1.

INTRODUCTION

Knowledge of atmospheric winds is fundamental to Meteorology, Climatology, Atmospheric research as well as for Navigation. The wind techniques relevant to the work described here can be summarized in:

The DWL technique of Aeolus is the most performant but also the most complex and challenging for both transmitter and receiver specifications. The transmitter laser operates in the UV @ 355 nm adequate energy, of around 100mJ, to ensure winds up the stratosphere. The requirements of spectral & thermomechanical stability are very stringent & demanding as well as the underlying issues of optical contamination & damage are severe at such energy level in the UV and in vacuum. The receiver focal plane itself, also relies on tightly specified optics & filtering components as well as durability & stability in environment & UV light. Last but not least, the associate data processing & wind retrieval algorithm is matching the system in complexity & susceptibility to errors & bias.

The need for a simpler system has motivated this work. A wind lidar based on simplicity but also flexibility for an agile Lidar system whose basic concept can be adapted to a multitude of space applications: from a relatively larger system for winds, to smaller & compact systems for altimetry, imaging, navigation, planetary mapping or landing assistance, etc. Basically, the concept proposed is based upon a scaling Tx, an array detector in photon counting, and a matched Data Processor whose Firmware, algorithm & software can be adapted to the multitude of applications.

Here we present a wind lidar concept able to provide vectors wind in the troposphere. The winds are extracted via cross-correlation of Point Cloud Images generated by a scanning (for a ground systems), or a multi-line-of-sight (LOS), elastic backscattering lidar. For Space, such instrument can be complemented by a (passive) Doppler Wind Imager in Limb Sounding Mode detecting O2 lines to cover the altitude range: 10 to 30 km⁵. The elastic lidar operates on Mie, Aerosol scattering only while the uniformly distributed, molecular scattering is filtered out on the receiver by a Rayleigh filter. Motion of atmospheric aerosol produced by winds are then translated into 3-D local wind information via cross-correlation of density variability in sequentially scanned or satellite moving atmospheric pixels. Cross-correlating all image pixels provide the required wind filed distribution. Hence Correlation Wind Lidar (CWL).

2.

THE CORRELATION WIND LIDAR CONCEPT

The CWL lidar relies on a low-energy high repetition rate infrared or visible laser (but a UV laser is used in the demonstrator here discussed) sending beams of ~ 0.1-1 mJ energy level @ 1-10 kHz repetition rate.

The receiver is based on a multichannel PMT detector array with single-photon-counting capabilities (in this study). The generated data sets, organized in matrices with indexes corresponding to pixel range & time, are then fed into a Processor performing cross-correlation between adjacent Matrices, A & B. In this work the DSP (Digital Signal Processing) package of MATLAB is being used. In this case, maximizing the elements cross-correlation provides the correlation coefficients whose indexes values provide the position of the pixel, in space & time, as it moves between A & B. A formula does then provide the sought vector wind components.

Here the Transmitter requirements are very relaxed when compared to a DWL system: a multi-mode longitudinal laser beam with no particular frequency stability requirements, of relatively low energy & high repetition rate is sufficient: AKA microlidar. So it is with the Receiver & Processor: here photon counting capability of less than one photon per pulse is required and for the Processor, an adequate Storage Capability is needed to handle the large data set generated and needed for processing: storage capacity of approx. 1 GB is deemed adequate for the cases studies here.

Fig. 1 gives a schematic of the CWL here studied, while Fig. 2 shows the concept for a wind space system.

Figure 1.

Schematics of the CWL instrument studied here: the transmitter laser scan the Atmosphere via an Azimuth scanning mirror. The Received Photons, are filtered of the Rayleigh components and detected/processed for wind acquisition.

Figure 2.

Schematics of the Wind Space concept. The Combo-Instrument satellite includes : the Multi line-Sight (two here) CWL + a Doppler Imager in Limb mode sounding the molecular Oxygen lines O2 in the altitude range 10-30 km (as HRDI in NOAA UARS wind Satellite)

Subsequent Figures 3, 4 & 5 provide details of the actual CWL lidar system used in this work to demonstrate the CWL concept. The lidar is operated at Park Cottage Observing Station, located South London, West Sussex County. It is a rural area, subject to urban aerosols flowing out of London⁶.

Figure 3.

The CWL used in the experiments. It can be seen, n the left, the small transmitter attached to the lidar frame together with the receiver telescope. To the right, it is visible part of the receiver electronics. The laser transmitter is a commercial French-built TEEM laser, multi-mode, un-stabilized, providing o/p energy of ~ 20μJ @ 1 KHz @ 355 nm. On top of the laser it can be seen beam expander with a Newport Scanner & mirror mount.

Figure 4.

The figure shows the MEADE telescope, two 45 degree mirrors that turns the bistatic into a monostatic optical lidar configuration, the elliptical scanning mirror. Not present in this configuration is the bandpass & Rayleigh filter.

Figure 5.

The figure shows on the left, a Photek 16 Channel array PMT-MCP array detector and electronics and, on the right, the 16 Channel Eventech single-photon-counting board electronics & preprocessing

3.

CWL DATA SETS & WIND PROCESSING: PRELIMINARY RESULTS

The PMT-MCP detector output electronics provides signal intensity profiles versus height and azimuth angle (equivalent to time in this configuration). Data on-board pre-processing is first carried out and consists in: a) data acquisition from all laser pulses, b) pulse averaging, c) R² multiplication & normalization, d) background calculation & subtraction. These data so generated are then used for wind processing. This is performed via NORMXCORR2 routine from MATLAB DSP package, but DSP also permits other functions, such as smoothing & filtering, etc. Figure 6. shows an example of CWL data file as generated by the Point Cloud Data file and ready for processing.

Figure 6.

The figure shows CWL data files: either as xls (as shown) or csv format

The CWL Lidar is typically operated at night to reduce solar background & noises. The scanner run continuously and scan a complete rotation of 360 degrees in 33 seconds, at a constant elevation angle of 40 degrees.

For our Lidar system, Telescope overlap occurs at about 50 meters, so the data are being considered as from this height range. The range bin can vary from 5 to 10 meters and 1000 pulses are being integrated in each lidar channel profile, 4096 range bins are generated corresponding to 20 km to 40 km Data sets. The first two bins, out of 4096, are not used for the correlation, due to an artifact induced by the electronics. Fig. 7 below show pictorially the determination of the wind in Polar coordinates together with the formula used for the determination of the three vector wind components. The two concentric circles shown, correspond to two adjacent scans. In the example shown, the pixel (indicated as a star) moves outwards to the right due to the wind, which is calculated once the maximum of the correlation coefficient is known. The indexes i and j, provide in fact the new range & azimuth position due to wind. The arrows in Figure 7 indicate the correspondence between correlation matrix indexes and pixel displacement.

Figure 7.

Pictorial view of wind vector calculation in CWL in polar coordinates. Left shown the motion of the aerosol feature moving with the wind during the scan and on the right it is shown the cross-correlation resulting Matrix with coefficient indexes providing the new coordinate position if the pixel.

Examples of wind calculations are now shown. Many Data sets were accumulated over the period May to November 2020. After this date the system went into refurbishment. In fact, after months of operation, the laser suffered UV optical degradation of the harmonic crystals which needed to be replaced. The energy had dropped to below 10 μJ as from September with the CWL still operating, but laser energy recovered to 25 μJ after repairs.

During the CWL runs over the May-November time frame, many data files were processed and atmospheric winds generated. Two show cases, run A & run B, are presented as examples of CWL winds. Both cases refer to dry and clear weather conditions on May and September 2020 runs. In both cases visibility was good for a suburban London location with a clear sky and no apparent clouds.

In Figure 8, it can be seen that the max correlation coefficient follows exactly the lidar profile. After an initial saturation the correlation coefficient drops then raises again in correspondence to probably a light cloud or urban aerosol coming from London⁶. The saturation of the cross-correlation coefficient is associated also with a large scale azimuth invariance (> 100s meters), suggesting denser urban aerosols out from London. The Figure 9 gives the corresponding Point Cloud Image while Figure 10, provides the wind calculated for this set.

The Figure 11 provides the wind calculated at September run. These two cases are given as example of wind blowing with or opposite to the Scanner direction (positive or negative winds in the graphs).

Figure 8.

Picture shows the magnitude of the correlation coefficient (blue curve) corresponding to the lidar profile (red curve).

Figure 9.

Picture shows the Point Cloud Image corresponding to above

Figure 10.

Picture shows the calculated vector magnitude of the vector wind, left with the corresponding atmosphere point cloud image. From the picture it can be seen that the wind runs opposite direction of the scanner.

Figure 11.

This run is shown for comparison to the wind in Figure 10, to show that in this case the wind blows in the same direction as the scanner rotation.

To further understand the nature and the role of the scattering aerosols at low altitude and its relative azimuth invariance, an empiric methodology has been developed, and is being used to solve the lidar equation and calculate aerosol & molecular extinction and scattering coefficients. This empiric method, based on actual lidar power profiles and local temperature & pressure, will be reported in a future publication. In Figure 12 below it is plotted the calculated backscattering & extinction coefficients for the run A (units are for backscattering: m.sr^-1 and for extinction: m^-1). From the plots shown it can be seen that the values of backscattering coefficients versus height well match the values measured in Ref. 6 for this London sub-urban area.

Figure 12.

Calculated scattering & extinction coefficients corresponding to Figure 8.

4.

PRELIMINARY CONCLUSIONS & ACKNOWLEDGMENT

The CWL lidar technique with DSP processing has demonstrated that it can provide atmospheric winds, offering significant system simplifications in terms of Hardware, Software & Processing than a DWL. These preliminary results show that further improvements on the present system concept can be obtained using a Rayleigh filter, to suppress molecular scattering, and the use of IR or Vis laser wavelengths to further reduce Molecular scattering. Although ~10 μJ @ 1kHz level has provided good result, energies in the range of ~25 μJ and/or higher repetition rate can significantly increase SNR and range.

For Space, a system study for the combination of CWL with a passive Doppler Imager in limb sounding is under consideration to fine tune a possible space wind instrument.

This work has been partially funded by ESA under PECS contract No. 4000130586/20/NL/SC.