Service overview
Geology, geophysics, GIS, remote sensing, hydrology, seismology and hazard modelling. This service is part of that division and is scoped individually around what you need to learn, decide or publish.
Geographic information systems organise data by location and let researchers combine layers that were collected for different purposes: elevation, land cover, soils, population, infrastructure. Remote sensing adds repeated, wide-area observations from satellites and aircraft. Together they support studies of land-use change, flooding, urban growth, vegetation, hazards and infrastructure planning.
The convenience of maps can hide problems. Layers differ in scale, coordinate system and accuracy, classification schemes change between years, and the same pixel may mix several land covers. Maps produced by classification contain errors that should be measured, not assumed away. This service supports spatial analysis and image analysis with attention to data preparation, accuracy assessment and appropriate use of spatial statistics.
If you plan to publish the outcome, see manuscript & publication support, for example computational & simulation manuscripts or original research article writing.
What this service covers
Problems this service can address
- Selecting suitable imagery and processing levels
- Classification accuracy and validation
- Integrating spatial layers consistently
Methods and analytical approaches
- Spatial data management and projection handling
- Image classification and change detection
- Spatial statistics and overlay analysis
- Accuracy assessment
Applications
- Land-use and land-cover studies
- Flood and hazard mapping research
- Environmental monitoring
Relevant engineering disciplines
Scope, inputs and deliverables
Scope of work
- Definition of the technical question, system boundary and success criteria
- Analysis, modelling or data work appropriate to your inputs
- Results with stated assumptions and limits of validity
- Documentation in a form suitable for reports or publication
- Feasibility check on: study area and objectives
Typical inputs required from you
- Study area and objectives
- Imagery or spatial data sources
- Reference or ground-truth data
Deliverables
- Maps and spatial analysis results
- Accuracy-assessment report
- Methods text
Highly specialised requests are first checked for feasibility from this information. Please do not send confidential data through the public form; describe it first and agree secure transfer afterwards.
Project workflow and method
- 01Enquiry and requirement assessment
- 02Technical scoping and proposal
- 03Execution
- 04Quality review and delivery, with an agreed round of revisions
Read the full project workflow.
In depth: how the work is done
Spatial data management and projections
Spatial analysis requires consistent coordinate reference systems and attention to datum and projection, since distance and area calculations depend on them. Resolution, extent and resampling method must be chosen deliberately, as resampling categorical data by averaging is meaningless. Metadata documents data sources, dates and processing so the dataset can be audited and reproduced.
Image preprocessing
Satellite images require radiometric calibration, atmospheric correction where appropriate, cloud and shadow masking, and co-registration for time series. Choice of sensor involves trade-offs among spatial, spectral, temporal resolution and cost. Radar imagery adds speckle filtering and different interpretation. Each preprocessing step affects subsequent classification and change detection, so settings are recorded.
Image classification and change detection
Supervised and unsupervised classification, including machine-learning classifiers, assign pixels or objects to classes. Training samples must be representative and independent of validation samples; spatial autocorrelation between nearby pixels can inflate accuracy if they are split randomly. Change detection compares images from different dates, requiring consistent preprocessing to avoid mistaking seasonal or sensor differences for real change.
Accuracy assessment
Map accuracy is estimated with an independent reference sample designed using probability sampling. The error matrix gives overall, user's and producer's accuracy and kappa-type indices, and area estimates should be adjusted for classification error because pixel counting is biased. Confidence intervals accompany all accuracy figures. Reporting only overall accuracy can conceal poor performance for rare classes.
Spatial statistics and overlay analysis
Overlay and multi-criteria methods combine layers into suitability or risk maps, with weights that are subjective unless justified. Spatial statistics, such as hot-spot analysis, spatial regression and point-pattern analysis, test whether patterns differ from random. The modifiable areal unit problem means results depend on the zoning used, which we test by varying scale.
Reproducible geospatial workflows
Scripted workflows with documented software versions and parameters replace manual point-and-click steps that cannot be repeated. Data licences, privacy of location data and sensitivity of mapped information are respected. Outputs include metadata describing lineage and accuracy.
Time-series analysis of satellite data
Dense time series support phenology, flood mapping and trend detection, but they require handling of clouds, sensor changes and orbit effects. Gap-filling and smoothing can introduce artefacts, so their settings and influence are reported. Trend tests should account for autocorrelation, and changes detected near the sensor's noise level are treated as uncertain. Combining optical and radar data can reduce cloud limits but demands careful alignment of their different physical meanings.
Ethics and privacy in spatial research
High-resolution imagery and location data can reveal sensitive information about people, property and protected sites. Researchers should consider whether data release could cause harm, comply with licences and anonymise locations where needed. Mapping of vulnerable populations or ecologically sensitive species may need aggregated outputs. These considerations are documented in the methods so readers see how the risks were handled.
Choosing resolution and scale
The scale of analysis should match the question. A coarse product may be adequate for regional trends but hide the small features that matter for a local study, while very high resolution adds cost and noise. We compare candidate datasets on pixel size, revisit time and spectral bands, and test whether conclusions change when the analysis is repeated at another scale. This makes the choice of data an explicit, defensible part of the method.
Quality control and good practice
Quality-control approach
- Models and analyses are checked against hand calculations, published benchmarks or your data where available.
- Assumptions, boundary conditions and sources are listed in the deliverable.
- Units, symbols and terminology are checked for consistency.
- Limits of validity are stated; the work is research-level analysis, not certification.
Common pitfalls this service helps avoid
- Random split of neighbouring pixels. Spatial autocorrelation inflates validation accuracy.
- Counting pixels to estimate area. Classification errors bias area estimates without adjustment.
- Comparing images without consistent preprocessing. Apparent change may be due to sensors, seasons or atmosphere.
- Arbitrary weights in suitability maps. Results can change substantially with weighting choices.
What good reporting looks like
A GIS or remote-sensing report describes the data with dates, resolution and processing, the classification or analysis methods, training and validation sampling, the accuracy assessment with confidence intervals, the area estimates with uncertainty, the sensitivity to scale and weights and the limitations. It does not constitute a legal survey, cadastral determination or regulatory mapping.
An illustrative scenario
Hypothetical example written for explanation. It is not a client case, a result or a testimonial.
Suppose a researcher maps urban expansion between two years. The first classification reports 94 percent overall accuracy using randomly split pixels. A spatially separated validation sample gives 85 percent, with low accuracy for the bare-soil class. Area estimates adjusted for classification error differ noticeably from pixel counts. The report presents the adjusted areas with confidence intervals and notes seasonal differences between the images.
Questions to settle before you begin
- Which coordinate system and resolution are right for my question?
- How will I design an independent validation sample?
- Are my images comparable in season and sensor?
- How will I adjust area estimates for classification error?
- Is my weighting of criteria justified?
Key terms
- Coordinate reference system
- A framework linking coordinates to positions on the earth.
- Error matrix
- A table comparing mapped classes with reference classes.
- Producer's accuracy
- The probability that a reference class is mapped correctly.
- Co-registration
- Aligning images so pixels correspond.
- Modifiable areal unit problem
- The sensitivity of spatial results to the zoning used.
Working together on Earth Sciences & Geosciences projects
Earth sciences engagements start with the dataset or field question: location, scale, sampling approach, measurement methods and the uncertainties in each. We ask for data tables, maps or logs in standard formats, the conceptual model you hold and any prior interpretations. Analyses make explicit how spatial and temporal sampling limits what can be concluded. Reports describe uncertainty and non-uniqueness, as in geophysical inversion, and separate observation from interpretation. Site-specific engineering or hazard decisions require qualified professionals and are outside the scope.
Frequently asked questions
Where does the imagery come from?
Often from open archives or data you hold; data access and licensing are discussed at scoping.
Can you classify land cover from satellite imagery?
Yes, with an independent, properly designed accuracy assessment.
Which software do you use?
Open-source and commercial tools may be used and documented; the choice depends on your needs.
Can you do change detection over several years?
Yes, with consistent preprocessing and uncertainty reporting.
Do you provide legal survey data?
No. Legal surveys require licensed surveyors.
How is this service scoped and priced?
After your enquiry we review what you need, ask for missing information and send a written scope and quotation. Price depends on complexity, deliverables and timeline. See the project workflow.