Changelog

The official Jiskta SDKs and example notebooks are now open source and available on PyPI and npm. The Python SDK supports DataFrames, wind derivation, area queries, exceedance, and trend analysis. The Node.js SDK has full TypeScript types and zero runtime dependencies. Jupyter notebook examples covering air quality trends, ERA5 climate, exceedance mapping, and a Node.js interactive map explorer are available in the examples repo.

CAMS Global (0.75°, global) now has near-real-time coverage for the last ~6 months via the ADS forecast dataset at leadtime_hour=0 (model analysis). This fills the gap between the validated EAC4 reanalysis (18-month lag) and today. Shown as purple cells on the coverage calendar.

Query by city or region name instead of a bounding box. area=paris, area=new-york, area=osm:54094 — all resolved from a global OSM-derived index of ~400K administrative areas with point-in-polygon filtering. Discover areas via GET /api/v1/areas?q=name.

/api/v1/link now supports global admin boundaries (not just EU NUTS3), CSV output format, per-year time series aggregation, and World Bank socioeconomic sources as join targets. Correlate climate data with GDP, population, or mortality at country and subnational level.

NASA Black Marble VNP46A3 monthly composite radiance is now available as variables=viirs. 0.005° (~500 m) resolution, global land coverage, 2020–present. Combine with CAMS air quality or ERA5 meteorology in a single query for environmental-economic analysis.

Public beta open. Sign up, get an API key, and start querying Copernicus CAMS air quality, ERA5 meteorological data, and NASA VIIRS nighttime lights via a single REST endpoint. Pay-as-you-go credit packs available from €10. New accounts receive 500 free trial credits.

Five ERA5 variables (2m temperature, boundary layer height, total precipitation, U/V wind components) now available via the same /api/v1/climate/query endpoint. Combine CAMS air quality with ERA5 meteorology in a single request using variables=no2,t2m,blh.

Rewrote the storage format from chunk-major to point-major layout. Combined with io_uring O_DIRECT reads, small-bbox queries (city/site monitoring) are now 34× faster on cold NVMe — from ~1,400ms to ~40ms for a 99-tile Europe-wide query. Full-tile E2E throughput: 130 req/s sustained at zero errors.

Aggregate raster climate data to NUTS3 administrative regions in a single API call. Compute means, sums, Pearson correlations, and top-N rankings across EU regions. Enables direct integration with Eurostat socioeconomic data at NUTS3 level.

Wide-bbox stats queries (format=stats over large areas) are now ~7× faster via AVX2 vectorised accumulation. Daily and monthly CSV aggregations replaced std::map with contiguous flat arrays — eliminating 36,500 per-query heap allocations. Peak throughput hit 19,715 req/s for stats mode.

Initial ingestion of the full Copernicus CAMS reanalysis for NO₂, PM2.5, PM10, and O₃ covering Europe at 0.1° resolution from January 2013 to the present. ~532 GB of annual tile files on NVMe RAID1. Daily auto-sync — new months are picked up as soon as ECMWF publishes them.