Lidar type uncertainty

It's been almost a decade since my first Wind Physics class at Carl von Ossietzky University of Oldenburg, where I was introduced to lidar systems. At that time, wind lidars had only a few years transitioned from academia to the real world. Since then, the lidar industry has matured significantly, expanding globally and proved to be a reliable source of wind measurements. This has played a crucial role in de-risking the development of numerous wind farms by reducing yield uncertainty.

Previously, the differences between continuous wave (ZX Lidars) and pulsed (Windcube) systems were largely theoretical, with both types delivering similar quality measurements in practice. However, as turbine hub heights have increased, particularly offshore, the distinctions between these technologies have become more relevant.

I won’t delve into pros, cons, and limitations of each technology here. The point is while the industry may have initially overlooked these differences, over the years lidar suppliers have diligently improved their systems' hardware and firmware. This ongoing development has aimed to enhance measurement quality, particularly at the system's operational limit.

However, these advancements have introduced complexity. It has become challenging to characterize the expected performance of a lidar type without detailed knowledge of factors such as firmware and hardware version, filtering configuration, data post-processing, configuration at calibration trials and more.

The complexity is further amplified in the floating lidar industry. The industry has set stringent requirements for floating lidar performance. Among other things, these requirements include minimum levels of data availability and correlation statistics against reference instruments. Such requirement intended to further drive industry and technology development.

However, it raises the question: Is it fair to penalize a floating lidar supplier for low data availability due to environmental factors like fog or low aerosols density? Effectively floating lidar suppliers have limited control over lidar performance in such conditions. They can opt for a different lidar type but that would only mean they'll need to handle a different set of operational limits.

Consequently, floating lidar suppliers have begun focusing on data processing to improve performance metrics. Today, most suppliers have dedicated R&D teams working on enhancing their data products, adding yet another layer of complexity to the characterization of expected lidar performance.

The immediate consequence is that aggregated analysis of lidar performance - that kind of analysis where you draw insights by comparing the performance of several systems over time - has become unviable due to the multitude of differences between units of the same type. The same is true when it comes to modelling and simulation: I dare to say that academic lidar type modeling may not represent the expected behavior of commercial systems available to the industry today given the lack of transparency on post-processing methodologies.

These layers of hardware, firmware and post-processing creates a complex environment to assess the performance of systems at the specific measurement campaign. Based on my recent experience, I already noticed differences in lidar behavior when comparing two systems of the same type and firmware but deployed by different FLS suppliers - both measuring within the same region.

As if there wasn’t enough, complexity increases further when a third-party lead measurement campaigns (i.e., governments, private investors, etc.). In some cases, not all necessary information is exposed to specialists, as such, preventing conclusive due diligence on the characterization of lidar performance.

The solution for this complex scenario is transparency and standardization.

Wind lidar specialists should always have access to raw measurement data (radial wind speeds and statistics). This will form a baseline measurement quality in which, ideally, only differences in hardware would introduce distinct behavior of the lidar type.

The development of post-processing techniques (filtering, motion compensation, etc.) aiming to bridge technology limitation is encouraged and necessary. However, post-processed data should be provided as additional data delivery to the raw measurement. Additionally, likewise outlined on Clause 6.2 of the to-be-released IEC 50-4, detailed processing algorithm documentation package should be an integral part of the delivery.

Newer post-processing algorithms should not automatically replace previous versions, specialists may benefit from having access to both datasets so that the confidence in the newer version is built over time by comparing it to former version and both against the raw data. Such “transition” is important to avoid blind spots inherent to new releases in scenarios in which it was not sufficiently tested.

On the standardization front, third-parties could assess applicability of a “Lidar/Floating Lidar Production-Line Certification Program" to standardize and control manufacturing lines and lidars BOM, thereby enabling the practice of random sampling. In the long term, this initiative has the potential to reduce the need of field verifications.

Finally, industry guidelines related to the commercialization of floating lidars could rethink requirements of data availability and prioritize system availability instead. This shift has the potential to refocus suppliers on system attributes they have full control of, such as redundancy, energy management, collision aversion technology and other relevant system design subjects.

I want to hear your thoughts and comments about this topic.

How do you perceive transparency and standardization on lidars / floating lidars?

Note: opinions expressed are solely my own and do not express the views or opinions of my employer.