status updates api

stac-elevation-bc serves British Columbia’s LidarBC elevation data as a SpatioTemporal Asset Catalog (STAC) — 102,460 tiles on the provincial objectstore, searchable by location and time from the rstac R package, QGIS (v3.42+), or any STAC-compliant client. The endpoint is https://images.a11s.one.

Each item carries up to two assets from the same flight over the same footprint:

asset what it is
dem bare-earth digital elevation model — every item has one
dsm digital surface model, where the delivery published one — 95,888 of 102,460

An item’s geometry is its footprint — the part of the tile that holds data, not the raster’s extent (#2). For tiles named on the BC Geographic System 1:2,500 grid it is the grid cell minus any no-data inside it, so a full tile’s footprint is exactly its cell and neighbours meet without overlapping; for the rest it is the valid data. A search for an area a tile has no data over no longer returns that tile. The dem asset carries raster:bands[0].statistics.valid_percent (the share of the raster holding elevations) and, where the whole file was read, file:checksum (sha256) and file:size. The DSM is assumed to cover what the DEM covers; its own coverage is not measured. Every item from a LidarBC delivery carries lidarbc:delivery ("092/092g/2016"), the key stac-pointcloud-bc items share, so a tile matches its delivery’s point clouds with one filter (#55); the lidarbc: fields are defined in crate.


The catalog refreshes monthly: a scheduled GitHub Actions workflow detects new tiles on the objectstore and appends them incrementally, with each month’s changes landing as a commit to data/ — a git audit trail of what was added, removed, and validated. How the automation works, its failure modes, and where the evidence lives are documented in scripts/, alongside a plain-language guide to the concepts behind the pipeline (COG, STAC, pgstac, validation caching).



Query the collection

Use bcdata to define an area of interest, then query the stac-elevation-bc collection for elevation tiles intersecting it. Below: all DEMs covering the Bulkley River watershed group between 2018 and 2020.

aoi <- bcdata::bcdc_query_geodata("freshwater-atlas-watershed-groups") |>
  bcdata::filter(WATERSHED_GROUP_NAME == "Bulkley River") |>
  bcdata::collect() |>
  sf::st_transform(crs = 4326)

date_start <- "2018-01-01T00:00:00Z"
date_end <- "2020-12-31T00:00:00Z"

# use rstac to query the collection
q <- rstac::stac("https://images.a11s.one/") |>
  rstac::stac_search(
    collections = "stac-elevation-bc",
    intersects = jsonlite::fromJSON(
      geojsonsf::sf_geojson(
        aoi, atomise = TRUE, simplify = FALSE
      ),
      simplifyVector = FALSE
    ) |> (\(x) x$geometry)(),
    datetime = paste0(date_start, "/", date_end)
  ) |>
  rstac::post_request()

# get details of the items
r <- q |>
  rstac::items_fetch()

# burn the results locally so we can serve it instantly on index.html builds
saveRDS(r, "data/stac_result.rds")
r <- readRDS("data/stac_result.rds")

# One row per ASSET, not per item. Every item carries a bare-earth `dem`, and
# most also carry a `dsm` from the same flight -- a dem-only column hid half of
# what the collection serves.
tab <- purrr::map_dfr(r$features, function(f) {
  purrr::imap_dfr(f$assets, function(a, key) {
    tibble::tibble(
      tile     = f$id,
      date     = substr(f$properties$datetime, 1, 10),
      type     = key,
      download = glue::glue('<a href="{a$href}" target="_blank">{basename(a$href)}</a>')
    )
  })
}) |>
  dplyr::arrange(tile, type)


Download links for every asset returned by the query above — one row per asset, so a paired item shows its dem and dsm together. NOTE: To view all columns in the table - please click on one of the sort arrows within column headers before scrolling to the right.


Every row above is a dem, because the Bulkley was flown between 2000 and 2020 and those deliveries published no surface model. Province-wide 95,888 of 102,460 items carry a dsm — the same flight over the same footprint at the same time — but the coverage is concentrated in the 2024 deliveries, so whether you get one depends on when your area was last flown rather than on where it is.

QGIS Data Source Manager (v3.42+)

QGIS 3.42 added native STAC support — connect directly to the catalog and filter by the current map view. See Lutra Consulting’s STAC-in-QGIS blog post for a walk-through.

QGIS draws each item’s bbox, not its geometry. Since the bbox is the bounds of the footprint, a partly empty tile’s box now shrinks to its data, but a tile whose data is a diagonal strip still draws as a rectangle.

Connecting to https://images.a11s.one

Connecting to https://images.a11s.one

Using the field of view in QGIS to filter results

Using the field of view in QGIS to filter results

Sister collections on the same endpoint

The same images.a11s.one STAC API serves several complementary BC collections:

Roadmap

The big one landed in 2026: the catalog is now self-updating (#23) — the goal open since the first build — and the July catch-up grew the collection from 58k to ~98k fully-validated items. Items now also carry the digital surface model alongside the bare-earth DEM (#31), paired on tile id and acquisition date, and each item’s geometry is its footprint — the tile’s grid cell minus its no-data — rather than its raster extent (#2). Still ahead:

Browse open issues for the full backlog.

License

MIT.