# Getting real elevation + soil for the Judean flank

## Why you have to fetch it (I can't, from here)
This environment's fetch tool only reaches **data.gov.il**, and only returns
JSON/text — not binary rasters. External DEM/soil APIs (OpenTopoData,
Open-Meteo, SoilGrids) time out, no GIS connector exists, and no browser is
connected. So a high-res DEM raster can't be streamed in. But `ingest_dem.py`
is built and tested — drop a GeoTIFF into a connected folder and it runs the
full pipeline (elevation, hillshade, slope, rivers, erosion, flyover) at true
resolution over hundreds of square miles.

## Recommended clip — Ein Auja / eastern Judean flank
Ties directly to the spring we already calibrated, with dramatic relief
(watershed crest ~+900 m → Jordan/Dead-Sea rift ~ −300 m).

    bounding box (WGS84):   west 35.15, south 31.70, east 35.55, north 32.00
    ≈ 38 x 33 km  ≈ 1,250 km²  ≈ 480 sq mi
    at 30 m  →  ~1300 x 1100 px

## Where to get it (free)
1. **Copernicus GLO-30** (30 m, best global DEM) — OpenTopography portal
   → https://portal.opentopography.org/raster?opentopoID=OTCOP.032021.4326.1
   Draw/enter the bbox above, output **GeoTIFF**, download. (Free login.)
2. **SRTM GL1** (30 m, reproducible baseline) — OpenTopography
   → https://portal.opentopography.org/raster?opentopoID=OTSRTM.082015.4326.1
3. **Soil composition** — SoilGrids (250 m): https://soilgrids.org
   Export **clay / sand / silt (0–5 cm), mean** as GeoTIFF for the same bbox.

## Then, once the file is in a connected folder
    python3 ingest_dem.py cop_glo30_auja.tif --erode --flyover
    python3 ingest_dem.py cop_glo30_auja.tif --soil soilgrids_clay.tif

Outputs: `dem_01_elevation_rivers.png`, `dem_02_hillshade.png`,
`dem_03_slope.png`, `dem_04_soil.png`, `dem_05_eroded_rivers.png`,
and `baked_dem*.npz` (feed to `flyover.py` for the cinematic pass).

## Alternative live path (lower res)
If you connect the Claude-in-Chrome extension, I can pull an elevation grid
live through your browser via a JSON elevation API — but point-query limits cap
it at a coarse grid (~hundreds of metres), not 30 m. A dropped GeoTIFF is far
higher quality.