Environmental Impact Assessments for renewable energies
DMP provide the statistical analysis services for many environmental impact assessments. Our clients include the European Marine Energy Centre Limited (EMEC Ltd.), Aurora Environmental Limited, and the Sea Mammal Research Unit Limited (SMRU Limited). We currently provide the EIA analysis to both the Strangford Lough and Falls of Warness undersea turbine installations.
Our clients typically seek answers to the following:
- What is the relationship between the animals and their environment in terms of environmental conditions, time of day/year, and location within the Lough?
- If a power installation were to be placed in this area, what type of impacts might be detectable using this observer system?
- What is the ability of this observation scheme to detect changes of interest?
- (Post installation) Has there been a true change in the underlying numbers of animals present in the study area? Or, have they altered their use of the space?
For example, the Strangford Lough data comprise of continual observer sightings of marine mammals and birds –species, numbers and locations. Oceanographic conditions are also recorded or interpolated. In total the data represent visual scans every few minutes, taken over many days/months/years. The data can require extensive cleaning and validation prior to analysis.
Answering the client’s questions requires relatively advanced statistical models that use custom-coded components in the statistical programming language R. Specifically DMP employ:
- Generalized Estimating Equations (GEEs) to account for the repeated (non-independent)
- measures through time. Failure to use the correct method here (i.e. naïve models that assume independence) will give grossly incorrect measures of uncertainty.
- Splines (some cyclical) to model animal relationships with the periodic changes in their environment.
- Model-based simulations under various impact scenarios to determine the probability of detected effects of marine installations. This is a very computer-intensive process requiring days of computer-time.
Statistical models, reports and simulation software.
The DMP models provide baseline relationships between the monitored species and their environment through time. These are accompanied with measures of uncertainty and collectively allow the client to determine what sequences of observations would be ‘unusual’ i.e. would indicate an impact.
The simulation models provide a range of installation-impact scenarios and the corresponding probability of detecting these. It is clear from this whether the client needs to consider alterations of the sampling strategies. For example, the client may:
- Reduce costs if current effort exceeds the desired goals.
- Increase effort to achieve the desired levels of detection.
- Investigate more complex alterations to the sampling design to optimise effectiveness.
Future post-installation observations can be fed to the models/simulation to
determine if these represent an impact, as well as the associated uncertainty
with this conclusion.
These projects require DMP to custom build regression models for cyclical non-linear patterns with spatial components and long-term repeated measurements. Bespoke simulation software is also required for projections of possible effects of turbine installation and the effectiveness of current monitoring regimes. No aspect of this is statistically trivial and naïve analysis of the problem would produce incorrect conclusions.