+91 98765 43210 | info@asav.com | Mon – Sat, 09:00 – 19:00 IST
Noida, Uttar Pradesh 201301
  1. Home
  2. Services
  3. GIS
  4. Land Use / Land Cover
GIS

Land Use / Land Cover

Classification and change detection for planning, environmental and agricultural programmes.

Land Use / Land Cover — representative deliverable
Overview

Land Use / Land Cover

Land-use and land-cover datasets get used for decisions with consequences — planning consent, subsidy allocation, environmental compliance, flood modelling. Which means the classification scheme, the minimum mapping unit and the accuracy assessment matter as much as the map itself.

We produce LULC classification from satellite and aerial imagery using whichever method the data and the scheme support: supervised classification, object-based segmentation, or manual interpretation where class definitions are too subtle for an automated approach to be trusted. Most real programmes end up using a combination.

Every delivery includes a proper accuracy assessment — a confusion matrix against independent reference samples, with producer and user accuracy per class and a kappa statistic. If one class is performing badly, you will see it in the numbers rather than discovering it in use.

What you receive

Every item below is issued in your own template, with your layer conventions, title block and revision scheme.

  • Classification to your scheme, or to CORINE, NLCD, IPCC or Anderson
  • Imagery pre-processing — radiometric and atmospheric correction, mosaicking
  • Training and validation sample design with spatial stratification
  • Classified raster and vectorised polygon output
  • Minimum mapping unit enforcement and generalisation
  • Accuracy assessment with confusion matrix and kappa per class
  • Multi-date change detection with change matrix and area statistics
  • Class area statistics by administrative or catchment boundary
  • Cartographic output and metadata to ISO 19115

Production detail

Key production parameters for Land Use / Land Cover
Typical turnaround3–8 weeks by area and class count
Imagery sourcesSentinel-2, Landsat, Planet, WorldView, aerial, UAV
Class schemesCustom, CORINE, NLCD, IPCC, Anderson Level I–III
Typical accuracy85–95% overall, class-dependent, reported honestly
Minimum mapping unitPer specification, from 0.01 ha upward
Output formatsGeoTIFF, GDB, SHP, GeoPackage, XLSX statistics, PDF maps
RevisionsIncluded until the accuracy target is met or shown unachievable

Software and formats

We work in your toolchain. If something you use is not listed, ask — the list below is only what we use most.

ArcGIS ProERDAS IMAGINEQGISGoogle Earth EngineeCognitionPython / scikit-learnGDALSNAP

Frequently asked

Overall accuracy of 85–95% is a realistic expectation, but the figure that matters is per class. Spectrally distinct classes like water and dense urban perform very well; distinguishing similar crop types or degraded scrub from grassland is much harder. We report per-class accuracy so you know which parts of the map to trust.

Yes, with the important caveat that classifications must be methodologically consistent to be comparable. We reprocess historic epochs on the same method rather than differencing two datasets produced differently, which otherwise measures method change as if it were land-use change.

Whichever the class scheme justifies, usually both. Automated classification handles the spectrally separable classes efficiently; classes defined by use rather than cover almost always need interpretation. We state the method per class.

Sentinel-2 and Landsat are free and adequate for many schemes. Finer minimum mapping units need commercial imagery, and we will advise on and help procure the most cost-effective source for your specification rather than defaulting to the most expensive one.
Step 4 of 4 in GIS

Send one land use / land cover job and see how we work

Scope, price and programme come back in writing before anything starts. Most first projects are a single deliverable so you can judge us on output, not on a pitch.

Call an engineer