PentlandX. David McCann Paul Bell. Field demonstration of the capabilities of X-band radar for coastal remote sensing.

Transcription

1 PentlandX Field demonstration of the capabilities of X-band radar for coastal remote sensing David McCann Paul Bell

2 Disclaimer Please be aware that the results presented in this presentation are the result of an interim analysis of the full range of data collected for the project. Therefore the final report contains results that differ (in some places considerably) from the final report.

3 PentlandX - Introduction Knowledge Exchange partnership between NERC and MeyGen ltd. Working towards a better understanding of the tidal energy resource of the Pentland Firth (N. Scotland) through application of NOC research methods Showcasing the capabilities of X-band radar oceanography in a challenging, remote operational environment Help MeyGen develop their models to a point of high confidence for turbine placement, and thus maximise energy generation on site.

4 Current capability Alongside the commercially available WaMoS HRC algorithms, the current NOC radar toolbox comprises a range of analyses, all performed on the same raw radar data Bathymetric survey (depth inversion of wave dispersion) Current measurement (dynamic effect of current on wave dispersion) Sea surface roughness analysis (tidal effect on surface roughness due to turbulence) Small target tracking (Marine wildlife, both airborne and seaborne)

5 Project area Pentland Firth Isle of Stroma John O Groat s Radar site Duncansby head

6 Co-tidal chart (POLPRED ORKM grid)

7 Project overview Deployment 5 th March to 11 th June 2013 Radar data recorded for 5 minutes every minutes continuously Compression routine run to reduce data rate to ~ 320Mb per hour WaMoS High Resolution Current (HRC) analysis run in real time All data stored and backed up to removable media Currently working through post-processing with NOC research-grade analyses

8 Equipment set-up Kelvin Hughes X-band marine surveillance radar equipped with a 2.4m antenna, mounted on a tower apprx. 12m above MSL WaMoS II radar computer and frame digitiser (OceanWaveS GmbH) GPS time-keeper for accurate data time-stamping Diesel generator (6kVa) connected to an inverter/charger and wet cell battery bank (OffGrid Energy)

9 Equipment set-up Radar antenna 4.2m Radar computer and data recorders Generator Battery bank

10 Equipment set-up GPS antenna WaMoS II Radar recorder PC Radar display Kelvin Hughes X-band radar GPS time-keeper

11 OffGrid Grid2Go Portable power supply system Remote location required a portable power solution Battery supply lasted ~ 3 days between each 6 hour charge 6kVa Diesel generator burned 371 lt of fuel over 91 days Equivalent use if generator was run continuously ~ 3325 lt Estimated CO 2 emission from generator ~ 971kg Approximate CO 2 emission from equivalent grid energy use ~ 732kg Estimated CO 2 emission if generator run continuously ~ 8.7 tonnes

12 Raw radar data Stroma Radar Duncansby Lighthouse

13 Raw radar data Example of the raw data that is fed to the analyses. Images like this are captured every 2.4 seconds for a total of 256 images in ~ 5 minutes. Importantly, the radar is simply a microwave camera, taking sequential images of not only waves but everything on the sea surface (and land) that reflects microwave energy. Every 256 image run, the compression routine packs the data for storage and the WaMoS HRC commercial toolbox runs before the next sample period. Critically, it is the wave signal that produces the data for most of the analyses. Without a certain threshold of wave activity there can be no current calculation. As waves are affected by refraction and diffraction the wave direction (swell) is vital for data quality in certain areas During the majority of the project the wave direction was Easterly, creating a significantly lowered area of data quality in the West of the inner sound. This should be kept in mind when interpreting output.

14 WaMoS HRC data (Flood tide)

15 WaMoS HRC data (Flood tide) Here is an example of a snapshot of tidal currents from the WaMoS HRC analysis for an arbitrary flood tide, highlighting the resolution and general data quality of the method. The radar s range is denoted by the red line and the red diamond denotes the antenna position. The Lease areas of Meygen (left) and Scottish Power (right) are denoted by black polygons. Data quality to the West of the inner sound has been severely degraded due to the lack of penetration of Easterly waves into the area. Data quality is also degraded towards the maximum range of the radar due to range effects. In short, the quality of the WaMoS HRC data for the majority of the survey was greatest in the centre and East of the inner Sound.

16 NOC research analysis (Bathymetry and currents)

17 NOC research analysis (Bathymetry and currents) This image is an example of the output from the NOC research-grade analysis for an arbitrary Ebb tide. Contour colours are bathymetry below Mean Sea Level (MSL) from the NOC bathymetric inversion algorithm and vectors are flow speed (no scale). Both resolution and overall data quality of the NOC analysis are greater than the HRC toolbox, however the execution time for the NOC analysis is an order of magnitude longer. As such, the NOC analysis area has been limited to that of the Meygen lease area in order to analyse as much data as possible for the project. The NOC analysis is more sensitive to smaller wave events, leading to an overall increase in data quality for secondary analyses.

18 Tidal currents harmonic analysis Radar derived currents are strongly affected by signal strength, which is related to wave energy Tidal current snapshots do not convey enough long-term information about a tidal resource Tidal harmonic analysis of radar-derived currents provide this information while acting to statistically minimise errors caused by occasional drops in signal strength

19 Tidal harmonic analysis, single point comparison WaMoS HRC Constituent % Energy Amplitude (m/s) Phase (deg) M S M N NOC research analysis POLPRED ORKM grid Constituent % Energy Amplitude (m/s) Phase (deg) M S N M Constituent % Energy Amplitude (m/s) Phase (deg) M S N L

20 Tidal harmonic analysis of WaMoS HRC data Harmonic analysis was performed at each point in the survey area, creating a map of tidal harmonic components Harmonic analysis is sensitive to record length the input data defines the output constituents so reductions in data quality cause reductions in constituent quality and quantity. The aforementioned signal strength problems in the West of the Inner Sound (and much of MeyGen s lease area) have severely reduced the computed amplitudes of tidal constituents and therefore can not be trusted However, closer to the antenna and through much of the Eastern area of the Firth, record length and data quality was high enough to produce a strongly-converging set of tidal constituents.

21 Tidal harmonic analysis of WaMoS HRC data

22 Tidal harmonic analysis of WaMoS HRC data The following images depict the M2 tidal velocity amplitude as well as the estimated maximum spring (M2+ S2) and minimum neap (M2 - S2) amplitudes. It should be noted that these are only two of the four most important constituents in the area. The maximum tidal flow speed is a combination of up to 40 constituents and meteorological effects and so will be greater than just the M2 component for example. However, tidal harmonics are periodic and if accurate can be used to predict tidal currents for any time. Radar data is unique in that tidal harmonic maps may be produced conventional survey techniques do not allow this. The residual tidal velocities represent the mean (residual) tidal flow after harmonic analysis has removed the harmonic component of the tide. The distinctive circulation seen in the residual vectors is believed to be responsible for a quasi-stationary dune field at this location.

23 Tidal harmonic analysis of WaMoS HRC data

24 Tidal harmonic analysis of WaMoS HRC data

25 Tidal harmonic analysis of WaMoS HRC data

26 Tidal harmonic analysis of NOC research data A similar analysis was performed on the radar-derived tidal velocities from the NOC research-grade analysis As the overall data quality is higher from the NOC analysis the record lengths being input into the harmonic analysis were longer and therefore an overall increase in predictive quality is observed Quality control techniques automatically reject data below a certain quality, denoted by grey areas in the following plots. Unfortunately, the increase in analytical quality is paired with a substantial computational cost and so not all of the data has been analysed. It is expected that eventually much of the greyed area will be filled in as more Westerly wave events are worked up.

27 Tidal harmonic analysis of NOC research data

28 Tidal harmonic analysis of NOC research data

29 Tidal harmonic analysis of NOC and HRC data

30 Tidal harmonic analysis of NOC and HRC data

31 Flood/Ebb misalignment Flood/Ebb misalignment is the angular difference from a rectilinear flow that is to say the angle between Flood and Ebb flow directions minus 180 degrees This angle is important for tidal turbine considerations as the turbine needs to face the flow with no oblique angle. Differences from a rectilinear tidal flow therefore require adjustments in design in order to maximise energy yield. The misalignment in the subset area of the MeyGen lease is predominantly due to inertial flow around the Isle of Stroma which is forced through the channel around a bend. Flood and Ebb currents flow around this bend from different incoming directions, increasing the misalignment.

32 Flood/Ebb misalignment

33 ADCP and NOC research data comparison Validation of radar-derived currents was achieved through co-incident ADCP surveys, an example of which is displayed in the following figure In this case the comparison is better during flood tides (higher peaks) as at this date the waves were Easterly which oppose a flood current. Opposing currents and waves are known to increase the signal to noise ratio in our radar analysis through a backing up of the waves.

34 ADCP and NOC research data comparison

35 Sea surface roughness analysis Application of experimental NOC research analysis to Pentland Firth data Identified tidally-modulated sea surface roughness signatures imaged by the radar Such signals are believed to be related to bathymetricallylocked vertical turbulent eddies and horizontal shear zones Could provide important information on large-scale turbulence which could severely affect turbine operation

36 5 minute (256 frame) mean image, Flood tide

37 5 minute (256 frame) mean image, Ebb tide

38 5 minute (256 frame) mean image, Ebb tide

39 Tidal surface roughness amplitude The following image is an example of the output from the new sea surface roughness analysis. Here the tidally-modulated component of the sea surface roughness signal is correlated to its relationship to either the flood or ebb tide In the image, red colours indicate signals that are correlated with the flood and blue the ebb tides The presence of tidally persistent signals over the dune field suggest the interaction between bathymetrically-locked vertical eddies created by flow separation over the dune crests Other features include wakes and horizontal shear zones that persistently occur every flood or ebb tide

40 Tidal surface roughness amplitude

41 Small target tracking Radar images any microwave reflector on or above the sea surface Seabirds and other airborne/seaborne targets can therefore be tracked Employs a relatively simple yet robust Global Nearest Neighbour (GNN) data associator and a target kinematic model Able to track multiple targets simultaneously in good weather, 24 hours a day Could provide a significant augment to traditional ecological impact surveys at MRE sites

42 Target tracks over 5 minutes (speed in m/s) The following figure shows the output of 5 minutes of bird tracking with individual target tracks coloured by their measured movement speed (m/s) The concentration of targets around the radar is predominantly a range effect the tracking method has a range of approximately 2km due to the range-dependent component of the signal to noise ratio. A lack of targets further than 2km does not necessarily denote a real lack of bird activity Further analysis on the bird tracking data will provide estimates of abundance maps, as well as correlated bird counts on flood/ebb tides and day/night cycles.

43 Target tracks over 5 minutes (speed in m/s)

44 Conclusions Demonstrable benefits to the MRE sector Preliminary site investigation reduces financial risk inherent in traditional point survey methods Cost effective monitoring Long term, site-wide data products without the financial overhead of vessel-based surveying Data products add a vast amount of information, context and texture to surveys and models, allowing the MRE sector to make a better informed decision on their resource

45 Thank you for your attention National Oceanography Centre Joseph Proudman Building 6 Brownlow Street, Liverpool, L3 5DA. Tel: + 44 (0) National Oceanography Centre University of Southampton Waterfront Campus European Way, Southampton, SO14 3ZH. Tel: + 44 (0)