the datum it guessed, and how far the terrain reads from GLO-30 in metres. On the tutorial tile that figure is around 26 m.
Convert to EGM2008¶
The badge offers three buttons:
- Convert to EGM2008 creates a new terrain, titled with
(EGM2008)appended, holding the same survey shifted into the project's vertical frame. Your original file is never modified, but once the converted copy is ready the original drops out of the Terrain list and the copy takes its place, so you do not end up with two versions of the same survey to choose between. Watch the Tasks panel for the new row to appear. This button only appears when Hydrata thinks the heights are ellipsoidal, because converting something already on EGM2008 would double-apply the shift. - Keep as-is dismisses the advisory and leaves the terrain alone.
- It's already correct dismisses it because you know better than the measurement.
Click Convert to EGM2008 for the tutorial tile, and use the converted terrain from here on. A model built on the ellipsoidal tile alone would still be valid, because the file is internally consistent. It stops being valid the moment you mix it with GLO-30.
The conversion is a shift, not a reprojection
Reprojecting changes the horizontal coordinate system. This changes the vertical reference only, by adding the local geoid separation, and it is reversible. The original raster is left exactly as uploaded.
Merge them into one surface¶
- In the Terrain section, click the Combine Surface icon
- The Combined surface panel opens with the Merge terrains list, labelled top = highest priority, bottom = base
- Add both terrains with + Add DEM to stack
Each row shows its own native resolution, so which one is the coarse one is answerable without leaving the panel, and a badge showing its place in the stack: TOP, BASE, or a number counting down from the top.
Order matters, and finer goes on top¶
The merge seeds from the bottom entry and pastes each higher entry over it inside that entry's own footprint. So the entry on top wins wherever it has data.
Put the converted 1 m LiDAR on top and GLO-30 at the base. Getting this backwards is the one mistake that looks like it worked: GLO-30 covers the whole grid, so a GLO-30 tile sitting on top of the survey replaces every pixel of it, and you get a 30 m surface with a 1 m file name.
Hydrata inserts a newly added DEM above the first entry it should outrank, so the default order is usually already finest-on-top. Use the up arrow on a row to move it, and the cross to remove it.
Hydrata also refuses to merge a stack where a coarser DEM sits above a finer one. The confirmation dialog names the offending pair with both resolutions, explains that this is legitimate for one case only, a design surface deliberately placed over a survey, and relabels its confirm button Derive anyway. It is an acknowledgement, never a hard block, so you can still do it on purpose.
The pencil on each row¶
The pencil toggle on a row says whether the feather merge may adjust that terrain's heights:
- Greyed pencil, unmodified: paste this terrain's elevations exactly, and never shift them. The terrain becomes a datum anchor.
- Green pencil, modifiable: the merge may reconcile this terrain onto the one below it with a vertical offset, so there is no step at the seam.
The base entry is always modifiable and its pencil is locked on, because something has to absorb the reconciliation. At least one entry must be modifiable, and Hydrata says so if you try to make them all anchors.
Every DEM you add starts with a green, modifiable pencil. For the tutorial, click the pencil on the 1 m survey's row so it turns grey: that makes the survey the anchor, with GLO-30 at the base absorbing the offset.
Settings and Create¶
| Field | What it does |
|---|---|
| Feather width (m) | How far outward from a higher terrain's edge the difference between the two surfaces fades to zero. Wider is smoother |
| Target resolution (m) | The output grid spacing. Set 1 for the tutorial, to keep the survey's detail |
| Set extent | Draw one rectangle to limit the output grid. Without it, the output covers the union of the inputs |
Click Create. A Confirm derive dialog shows the estimated output size and, if the stack has a coarser-above-finer pair, the warning above. Outputs above 10 GB are refused, and the dialog tells you to reduce the extent or coarsen the target resolution.
The extent is the size lever
A 1 m target resolution over a whole GLO-30 tile is an enormous raster. Set an extent a little wider than the area you will actually model.
What you get¶
The merge produces a new terrain, titled with (derived) appended, and it is
an ordinary terrain row: you select it as a Scenario's Terrain exactly like any
other. Neither input is modified.
Inside its own footprint the higher terrain is untouched. Outside it, the merge interpolates the difference between the two surfaces, full at the seam and fading to zero outward, and adds that to the surface below. The result is continuous rather than blended, so the survey's detail survives.
The merge measures the largest remaining step at each seam and the vertical offset it applied, and stores both with the derived terrain, but it does not show them on screen. What you can read is the merge's log in the Tasks panel and the surface itself: a visible cliff along the survey's edge means the datums still disagree. Go back and convert.
The merge warns about datums, it does not fix them
If the recipe still mixes an ellipsoidal terrain with an EGM2008 one, Hydrata tags each input with the datum it inferred and records a mismatch warning on the merge, then carries on. The warning is not blocking and nothing is converted automatically: converting a terrain is always something you ask for. Read the merge's log in the Tasks panel if you are not sure.
Next steps¶
Select the derived terrain from here on. Next: define boundaries around the area you want to model.