Remote Sensing for Urban Growth Monitoring Training Course

Participants can select suitable satellite data for defined urban monitoring objectives.

• Electromagnetic spectrum properties relevant to vegetation, impervious surfaces, and water
• Spatial, spectral, temporal, and radiometric resolution trade-offs
• Sentinel-2 MSI bands selected for urban land-cover discrimination
• Landsat Collection 2 surface reflectance compared with Sentinel-2 products
• SAR backscatter characteristics for cloud-prone urban monitoring areas
• Sensor selection matrix matched to monitoring objectives and budget

Participants can locate, assess and document imagery and supporting datasets for an urban monitoring project.

• Copernicus Data Space catalogue searches structured by date, cloud, tile
• USGS EarthExplorer queries filtered by collection, processing level, and footprint
• OpenStreetMap, WorldPop, and administrative boundary data assessed for fitness
• Metadata reviewed for acquisition conditions, processing lineage, and licensing
• STAC item fields used to document imagery discovery and access
• Source inventory records datasets, dates, licences, coordinate systems, and limitations

Participants can prepare consistent raster inputs for classification and change analysis.

• Coordinate reference systems selected using UTM zones and local requirements
• Surface reflectance products distinguished from top-of-atmosphere reflectance
• Cloud, cirrus, and shadow masks applied using quality assessment bands
• Raster clipping, mosaicking, resampling, and band stacking performed consistently
• Sentinel-2 Scene Classification Layer used for cloud-screening decisions
• Preprocessing log records parameters, software versions, and source scenes

Participants can interpret urban land-cover features and document image interpretation confidence.

• Visual interpretation keys distinguish roofs, roads, bare soil, and vegetation
• True-colour, false-colour, and SWIR composites selected for target features
• Image enhancement through contrast stretching and histogram inspection
• Ancillary maps and street-level imagery used to resolve ambiguous classes
• Minimum mapping unit defined from sensor resolution and decision needs
• Interpretation confidence documented for mixed pixels and seasonal confusion

Participants can configure a QGIS project for reproducible urban growth analysis and map production.

• QGIS project structure separates raw data, derivatives, layouts, and exports
• GeoPackage layers organised with field aliases, domains, and metadata
• Raster Calculator derives indices with reproducible expressions and naming
• Sampling points created through random, stratified, and systematic designs
• Symbology communicates classification certainty, change direction, and monitoring dates
• Print layouts combine maps, scale bars, legends, sources, and annotations

Participants can use Google Earth Engine to filter, composite and export satellite imagery.

• Google Earth Engine assets, ImageCollections, and geometries organised for projects
• JavaScript scripts filter Sentinel-2 collections by area, date, and cloud
• Median composites reduce cloud contamination while retaining seasonal signal
• Band math calculates indices using Earth Engine image expressions
• Export tasks produce GeoTIFF rasters and tabular zonal statistics
• Script comments and parameter blocks support repeatable monitoring runs

Participants can build a supervised land-cover classification for urban and peri-urban areas.

• Land-cover scheme defines mutually exclusive classes and class definitions
• Training samples split into development and independent validation sets
• Random Forest classification configured with predictor bands and indices
• Class imbalance addressed through stratified sampling and sample review
• Feature selection compares spectral bands, indices, texture, and ancillary data
• Post-classification filtering removes isolated pixels without erasing narrow features

Participants can quantify classification accuracy and diagnose errors affecting urban growth estimates.

• Confusion matrices summarise agreement between mapped and reference classes
• Producer's, user's, and overall accuracy calculated from validation samples
• F1 score reported where class prevalence makes overall accuracy misleading
• Area-adjusted accuracy estimates account for stratified sampling design
• Reference labels verified against high-resolution imagery and field evidence
• Error analysis identifies systematic confusion requiring revised classes or predictors

Participants can detect and quantify urban land-cover changes between observation dates.

• Post-classification comparison quantifies transitions between two mapped dates
• Change matrices isolate conversion from vegetation, agriculture, and water
• Image differencing detects spectral change before thematic classification
• Change thresholds calibrated using stable areas and reference observations
• Registration errors assessed before interpreting boundary-level change patterns
• Change polygons attributed with dates, previous class, new class, and area

Participants can use spectral indices to screen and assess built-up land expansion.

• NDBI, NDVI, and MNDWI calculated for built-up change screening
• Normalized difference indices interpreted alongside seasonal and soil effects
• Index thresholds tested against reference samples rather than assumed
• Impervious surface proxies separated from dry bare land using composites
• Multi-index decision rules document conditions for built-up candidate pixels
• Index maps compared across dates using matched seasonal acquisition windows

Participants can compare pixel-based and object-based methods for complex urban land cover.

• Object-based image analysis segments imagery into meaningful urban land parcels
• Segmentation scale, shape, and compactness parameters tested against block patterns
• Object attributes combine mean spectra, texture, geometry, and context
• Building footprints and road networks refine parcel-level classification results
• Rule-based classification documents thresholds for roads, roofs, and open land
• Pixel and object approaches compared for heterogeneous peri-urban areas

Participants can calculate growth rates, land consumption indicators and spatial pattern metrics.

• Urban expansion rate calculated from built-up area across observation dates
• Annual land consumption rate compared with population growth indicators
• SDG 11.3.1 concepts applied to land consumption and population growth
• Landscape metrics measure patch density, edge density, and fragmentation
• Buffer analysis measures growth near roads, transit, rivers, and protected areas
• Zonal statistics summarise built-up change by ward, district, and municipality

Participants can analyse multi-date imagery to identify persistent urban growth and unusual change.

• Annual composites constructed from matched seasonal Sentinel-2 observations
• Time-series charts reveal acceleration, pauses, and seasonal classification artefacts
• Breakpoint detection distinguishes persistent conversion from temporary disturbance
• Rolling baselines flag unusual built-up expansion for analyst review
• Cloud frequency and missing observations documented within temporal coverage assessments
• Date selection rules protect comparisons from phenology and flood effects

Participants can plan verification, quality assurance and peer review for urban monitoring outputs.

• Field verification protocols specify sites, observations, photographs, and consent
• Mobile data forms capture coordinates, land-cover labels, and evidence
• Positional accuracy checks compare reference geometry with imagery resolution
• Quality assurance checklist reviews inputs, methods, outputs, and documented limitations
• Peer review procedures test reproducibility of maps, code, and calculations
• Uncertainty statements separate observed change, model error, and interpretation limits

Participants can interpret mapped urban growth against planning, infrastructure and hazard constraints.

• Growth hotspots assessed against zoning, infrastructure capacity, and hazard exposure
• Suitability criteria identify conflicts with floodplains, farmland, and protected land
• Scenario maps compare likely expansion under alternative policy assumptions
• Evidence packages distinguish monitoring findings from planning recommendations
• Parcel-level inference avoided when imagery resolution cannot support conclusions
• Planning brief links mapped change to inspection, service, and enforcement priorities

Participants can communicate urban growth evidence clearly to technical and executive audiences.

• Map design applies visual hierarchy, colour conventions, and legible classification labels
• Small-multiple layouts compare baseline, current extent, and detected change
• Area statistics presented with units, dates, uncertainty, and data sources
• Technical methods notes state sensors, processing steps, and accuracy results
• Executive briefing structures findings, implications, caveats, and recommended actions
• Interactive web-map requirements defined for layers, filters, pop-ups, and access

Participants can design an operational procedure for recurring urban growth monitoring.

• Monitoring standard operating procedure assigns triggers, frequency, inputs, and owners
• Data governance matrix records custodians, access rights, retention, and licences
• Version control conventions manage scripts, raster outputs, and published maps
• Processing time and cloud constraints incorporated into delivery schedules
• Escalation criteria route suspected encroachment or hazard exposure for review
• Performance indicators track timeliness, accuracy, coverage, and decision uptake

Participants can apply ethical, privacy and licensing controls to urban monitoring work.

• Privacy risks assessed when combining imagery, addresses, and demographic datasets
• Purpose limitation and data minimisation guide urban monitoring data collection
• Bias checks examine unequal imagery quality and representation across neighbourhoods
• Sensitive-site protocols protect shelters, informal settlements, and critical infrastructure
• Licensing terms govern redistribution of commercial imagery and derivative products
• Disclosure notes communicate uncertainty without overstating automated detection results

Participants can build, review and present a complete validated urban growth monitoring project for a defined decision question.

• Project brief fixes area of interest, monitoring question, and decision users
• Imagery inventory and preprocessing record establish traceable analytical inputs
• Classification or change-detection method produces a documented urban growth map
• Validation evidence quantifies accuracy, uncertainty, and material mapping limitations
• Indicator dashboard summarises growth patterns, hotspots, and planning implications
• Technical report and executive presentation defend methods, findings, and actions