Readable geoprocessing for everyone

Spatial analysis a non-programmer can write, reproduce, and share.

Easy wins every time

Decision-maker & analyst briefing

The opportunity

Spatial questions are everywhere

The people who need the answers — analysts, scientists, planners — mostly aren't programmers.

Yet to automate the work — to make it repeatable and shareable — today's tools demand that they write code. So the work waits, or doesn't happen.

niva · Easy wins every time

The problem

Automating QGIS means writing PyQGIS

To script even simple geoprocessing today, you must:

  • Initialize a QgsApplication — boilerplate before any work begins
  • Memorize algorithm IDs — native:buffer, native:dissolve, native:clip
  • Build ALL_CAPS parameter dicts — and juggle TEMPORARY_OUTPUT
  • Thread each output into the next — by hand, step after step

That's a programming task — out of reach for the GUI-first analysts who need it most, and tedious even for those who can.

niva · Easy wins every time

The shift

Automation shouldn't require a programmer

niva · Easy wins every time

The product

Meet niva

A concise, readable text-pipeline grammar for QGIS geoprocessing — for people who don't want to write PyQGIS.

load roads.gpkg | buffer 100 dissolve | clip city.gpkg | save roads_local.gpkg

A whole pipeline on one line — that a non-programmer can write and read — running on QGIS's own Processing algorithms underneath.

Easy wins every time.

niva · Easy wins every time

Audience

Who niva is for

Decision-makers

Faster answers without hiring scarce GIS programmers. Work that's auditable, reproducible, and shareable across the org.

Data scientists
(non-programmers)

Already use Marimo for repeatable, shareable analysis. Now add geoprocessing — in the same readable, reactive workflow.

Downstream consumers

Explore the results in QGIS, in Marimo, or as static files — whichever fits, no GIS expertise required.

niva · Easy wins every time

Marimo-native

It fits the workflow you already have

Marimo is how your team makes analysis repeatable and shareable. niva runs right inside it:

  • Write niva in a Marimo cell — beside your Python and charts
  • Reactive — change an input, the map and tables update live
  • Repeatable — the notebook is the analysis; re-run any time, same result
  • Shareable — hand someone the notebook, or export it; no setup ritual

No new tool to adopt — niva meets your people where they already work.

niva · Easy wins every time

Flexible by design

One grammar, many surfaces

niva · Easy wins every time

Trust the result

Reproducible & shareable — by design

The value decision-makers care about most: results you can re-run, audit, and defend.

  • Pipelines are the record — the one-line flow (or YAML) is the documentation
  • Provenance as a byproduct — every step auto-recorded as lineage metadata
  • Assess your inputs — profile incoming data for quality before you trust it
  • Version-controlled flows — diff and review analysis like any other text
niva · Easy wins every time

For every stakeholder

Produce once — explore it anywhere

niva · Easy wins every time

Without niva

The workflow today

niva · Easy wins every time

With niva

The workflow with niva

niva · Easy wins every time

Thin, not limiting

All of QGIS — in plain language

878
Processing algorithms
406
expression functions
SQL
SpatiaLite & PostGIS

niva is a thin wrapper — friendly verbs mapped onto QGIS's own algorithms, not a re-implementation of GIS. The full power of the world's leading open-source GIS, in one readable line.

niva · Easy wins every time

Beyond the vector layer

Every data type QGIS reads

Point clouds

✓ each/show/catalog see LAS · LAZ · COPC · E57 · …
✓ Friendly dtm · dsm · hag — raw LiDAR → raster

Raster · mesh · vector

✓ Discovery spans all GDAL formats + MDAL mesh
✓ One inventory: catalog data/ to=report.md

A folder of LiDAR tiles is just another pipeline input:

each "tiles/*.las" | dtm resolution=1 | save "dtm/{name}.tif"
niva · Easy wins every time

A real study · end to end

From a point cloud to a watershed

One bare-earth DTM → a full hydrologic stack, in plain lines:

each "*.las" | dtm resolution=1 | save "dtm/{name}.tif"
run gdal:merge INPUT="dtm/*.tif" OUTPUT=dtm.tif
run grass:r.watershed elevation=dtm.tif accumulation=accum.tif drainage=dir.tif basin=basin.tif
run grass:r.to.vect input=basin.tif type=2 output=catchments.gpkg
run saga:ta_channels:5 DEM=dtm.tif ORDER=strahler.tif SEGMENTS=streams.gpkg BASINS=basins.gpkg

Flow direction · flow accumulation · vector catchments · Strahler-ordered streams — GRASS & SAGA, one grammar.

niva · Easy wins every time

Fuse anything

Terrain, imagery & features — together

Multispectral

✓ Landsat · Sentinel-2 · NAIP → NDVI
✓ Zonal stats onto parcels — greenness per lot

LiDAR → vector

✓ Contours · slope zones from the DTM
✓ Elevation & canopy per building footprint

Rasters become lines and polygons; every vector product lands in one GeoPackage.

niva · Easy wins every time

The tailwinds

Why niva, why now

  • QGIS is the de-facto open GIS — ubiquitous, trusted, free; the foundation we build on
  • Reproducible analysis is the norm — Marimo and notebooks are how modern teams work
  • The automation gap is real & unserved — GUI analysts still can't automate without code
  • Open and clean-room — GPLv3, built on PyQGIS; no proprietary lock-in to fear
niva · Easy wins every time

The value

What niva delivers

Speed

Answers the same day — no queue behind a programmer.

Capacity

Existing analysts automate their own work.

Trust

Reproducible, provenance-tracked, auditable results.

Reach

Share to QGIS, Marimo, or static files — for anyone.

Lower cost

Less custom scripting, less specialized hiring.

No lock-in

Open source, on open foundations.

niva · Easy wins every time

Status · built and shipping

What's working today

niva ships as a QGIS plugin (v0.62) and on PyPI, and runs real analysis end-to-end on QGIS's own algorithms:

Grammar & engine

✓ Readable lexer + parser → pipeline stages
✓ Pipe-chaining engine: layer handles, lineage
✓ PyQGIS backend — runs real geoprocessing

Verbs & algorithms

✓ 48 alias verbs + built-ins (vector · raster · point cloud)
✓ sql @conn — SELECT → layer, and server-side writes
✓ run — reach ANY of QGIS's 878 algorithms

Every data type

✓ Vector · raster · point cloud · mesh discovery
✓ LiDAR: dtm/dsm/hag · GRASS + SAGA hydrology
✓ Landsat / Sentinel-2 / NAIP fusion → NDVI

Data, cartography & delivery

✓ PostGIS & SpatiaLite — read · write · analyse
✓ project / style · figure / map · show / catalog
✓ Plugin · CLI · Python · LSP · 660+ tests · full docs

niva · Easy wins every time

Roadmap · shipped & next

The road ahead

  • v0.1 – 0.62 — Shipped · available now, on PyPI
    Grammar · 48 verbs + raster · run → 878 algorithms · point clouds + mesh + all-types discovery · GRASS/SAGA hydrology · multispectral fusion · PostGIS read·write·analyse · project / style · figure/map · LSP · QGIS plugin + CLI + Python · provenance · full docs
  • v1.0 — Stable release
    Grammar freeze (SemVer) · worked Marimo–QGIS integration
  • v2.0 — Power features
    Named intermediates & variables · SQL-driven quality rules & constraints
  • v2.x — Service mode
    Service / daemon mode · richer layout & symbology export
niva · Easy wins every time

Let your analysts do the analysis.

Bring readable, reproducible geoprocessing to the people who need it.

Easy wins every time.

github.com/johnzastrow/niva  ·  Let's run a pilot.