Remote Sensing for Natural Resource Management Training Course

Participants can select appropriate earth observation sources for a defined monitoring requirement.

• Electromagnetic spectrum relationships with vegetation, soil, water and built surfaces
• Spatial, spectral, temporal and radiometric resolution trade-offs examined
• Landsat, Sentinel-1 and Sentinel-2 mission characteristics compared
• Passive optical sensors contrasted with active synthetic aperture radar
• Revisit intervals matched to crop cycles, flood events and disturbance
• Scene footprints, swaths and spatial coverage planned for study areas

Participants can prepare spatial reference data that align correctly with imagery and analysis outputs.

• Geographic coordinates and projected coordinate reference systems distinguished
• UTM zones selected for local natural resource monitoring projects
• Raster cells, vector features and attribute tables linked correctly
• Study boundaries built from concessions, catchments and administrative polygons
• Scale and minimum mapping unit defined before image classification
• Spatial metadata recorded for source, date, resolution and processing status

Participants can find, inspect and document suitable imagery and supporting geospatial datasets.

• Catalog searches filtered by location, acquisition date and cloud cover
• Scene metadata checked for processing level, projection and quality flags
• CEOS Analysis Ready Data principles applied to imagery selection
• Digital elevation models sourced for terrain and hydrological analysis
• Field observations, cadastral boundaries and thematic layers assessed for fitness
• Data licences, access restrictions and attribution requirements documented

Participants can prepare optical satellite imagery for consistent multi-date analysis.

• Top-of-atmosphere and surface reflectance products distinguished before analysis
• Cloud, cirrus and cloud-shadow masks applied using quality bands
• Atmospheric correction purpose assessed across multi-date image comparisons
• Image mosaics assembled with consistent dates, sensors and projection
• No-data areas, scan gaps and saturation artefacts identified
• Visual quality checks completed using true-colour and false-colour composites

Participants can interpret radar imagery and choose suitable applications for cloud-prone areas.

• Backscatter explained through surface roughness, moisture and vegetation structure
• VV and VH polarisation compared for land and vegetation applications
• Speckle reduction balanced against preservation of spatial detail
• Terrain correction applied before geographic measurement and map production
• Radar time series used for flood extent and forest disturbance
• Layover, shadow and incidence-angle effects identified in mountainous terrain

Participants can calculate and interpret indices for vegetation, water, burn and soil conditions.

• NDVI calculated for vegetation greenness and seasonal condition monitoring
• EVI selected where dense canopy reduces NDVI sensitivity
• NDWI and MNDWI compared for open-water delineation
• Normalized Burn Ratio applied to wildfire severity assessment
• Bare Soil Index interpreted alongside field and seasonal context
• Index thresholds calibrated using reference observations and image histograms

Participants can identify natural resource features through systematic visual interpretation.

• Tone, texture, pattern, shape and context used systematically
• False-colour composites configured for vegetation, water and exposed soil
• Image segmentation concepts introduced for object-based land-cover mapping
• Training polygons drawn around internally consistent land-cover examples
• Interpretation keys documented for recurring habitats and land uses
• Ancillary terrain and drainage data used to resolve ambiguous features

Participants can design field-reference data collection for classification and validation.

• Reference data roles separated between training, validation and calibration
• Probability sampling methods compared for unbiased accuracy estimation
• Stratified random samples allocated across rare and common classes
• Field forms specify class definitions, photographs and positional accuracy
• GPS positional error considered against pixel size and boundary complexity
• Seasonal timing aligned between field visits and image acquisition

Participants can produce an interpretable supervised land-cover map from labelled imagery.

• Class schemas designed around management decisions and map scale
• Random Forest classification configured with labelled spectral training data
• Training samples reviewed for class purity, balance and spatial independence
• Predictor bands, indices and terrain variables selected deliberately
• Classification outputs inspected for salt-and-pepper noise and implausible patterns
• Post-classification filtering applied without obscuring meaningful small features

Participants can use unsupervised and rule-based methods where labelled data are limited.

• K-means clustering used to reveal spectral groupings before class labelling
• Cluster outputs merged using field knowledge and image interpretation
• Decision-tree rules combine index thresholds, elevation and proximity variables
• Object-based rules applied to ponds, roads and riparian vegetation
• Rule thresholds tested against independent reference observations
• Method selection justified by data availability, repeatability and management purpose

Participants can quantify map accuracy and communicate limitations to decision-makers.

• Confusion matrices constructed from independent validation sample records
• Producer’s accuracy and user’s accuracy interpreted for each class
• Overall accuracy and balanced accuracy reported with appropriate caveats
• Area-adjusted estimates calculated when sample allocation is disproportionate
• Classification uncertainty mapped to identify unreliable geographic areas
• Map limitations stated without overstating confidence in management conclusions

Participants can detect, verify and quantify meaningful change between observation periods.

• Image dates matched to comparable phenological and hydrological conditions
• Post-classification comparison produces land-cover transition matrices
• Image differencing identifies vegetation loss and surface-water change
• Time-series charts separate persistent trends from short-term anomalies
• Change polygons screened for registration errors, clouds and shadows
• Change rates calculated by area, period and management unit

Participants can assess forest extent, disturbance and restoration signals using remote sensing evidence.

• Forest definitions aligned with policy thresholds for canopy and height
• Canopy-cover indicators mapped from optical imagery and reference plots
• Radar backscatter time series used for disturbance screening
• Forest loss alerts verified using higher-resolution imagery and field evidence
• Restoration plots compared with baseline condition and reference ecosystems
• Fragmentation metrics calculated for core forest, edges and connectivity

Participants can map surface-water dynamics and catchment condition for water management decisions.

• Water extent extracted using spectral indices and radar backscatter
• Seasonal water persistence mapped from multi-date satellite observations
• Wetland boundaries interpreted using inundation patterns and vegetation signatures
• Digital elevation models used for flow direction and catchment delineation
• Riparian buffer condition assessed from vegetation cover and land use
• Sediment plumes and turbidity proxies interpreted with field sampling limits

Participants can derive management indicators for crops, grazing land and exposed soils.

• Crop calendars linked to image dates and phenological signatures
• Vegetation trajectories compared across fields, seasons and rainfall conditions
• Rangeland cover assessed using fractional vegetation and bare-ground indicators
• Grazing pressure interpreted alongside stocking records and field observations
• Erosion-prone areas screened using slope, cover and drainage concentration
• Drought indicators combined from rainfall anomalies and vegetation indices

Participants can monitor disturbance footprints and prioritise follow-up inspections.

• Burned-area mapping combines pre-fire and post-fire spectral indices
• Fire severity classes validated against field observations and recovery patterns
• Mine expansion detected through bare ground, water and vegetation changes
• Linear infrastructure corridors monitored for clearing, erosion and encroachment
• Disturbance alerts prioritised using proximity to sensitive receptors
• Inspection routes planned from change polygons and access constraints

Participants can create habitat-condition evidence while recognising remote sensing inference limits.

• Habitat variables derived from land cover, structure and connectivity
• Species distribution inputs distinguished from direct species observations
• Landscape metrics applied to patch size, isolation and edge effects
• Habitat suitability models documented with assumptions and reference data
• Protected-area pressures screened through roads, settlement and land-cover change
• Biodiversity claims bounded by spatial resolution and ecological uncertainty

Participants can prepare spatial evidence that supports land-use and carbon reporting processes.

• Activity data derived from land-use change maps and area estimates
• Emission factors distinguished from remotely sensed land-cover observations
• IPCC land-use categories aligned with locally defined map classes
• Carbon stock proxies interpreted without substituting for field inventory
• MRV workflows document source data, assumptions and uncertainty
• Environmental reports link maps, statistics, methods and decision implications

Participants can design repeatable remote sensing workflows for routine organisational monitoring.

• Workflow diagrams specify inputs, processing steps, outputs and approvals
• Google Earth Engine scripts organised with parameters and clear comments
• QGIS projects structured with named layers, styles and relative paths
• Version control principles applied to code, datasets and map products
• Quality-assurance checklists cover geometry, dates, units and classification logic
• Monitoring calendars assign acquisition windows, review roles and reporting deadlines

Participants can build, review and present a complete remote sensing monitoring project with defensible recommendations.

• Project scope defines decision question, users, indicators and geographic boundary
• Imagery and reference datasets selected with documented suitability criteria
• Processing workflow produces indices, classifications or detected changes
• Validation evidence quantifies accuracy, uncertainty and outstanding data gaps
• Management maps communicate findings through clear legends and concise annotations
• Recommendations prioritise field verification, operational actions and future monitoring
• Peer review tests reproducibility, interpretation logic and decision relevance