Remote Sensing Techniques for Renewable Energy Projects Dr Stuart Clough APEM Ltd
What is Remote Sensing? The use of aerial sensors to detect and classify objects on Earth Remote sensing for ecological applications essentially focuses on the classification of either habitats or of the species themselves
Remote Sensing for Renewables - examples Habitat mapping for linear assets such as onshore cable routes from offshore devices Very high resolution imagery of potential routes from landfall to substation Automated classification into land cover types Allows route selection to be refined, fewer landowner contacts Allows for ground surveying to be targeted
Remote Sensing for Renewables - examples Habitat mapping for hydropower projects Very high resolution imagery of river channels Automated extraction of water depth and grain size maps, and fish habitat maps Allows for detailed assessment of sediment transport issues Allows effect of changes in flow to be quantified
Remote Sensing for Renewables - examples 3D Terrain Mapping - automated digital terrain model extraction from stereo imaging
Remote Sensing for Renewables - examples Thermal imaging of seals for marine renewables projects abundance, distribution and habitat use
Case Study the evolution of remote sensing techniques for offshore wind studies Historically most studies used observer based methods from either a vessel or aircraft platform Used statistical methods to correct for reduced detection ability with distance from the platform The problem with boats some birds are attracted, others are repelled, cause disturbance (high % in flight), post construction bias, unquantifiable error The problem with visual aerial health and safety (200ft), disturbance, post construction bias, unquantifiable error Improvements in sensor technology meant these problems could be overcome
Benefits of the switch to digital Operational benefits Health & Safety; survey altitude, less time at sea Survey design consistent; no change post-construction More flexible - responsive to weather windows Cheaper than boat based surveys Biological benefits No disturbance to animals being surveyed Statistical benefits No attraction / repulsion from platform Allows randomised designs (the image is a quadrat!) Quicker reduced effect of temporal change Data can be matched to environmental gradients for modelling Error is quantifiable, population change can be measured Confidence Fully auditable, quality assurance Fully objective; human error reduced
Technology continues to improve Moores law 60 megapixels First camera system to provide forward motion compensation on 2 axes: Stability and accuracy Multispectral (RGB and NIR bands) and very high resolution Leica RCD30
Survey design Case-by case-basis Consider site, species, consenting risks How best to answer the questions Specialist software plots flight lines and nodes Camera only fires when target location reached
Survey approach: Grid-based survey Back to first principles Classic design quadrat style Independent samples of population Statistically more powerful than transect based methods Trade off between percentage coverage and processing cost
Project specific outputs: Which species are present? Distribution and abundance? Bird flight altitude Proportion of adults/juveniles Direction of bird movement (SPA linked?) Population estimates
Example images - high resolution
Automated image analysis Automated Object Recognition Bird wing span and body length Bird flight direction Bird flight height
Example GIS Output
Quality Assurance APEM image assessment by trained observers Number of birds/marine mammals Grouping/species Individual Position (GIS based) External independent QA BTO and SMRU
UKAS Accreditation UKAS accredited for the Identification & Enumeration of Birds from Aerial Photographs Extension of existing Lab Accreditation Ensures Quality of Results Ensures Reproducibility & Traceability Drives continuous improvements Evidence of high standards client and regulator confidence in our data
Additional outputs - bird flight height Altitude of Aircraft (known) Camera Bird Altitude (calculated) Bird Resolution (Calculated from average size) Surface Resolution (known)
Additional outputs - flight direction & SPA connectivity Individual birds are georeferenced Bearing is automatically determined from headtail axis Extraction to GIS Rose diagrams produced for defined areas
Statistical methods for population estimates Generalised Additive Models (GAMs): Method of fitting a smooth relationship between two or more variables to assess complex relationships. Geographical, physical and environmental covariates used to inform model E ( s j ) exp 0 fk ( z jk ) k Where Ѳ 0 is the intercept, f k are smoothed functions of the explanatory covariates, and z jk is the value of the k th explanatory covariate in the j th group Fitted with a suitable error distribution to cope with the non-parametric nature of counts
Data preparation A template grid is created, with an appropriate cell size to cover entire survey area Cell size chosen to minimise the number of assumptions about the spatial distribution of individuals Environmental parameters (depth, X, Y, and distance to shoreline/spas) are spatially joined to the survey grid
GAM outputs 5 km 25
GAM outputs Animal abundance within each cell predicted by a GAM based on significant environmental parameters GAM-derived values spatially joined to the template grid to create GAM prediction maps for individual bird groups/season and imported into ArcGIS Enables sensitive planning of renewable energy projects as well as population estimates / distribution
Conclusions Sensor technology continues to improve New sensors and better sensor resolution will result in new applications Remote sensing produces a permanent record and statistically robust, auditable data Expert ecological interpretation will always be required Remote sensing will have an increasing role in ecology and environmental management projects in the future
Thank you Dr Stuart Clough Director APEM Limited Tel: (0161) 442 8938 Email: s.clough@apemltd.co.uk www.apemltd.co.uk