ERDAS IMAGINE Satellite Classification and Raster Analysis Training Course

Course overview

Satellite imagery programmes routinely produce more data than GIS teams can turn into defensible land-cover, change-detection, and suitability outputs. Analysts need to determine whether an observed spectral pattern is a crop, settlement expansion, burned area, water body, or sensor artefact; select an appropriate classification approach; and document accuracy well enough for planning, environmental reporting, or operational decisions. This course addresses that workflow in ERDAS IMAGINE, from preparing multispectral imagery to delivering validated raster products.

Participants work through ERDAS IMAGINE’s raster environment, including layer-stack creation, radiometric and geometric preparation, image enhancement, spectral indices, training-sample design, supervised and unsupervised classification, object-based image analysis concepts, post-classification refinement, and accuracy assessment. They learn to use signature evaluation tools, produce confusion matrices, calculate user’s, producer’s, and overall accuracy, and apply raster modelling to combine classified imagery with elevation, proximity, and constraint layers. The emphasis is on making technically appropriate choices that can be explained to project managers and data users.

Instruction combines guided demonstrations with sustained hands-on work using a realistic satellite-imagery case study. Each participant builds an ERDAS IMAGINE project, prepares imagery, creates a classified land-cover map, tests its accuracy against reference data, and produces a raster analysis model and map-ready outputs. They leave with an exportable project structure, documented processing steps, classification outputs, accuracy-report template, and a practical workflow they can adapt to their organisation’s imagery and reporting standards.

The course is suited to GIS and remote-sensing professionals who already work with spatial datasets and need a reliable ERDAS IMAGINE workflow for analysing optical satellite imagery. It is equally useful for teams moving from visual interpretation or basic raster display to repeatable, quality-controlled image classification.

Course objectives

By the end of this course, participants will be able to:

• Configure an ERDAS IMAGINE project with organised raster inputs, projections, metadata, and output directories
• Prepare multispectral imagery using layer stacking, subsetting, resampling, and radiometric enhancement methods
• Calculate NDVI, NDWI, band ratios, and other spectral indices using ERDAS IMAGINE raster functions
• Create and evaluate training samples and spectral signatures for supervised land-cover classification
• Run unsupervised and supervised classifications using ISODATA and maximum-likelihood methods
• Refine classified rasters with recoding, clumping, sieving, majority filtering, and ancillary GIS layers
• Produce confusion matrices and calculate overall, producer’s, and user’s accuracy from reference samples
• Build a documented raster model and deliver map-ready classified imagery with supporting accuracy evidence

Benefits of attending

Target competencies

Who should attend

• Remote Sensing Analysts — who need to turn satellite scenes into validated land-cover and change-analysis products
• GIS Analysts — who support planning, environmental, infrastructure, or asset teams with raster-based evidence
• Geospatial Data Scientists — who need a practical desktop workflow for preparing labelled raster inputs and assessing results
• Environmental Monitoring Officers — who track vegetation condition, water extent, habitat change, or disturbance from imagery
• Land-Use Planning Specialists — who require defensible classifications to measure development, settlement growth, and land conversion
• Survey and Mapping Professionals — who integrate satellite-derived layers with vector mapping, elevation, and field-reference data

Requirements and prerequisites

Participants should be comfortable working with GIS data: opening raster and vector layers, understanding coordinate reference systems, navigating map displays, and managing project folders. Prior experience with ArcGIS Pro, QGIS, or another desktop GIS is useful, as is a basic understanding of satellite bands, pixels, and attribute tables. Familiarity with ERDAS IMAGINE is helpful but not required; the course introduces its interface and core raster workflow. Programming, machine learning, advanced statistics, and prior image-classification experience are not required. Participants should be prepared to interpret simple accuracy tables and compare analytical outputs.

Training methodology

The instructor demonstrates each ERDAS IMAGINE workflow on satellite imagery before participants repeat it in individual guided exercises. Short technical briefings explain the decisions behind band selection, training data, classifier choice, and validation rather than treating buttons as a fixed recipe. A continuing land-cover case study supplies imagery, reference points, and ancillary layers for hands-on analysis. Participants compare classification results in small groups, troubleshoot common errors with the instructor, and finish by adapting a raster workflow and delivery checklist for a current or anticipated workplace use case.

Course outline

Day 1: ERDAS IMAGINE raster foundations and image preparation

• ERDAS IMAGINE interface, viewers, project organisation, and raster metadata inspection
• Satellite sensor characteristics for Landsat and Sentinel-2 multispectral imagery
• Raster coordinate systems, map projections, pixel size, and spatial alignment
• Layer-stack creation and multiband image subsetting
• Radiometric display enhancement using histogram stretch and contrast tools
• Band combinations for vegetation, water, built-up land, and false-colour interpretation
• Resampling methods and their effect on analytical raster outputs

Workshop: Participants create an organised ERDAS IMAGINE project, layer-stack and subset a multispectral scene, and prepare enhanced image views for interpretation.

Day 2: Spectral analysis and classification design

• Spectral response patterns and separability of common land-cover classes
• Interactive feature-space analysis and pixel-value interrogation
• Spectral index calculation with NDVI, NDWI, and normalised difference built-up index methods
• Training-sample design using regions of interest and reference data
• Signature editor functions, class naming, and signature-file management
• Signature statistics and separability measures for training-data quality control
• Unsupervised ISODATA clustering and cluster-labelling strategies

Workshop: Participants calculate spectral indices, digitise training regions for a defined class schema, and assess whether their signatures are sufficiently separable.

Day 3: Supervised classification and class refinement

• Classification scheme design for operational land-cover mapping
• Maximum-likelihood classification parameters and probability thresholds
• Supervised classification execution from validated signature sets
• Comparing supervised and unsupervised classification outputs
• Class recoding and thematic aggregation for reporting requirements
• Clump, sieve, and majority-filter operations for salt-and-pepper reduction
• Ancillary-layer integration using elevation, slope, masks, and proximity constraints

Workshop: Participants run a maximum-likelihood classification, compare it with ISODATA results, and produce a refined thematic land-cover raster.

Day 4: Validation, change analysis, and raster modelling

• Reference-sample selection and independent validation-set design
• Accuracy assessment setup using classified maps and ground-reference points
• Confusion matrices, overall accuracy, producer’s accuracy, and user’s accuracy
• Interpreting omission and commission errors for class-improvement decisions
• Post-classification comparison for land-cover change detection
• Raster algebra and conditional functions in ERDAS IMAGINE Spatial Modeler
• Suitability modelling with weighted criteria, exclusion masks, and reclassification

Workshop: Participants validate their classification with reference points, produce a confusion matrix, and build a simple suitability model from classified and ancillary rasters.

Day 5: Production workflows and workplace application

• Quality assurance checks for source imagery, class labels, projections, and raster outputs
• Map composition and thematic symbology for classified raster communication
• Raster export settings for GeoTIFF, IMG, and GIS interoperability
• Metadata and processing-log requirements for reproducible image analysis
• Batch processing concepts and reusable model design in Spatial Modeler
• Common classification failure modes and structured troubleshooting methods
• Operational workflow planning for monitoring, reporting, and stakeholder review

Workshop: Participants complete and present a documented mini-project containing a classified map, accuracy report, raster model, export package, and workplace implementation plan.

Tools & standards covered

ERDAS IMAGINE, ERDAS IMAGINE Spatial Modeler, ArcGIS Pro, GDAL

A typical training day

Live online deliveries follow the same structure in the East Africa Time zone, with shorter screen blocks and longer breaks.

What the fee includes

How you can take this course

Certification

Participants who complete the full five days receive the Skillset Development Certificate of Completion, stating the course title, course code, dates and delivery format — suitable for professional-development records and employer reimbursement.

Frequently asked questions