A Review of Sigma Space’s HRQLS, HAL and HRQLS2 sensors and Their Conformance with the USGS LiDAR Base Specification
A 1.640Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE
The photon LiDAR technology continues to garner much interest and will continue to do so. It is unfair at this point to lump this exciting new technology into just one category. There is Single Photon LiDAR (SPL) and there is Geiger Mode LiDAR (GML). As a result of the "Are We There Yet?" article in the May issue of LIDAR News it is important to look at these two technologies separately. Additionally, the work required to get these two types of technology to conform to the USGS LiDAR base specification is very different and it looks like the SPL sensors are closer than we thought.
The Sigma Space website (www. sigmaspace.com) indicates that this company is much more than the maker of airborne SPL sensors. They have developed components for the Advance Topographic Laser Altimeter System (ATLAS) instrument in NASA’s Ice, Cloud, and land Elevation Satellite-2 (ICESat-2). These components include the photon timing electronics, and the LRS (Laser Reference System) The ATLAS is designed to fly on ICESat-2 at an orbit, 500+ kilometers above the earth’s surface. This requires extraordinarily accurate LiDAR footprint positioning from moment to moment, which is possible with the information from the LRS since it can measure the pointing of the laser beams with respect to the stars. For this project, Sigma Space developed a Star-simulating thermal vacuum chamber on-site, calibrated to sub-arc second accuracy, in order to assess the performance and accuracy of that sensor.
The SPL sensors provided by Sigma Space are the HRQLS, HAL and soon to be released HRQLS-2. One of the difficulties for most LiDAR users is to understand the technology and terminology that goes along with this new technology compared to what we are used to with conventional LIDAR. One example of this would be the repetition rate of these systems. For instance, HRQLS2 operates at 5.0 megapixels/ second. The simple conversion of this as it relates to conventional LIDAR is that it produces 5 million measurement points per second. The same conversion is not valid for a Geiger Mode Lidar, but that is not the subject of this article. The table above details the operational characteristics of all Sigma sensors.
Unlike other photon LiDAR sensors, the SPL can operate during day or night with similar performance. Although, with the photon LiDAR sensors the solar noise (returns from solar radiation) can be higher than conventional lidars, Sigma Space has developed a way to minimize this better than other providers, accurately filtering and classifying the solar noise, and using the information for surface reflectivity measurements.
Another nice feature of the HRQLS, HAL and HRQLS-2 SPL LiDAR sensors is that they operate at a wavelength of 532nm. At this wavelength both topographic and bathymetric data can be collected with a single sensor and on the same flight. The exact bathymetric performance of the sensors is currently being calibrated but initial tests indicated that the sensor works up to 15 -20 meters in ocean water and several meters in riverine waters, obviously this is dependent on clarity and other water characteristics. Sigma Space continues to test the sensors in different water characteristics and will provide calibrated results soon.
Okay, so this all sounds pretty awesome and Sigma Space has conducted several projects with very good results and does a lot of work for the government (DOD) using HAL. The big question continues to be the ability to make these sensors and the data produced from them compliant with the USGS Lidar Base Specifications. The major hurdles right now may be the amount of return noise compared to conventional sensors and whether or not the data can be delivered in fully calibrated swaths.
The SPL sensors provided by Sigma Space should not have a problem delivering the data in calibrated swaths as this has been their method of operation from the beginning. Given that all Photon LiDAR sensors have significantly more return noise than conventional sensors this potentially can be perceived as an issue. But is it? Sigma Space has developed a rigorous compliance table based on USGS-LBS and is, one by one, meeting the specs with their systems or suggesting what variances may be requested. That is a humble and refreshing attitude from a hardware manufacturer.
Consider a project like Garrett County Maryland: the HRQLS sensor was used to collect Garrett County a forest covered area equal to 1,699 km2, in 12 hours at 12 ppm. Depending on the latest conventional LiDAR used, the same collection would take roughly the same time to collect at a QL2 data set at 2ppm. Yes, there would be more noise from the SPL sensor but the SPL sensor is collecting 6 times more points so what would be the noise ratio between the two resulting data sets versus the actual point density. Also, how would the two resulting bare earth models compare? In fact Sigma generated a bare earth model for a collection in Monterey, California where a very dense ground truth grid existed, and the RMSZe when compared to twenty control points measured with static GPS met the USGS QL1 standard.
The compliance with USGS Lidar Base Specifications go much further than calibrated swaths and noise. There will have to be some changes made to the specifications or the Photon LiDAR Groups will have to figure out how to comply as required. Sigma Space is in favor of the later. There is no advocacy by the author for either at this point but it is important to state the facts about the compliance and non-compliance. Sigma Space has been working on this for some time now and is very interested in presenting a commercially viable solution based on their technology. So much so, that Sigma Space has provided a detailed analysis of the USGS Lidar Base Specification (USGS LBS) and the SPL compliance to this specification. The following are highlights and overview of that analysis and if a full analysis is required please contact Sigma Space directly.
The USGS LBS requires that laser intensity be attributed for each individual return. This is a secondary requirement that does not affect accuracy but helps with classification. SPL will need a variance for their shot channel pseudo-intensity, which is actually derived from the amount of solar noise captured by the instrument on a per pixel per shot basis. The intent of the "intensity" specification is to have an assessment of reflectivity, and clearly the reflectance of sun illumination is a good measurement to that effect. The ability of Sigma’s SPL to work during full daylight condition is a clear advantage in this regard. How this relates to what Leica, Optech and Riegl do regarding 8-bit, 12-bit and 16-bit intensity respectively may not be as big an issue as believed if the represented results can be of similar qualification as it relates to the current variations of the conventional sensors.
The nominal pulse spacing (NPS) as it relates to multiplexed pulses will have to be quantified to "pulse" spacing/density determination. Additionally, there should be no issue as it relates to NPS referring to single pass, multi-channel, first return and ANPS referred to multi-pass, multi-channel, first returns. The SPL will be able to comply with this easily. The USGS LBS specification states: QL pulse density specs are on ANPD: QL1 is 8 pulse/m2 (35cm spacing); QL2 is 2 pulse/m2 (71cm spacing). These are to be measured on a "representative" polygon at least 1km2 (equivalent to a 1000m cell)".
Currently, Sigma Space checks this using a 100 m2 and would just have to change the parameters for checking this for compliance. The USGS data void for single pass collection shall be complied with the SPL and the spatial distribution and regularity will also be complied with. Minor modifications to the testing algorithms just need to be made to conform. The requirement of the fully complaint LAS 1.4 standard should be easily accomplished but modifications to adjusted time information as it relates to week time will have to be made but conventional LiDAR had to do this once too. Multiple returns from a given pulse will be stored in sequential order as required by USGS LBS. Currently, SPL data is not stored this way but it will be as a result of the fully complaint LAS 1.4 update.
The USGS LBS specification states "The spatial distribution of geometrically usable points will be uniform and regular … collections shall be planned and executed to produce an aggregate first return point cloud that approaches a regular lattice of points, rather than a collection of widely spaced, high-density profiles of the terrain." SPL routinely produces 10 x 10 arrays, nominally spaced at 50 cm between points, or closer. The grid, even before any "gridding" looks quite regular. , While SPL systems use a conical scan. It would seem that this specification–and how it relates to the SPL–will require additional analysis based on the current conventional scan patterns, but since it is not different that a Palmer scan it should not present any issues.
The positional, relative and absolute accuracy considerations as specified in the USGS LBS and how they relate to the SPL need further examination. How these are interpreted as it relates to the resulting data regardless of type of sensor used may not require a variance for the SPL sensor and it would be presumptive to say otherwise. Sigma Space has indicated that they may need a variance for some of the accuracy testing criteria, but only as it relates to solar noise. This may not be the case given the SNR (Signal to Noise Ratio) as it relates to collected points and all noise points whether that is solar noise, multi-path or any other noise and how it relates to the statement "isolated noise is expected within sample areas and will be disregarded" What is the measurement . of "isolated Noise"? How is solar return accuracy measured? These are noise as it relates to the intended 3-D model but are they accurate? Since these returns are considered noise and the ratio of this noise as it relates to the information collected by the SPL versus similar noise and information collected by a conventional LiDAR, may or may not be proportional, does this mean this noise is or is not part of the positional and relative accuracy of the SPL or Conventional LIDAR as it relates to the USGS LBS?
The conclusion to this will require more involvement within the LiDAR profession to determine how this complies or doesn’t comply. It should be noted that in the evolution of conventional LiDAR this author has observed some significate solar noise in accepted LiDAR data sets over the past 19 years. The amount of noise as it relates to the actual usable returns was in the neighborhood of between five and ten percent. The report positional and relative accuracy of these data sets was well within the required stated accuracies for these jobs. This is not to say these data sets would pass the USGS LBS requirements and I am not sure as to the percentage of noise to useable data as it relates to any Photon LIDAR. It would just be wise as a community to assess this further as it relates to what we do now and what we actually are doing and will be doing since this new technology including the SPL sensors are part of us now. Absolute Vertical accuracy should be complied with on any LiDAR sensor and it has been indicated that the SPL sensors comply with this requirement.
All the other USGS LBS requirements should be met when using an SPL sensor regardless of type. These specifications would include but are not limited to LAS withheld flag, LAS overlap flag, point classification, Classification accuracy, metadata, raw point cloud, classified point cloud, bare earth surface Raster DEM, breaklines including hydro-flattening and hydroconditioning and reporting as required, by the USGS LBS specification.
Sigma Space with its SPL sensors and their ability to familiarize themselves with the commercial LiDAR profession provides some refreshing optimism and it would appear that they are a lot closer than previously believed. There is still an uphill battle for this new technology and to lump all Photon LiDAR into one bucket is nave to do. As stated several times before, whether it be conventional LiDAR technology, Geiger Mode LiDAR, or Single Photon LIDAR, it is very important to get as much information from as many different professionals as possible. The SPL sensors provided by Sigma Space have their place in the LiDAR profession and this technology continues to be exciting and impressive. And even more so as we will have more tools in the tool box to provide the users with more options to meet their requirements in innovative and effective ways.
James Wilder Young (Jamie) is currently a Senior Geomatics Technologist for Merrick & Co. located in Greenwood Village, Colorado. His experience includes all aspects of LiDAR including sensor development, applications development, data acquisition, data processing and project management.
A 1.640Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE