Remote sensing standards: their current status and significance for offshore projects Peter J M Clive Technical Development Consultant SgurrEnergy Ltd 225 Bath Street Glasgow G2 4GZ E: peter.clive@sgurrenergy.com T: +44.141.227.1724 Remote sensing technologies such as Sodar and Lidar have proved to be cost-effective methods of acquiring critical project datasets from locations that present access challenges. As a result they are increasingly being adopted in the offshore environment where their compact, portable and robust characteristics can be maximally leveraged to acquire data that previously would have represented greater cost or would not have been available at all. The codification of best practice in relation to remote sensing has made significant progress and currently is the focus of efforts made by, for example, the IEC, the IEA and MeasNet, among others. This process will provide both informative recommended guidelines and a regulatory framework of normative procedures which remote sensing practitioners will need to comply with to ensure, for example, the robustness of calculations of project uncertainty, representing in a complete and unbiassed manner all sources of project uncertainty. This paper outlines the status of the regulatory environment regarding the use of remote sensing offshore as well as discusses the benefits that accrue through compliance. In addition key sources of uncertainty are discussed. Is remote sensing bankable? Bankability requires two things to be true: 1. The long term productivity and asset longevity is sufficient to service the debt raised to finance the project; 2. The percentile chosen to represent the productivity that can be relied on for example the P90 is derived from a robust uncertainty analysis that represents all the sources of project uncertainty in a complete and unbiased manner. Every instrument introduces uncertainty as a result of measurement errors and removes uncertainty by virtue of the data acquired, including conventional cup anemometry. Bankability depends on being able to quantify uncertainty, not on the specific instruments used. This need to quantify uncertainty largely drives progress towards clear, open and transparent standards describing procedures that can be universally implemented. Performance verification The quantitative significance of measurements acquired using a remote sensing device relies on its ability to deliver accurate and reproducible data which can be traced back to reference sensors calibrated in facilities that are compliant with national standards. The wind flow parameters the remote sensing device should be able to measure in this manner should include 1. Horizontal wind speed 2. Wind direction 3. Relative wind speed (wind shear)
The third of these requires not just the remote acquisition of wind speed measurements but the accurate recording of the range or height at which these measurements are made. In general the sensitivity of the remote sensing measurements to various environmental factors should be investigated to determine the circumstances under which the performance of a remote sensing device is sufficiently well understood to allow robust uncertainty calculations. One such factor is shear and the response of a remote sensing device to shear is related to the variation in its performance with height as discussed below. Example of sensitivity: height dependent bias Bias that varies with height is significant as it will lead to spurious assessments of wind shear (the variation of wind speed with height). There are four principal sources of height dependent bias in 1st generation wind Lidar measurements. 1. Inhomogeneous flow within the volume defined by the scan geometry [2] due to complex terrain, wakes, variable flow inclination, obstacles and other influences on the flow within measurement volume. This is already relatively well documented [1]; 2. Averaging of a non-linear shear profile over a range of heights corresponding to the projection of the probe length/range gate onto a vertical axis and assigning the resulting average value to the height corresponding to the mid-gate range. This has been discussed in detail previously [3,4]; 3. Turbulence induced vector averaging bias incurred when the diameter of the VAD or DBS disc approaches turbulent length scale. This is discussed below for the first time to the author s knowledge; 4. Cloud anomalies. This only affects continuous wave Lidars as their ranging systems rely on a variation in sensitivity with height which may be obscured by a variation in signal strength with height associated with, for example, backscatter from clouds. This has been discussed previously [5] Turbulence induced vector averaging bias The red circles in the plot Figure 1 below show the slopes derived from the regression of ZephIR Lidar measurements against measurements acquired by calibrated reference sensors mounted on an 80m mast, as reported in the Myres Hill Remote Sensing Inter-comparison Study [6]. These seem to vary with height in a manner suggestive of a height dependent bias relative to the reference sensors, resulting in a decreasing slope with increasing height. The comparison of the ZephIR with a second ZephIR, shown by the blue squares, does not exhibit the same trend, which suggests the bias relative to the mast mounted reference sensors is an artefact of the remote sensing methodology implemented by the ZephIRs. An assessment of the local orography suggests that the variation in flow inclination is insufficient to account for the effect observed above. A simple highly preliminary model was developed to account for the bias. A turbulence profile was assumed with turbulence intensity TI decreasing with height as follows TI = TI 0 1 where TI 0 is the turbulence intensity at ground level, h is the height, H BL is the height of the boundary layer and k is an exponent describing the profile. It is observed that turbulence will cause a variation σ in the magnitude and direction of the wind velocity vector at points around the circumference of the disc described by a 1st generation scan geometry such as a VAD (Velocity Azimuth Display) at a given height h. This will reach a maximum when the diameter of the disc D is comparable to the turbulent length scale L h H BL k
D σ = TIsin π 2L where D varies with h. Figure 1: Performance verification regression slope as a function of height
Then the vector averaging bias incurred by the variation in the velocity vector around the disc is given by [4] v v Vector Scalar sin = The dotted red line above shows a fit of this model to the data presented. This suggests a more rigorous analysis, perhaps considering the coherence function of the turbulent flow, may prove fruitful. 3σ 3σ Example verification scheme The performance of a well sited remote sensing device can be verified by comparison with concurrent data from a co-located reference calibrated cup anemometry mounted on a tall mast. Comparisons are made for each 1. 10 degree direction sector 2. 0.5ms -1 wind speed bin between 4ms -1 and 16ms -1 Data are acquired until there are at least 3 data points in each wind speed bin. The bin averages are regressed and the bin-wise statistics support the calculation of comparison statistics upon which uncertainty calculations can be based. The validity of the performance verification at sites other than the test site is supported by sensitivity analyses which examine the influence of, among other, flow inclination, precipitation, flow inhomogeneity, shear, veer, turbulence, temperature, atmospheric stability, amongst other parameters. Cost-benefit The use of remote sensing in a manner that complies with the developing guidelines and standards requires performance verification by comparison with calibrated cup anemometry mounted on temporary meteorological masts. However, the best use of remote sensing, in terms of cost-benefit and the reduction in project uncertainty that can be achieved for a given investment in data acquisition, will often entail modes of operation that have no direct analogy with met masts. The installation of remote sensing in basic 1 st generation mast replacement roles, where the device acquires data from the volume of air immediately above it and records wind profile information in a manner and format that is comparable with mast methodologies, does not necessarily compete with masts themselves in terms of the reduction in project uncertainty achieved for a given spend on data. This is notwithstanding the ability to acquire taller profiles, making measurements at heights above levels at which masts can be installed economically, and also takes into consideration the compact portable configuration of many remote sensing devices which can be moved to multiple locations in succession with greater ease than masts. In many instances the data acquisition budget would have as effectively been spent installing additional masts, and a project and site specific assessment of the benefit of conducting 1 st generation remote sensing must be conducted. The best use of remote sensing devices, which delivers an optimal cost-benefit that cannot be achieved using masts alone, requires the implementation of 2 nd generation techniques that have no direct analogy with mast measurements. In order to get the most from remote sensing devices one must think in broader terms than those associated with masts. For example, surveys of large areas to rapidly investigate the development of wind shear across a project site, using multiple PPIs with different elevation angles, allows the acquisition in a few weeks of detailed information that would otherwise take months to obtain, even with the use of 1 st generation devices. Measurements and assessments that may not previously have been considered, because their implementation using masts and 1 st generation remote sensing
represented too high a cost, can be delivered within budget and may even become routine using 2 nd generation techniques. This approach delivers fundamentally new information to a resource or site assessment which would not otherwise have been available, as opposed to merely replicating the capabilities of masts. As a result more dramatic benefits become available. This presents a fundamental challenge for the development of guidelines and standards: how to bring 2 nd generation techniques within the scope of recommendations whose development has been initiated and progressed precisely to quantify the extent to which 1 st generation remote sensing devices perform like masts. The need to incorporate 2 nd generation techniques within this process is more pressing given the suitability of these techniques for offshore measurements, where the relative inaccessibility of much of a project site and the expense of monitoring it can be mitigated by the long range of a 2 nd generation device implementing a 2 nd generation scan geometry. Conclusions The growth of confidence in remote sensing is being supported by the development of a framework of standards, guidance and recommendations regarding best practice. A key development will be the anticipated publications of a revision of IEC 61400-12-1 including an Annex detailing procedures for remote sensing device performance verification that includes sensitivity analyses such that the devices can be fully incorporated in a complete, robust and unbiased uncertainty analysis. The application of remote sensing more widely requires such an assessment. References 1. Clive, PJM. Remote Sensing Standards: the Current Status, EWEC, 2010 2. Clive, PJM. 2nd generation Lidar techniques, EWEC, 2010 3. Clive, PJM. Highlighting uncertainty with Lidar, AWEA Wind Resource and Project Energy Uncertainty Assessment Workshop (Invited Paper), 2007 4. Clive, PJM. Compensation of bias in Lidar wind resource assessment, Wind Engineering 32-5, 2008 5. Clive, PJM. Mitigating the impact of complex terrain on Lidar wind resource assessment, BWEA30, 2008 6. L. Monaghan, Myres Hill Remote Sensing of Wind Speed Intercomparison Study Final Report, Garrad Hassan Report Number 1026/GR/1