---
title: 'Aeolus Satellite: Global Wind Lidar Mission'
url: https://www.emergentmind.com/topics/aeolus-satellite
type: topic
---

# Aeolus Satellite: Global Wind Lidar Mission

Aeolus was a European Space Agency Earth-observation mission that operated from its launch on 22 August 2018 until re-entry on 28 July 2023, carrying a single payload, the Atmospheric LAser Doppler INstrument (ALADIN), to demonstrate space-borne wind lidar and deliver global wind-profile data for numerical weather prediction [2310.08616]. It was also the first satellite to carry a Doppler Wind Lidar into low-Earth orbit, with the primary objective of providing global horizontal line-of-sight (HLOS) wind profiles for weather forecasting, data assimilation, climate research, air-quality monitoring, hazard prediction, and as a demonstrator for future operational wind lidar missions [2312.00190]. During the mission, Aeolus established the practical viability of ultraviolet Doppler wind lidar from space, while also exposing the calibration, geolocation, and optical-degradation issues that govern long-duration in-orbit performance.

## 1. Mission definition and observational role

Aeolus was conceived as a wind mission: its defining purpose was to measure atmospheric wind profiles with global coverage using ALADIN, a space-borne lidar instrument [2310.08616]. The resulting observations contributed to improving the accuracy of numerical weather prediction, and the mission’s Level-2B wind products were subsequently incorporated into broader intercomparison and validation frameworks alongside sondes, aircraft, balloons, and atmospheric motion vectors [2312.00190].

The mission timeline is central to its scientific interpretation. Aeolus launched in August 2018 and remained operational until July 2023, exceeding its three-year design life by 18 months [2310.08616]. This prolonged operation was significant because it enabled observation of both nominal performance and progressive in-orbit degradation. A plausible implication is that Aeolus served not only as a technology demonstrator but also as a long-baseline case study in the lifecycle behavior of orbital Doppler wind lidar systems.

Aeolus is often associated primarily with weather forecasting, but the mission description was broader. The Level-2B wind products were relevant to climate research, air-quality monitoring, and hazard prediction, and the mission was explicitly framed as a demonstrator for future operational wind lidar missions [2312.00190]. That dual role—operational utility and pathfinding instrumentation—became particularly evident in later analyses of signal decline and calibration.

## 2. ALADIN architecture and Doppler wind measurement

ALADIN was built around a master-oscillator power-amplifier Nd:YAG laser with third-harmonic generation at $\lambda = 354.8\ \mathrm{nm}$, operating at $50.5\ \mathrm{Hz}$ [2310.08616]. Behind the laser bench and a flip-flop mechanism that selected between two redundant flight models, FM A and FM B, the ultraviolet beam passed a photodiode for onboard energy monitoring, a quarter-wave plate to produce circular polarization, and a beam expander of $\times 3.4$ before entering a $1.5\ \mathrm{m}$ Cassegrain telescope in monostatic configuration [2310.08616]. The telescope further expanded the beam to $0.92\ \mathrm{m}$ diameter and approximately $20\ \mu\mathrm{rad}$ divergence, yielding an approximately $8\ \mathrm{m}$ surface footprint from $320\ \mathrm{km}$ altitude at $37.7^\circ$ off-nadir [2310.08616].

The measurement principle was Doppler wind lidar. ALADIN transmitted ultraviolet pulses into the atmosphere, and backscattered photons from molecules and particulates returned to the telescope, where the Doppler shift $\Delta \nu$ of the return signal was measured [2312.00190]. The relation between Doppler shift and along-beam velocity was given as

$$
v_{\mathrm{HLOS}} = (c/2\nu_0)\cdot \Delta \nu,
$$

where $\nu_0$ is the transmitted frequency and $c$ the speed of light [2312.00190].

Aeolus generated two Level-2B wind products: Rayleigh-clear winds, derived from molecular backscatter in clear air, and Mie-cloudy winds, derived from aerosol backscatter in clouds [2312.00190]. On the receive side, backscatter from molecules (“Rayleigh”) and particles (“Mie”) was collected through an $88\ \mu\mathrm{m}$ field stop, corresponding to an approximately $18\ \mu\mathrm{rad}$ field of view, and separated into two channels with full widths at half maximum of approximately $3.8\ \mathrm{GHz}$ and approximately $50\ \mathrm{MHz}$, respectively, using CCD detectors to retrieve Doppler-shifted wind speed profiles [2310.08616].

This architecture is important because later performance anomalies could not be interpreted purely from onboard energy telemetry. Aeolus had distinct transmit and receive subsystems, redundant laser flight models, and separate Rayleigh and Mie retrieval channels. The eventual diagnosis of the mission’s signal loss depended precisely on disentangling these instrument-path contributions [2310.08616].

## 3. Level-2B winds, collocation methodology, and statistical evaluation

Aeolus Level-2B winds were evaluated in the System for Analysis of Wind Collocations (SAWC), a jointly developed framework by NOAA/NESDIS/STAR, UMD/ESSIC/CISESS, and UW-Madison/CIMSS for intercomparison of winds from multiple observing platforms [2312.00190]. SAWC included a multi-year archive of Aeolus winds together with rawinsondes, commercial aircraft, Loon stratospheric superpressure balloons, and satellite-derived atmospheric motion vectors. All archived winds except Aeolus BUFR were provided in netCDF-4 with common variables such as time, latitude, longitude, height and/or pressure, wind speed, wind direction, or $u/v$ components [2312.00190].

The SAWC collocation tool ingested a “Driver” dataset and one or more “Dependent” datasets, applied user-selectable quality controls, performed four-dimensional matching in latitude, longitude, height or pressure, and time, and output netCDF index files listing matched observations and their $\Delta t$, $\Delta p$ or $\Delta z$, and great-circle distance $\Delta x$ [2312.00190]. The plotting tool then extracted matched winds, projected non-Aeolus winds onto the Aeolus HLOS direction if Aeolus was involved, applied a $\pm 25\ \mathrm{m\,s^{-1}}$ gross-check filter to wind differences, optionally super-obbed multiple dependent observations per driver, and computed statistics globally, by region, and by season [2312.00190].

The default collocation criteria were a maximum time difference of $60\ \mathrm{min}$ for Aeolus-aircraft/AMV/Loon and $90\ \mathrm{min}$ for sondes, a maximum horizontal separation of $100\ \mathrm{km}$ or $150\ \mathrm{km}$ for sondes, a maximum $\Delta \log_{10}(p)$ of $0.04$, and a maximum height difference of $1\ \mathrm{km}$ [2312.00190]. The spherical horizontal-separation metric was

$$
\Delta x = R_\oplus \cdot \arccos[\sin \phi_D \cdot \sin \phi + \cos \phi_D \cdot \cos \phi \cdot \cos(\lambda_D - \lambda)].
$$

Matches were accepted only if all four criteria were met [2312.00190].

For Aeolus, SAWC applied specific Level-2B quality controls. Mie-cloudy winds were rejected if the uncertainty exceeded $5\ \mathrm{m\,s^{-1}}$. Rayleigh-clear winds were rejected if the uncertainty exceeded $8.5\ \mathrm{m\,s^{-1}}$ for $200\ \mathrm{hPa} < p \le 800\ \mathrm{hPa}$ or $12\ \mathrm{m\,s^{-1}}$ for $p < 200\ \mathrm{hPa}$; a minimum vertical-bin thickness of $z \ge 0.3\ \mathrm{km}$ and integration length $\le 60\ \mathrm{km}$ were also enforced [2312.00190].

SAWC computed mean bias, root-mean-square error, correlation coefficient, and the standard deviation of the differences. In the one-year evaluation from September 2019 to August 2020, with Aeolus as Driver and aircraft, AMVs, sondes, and Loon balloons as Dependents, the global Rayleigh-clear statistics were: aircraft, $N \approx 9.1\times 10^5$, bias $= +0.02\ \mathrm{m\,s^{-1}}$, $\mathrm{RMSE} \approx 6.4\ \mathrm{m\,s^{-1}}$, $r = 0.93$; AMVs, $N \approx 5.3\times 10^6$, bias $= -0.02\ \mathrm{m\,s^{-1}}$, $\mathrm{RMSE} \approx 5.1\ \mathrm{m\,s^{-1}}$, $r = 0.93$; sondes, $N \approx 4.9\times 10^5$, bias $= -0.06\ \mathrm{m\,s^{-1}}$, $\mathrm{RMSE} \approx 6.5\ \mathrm{m\,s^{-1}}$, $r = 0.95$; Loon, $N \approx 7.4\times 10^3$, bias $= -0.85\ \mathrm{m\,s^{-1}}$, $\mathrm{RMSE} \approx 7.4\ \mathrm{m\,s^{-1}}$, $r = 0.84$ [2312.00190]. The global Mie-cloudy statistics were: aircraft, $N \approx 5.7\times 10^5$, bias $= +0.26\ \mathrm{m\,s^{-1}}$, $\mathrm{RMSE} \approx 5.6\ \mathrm{m\,s^{-1}}$, $r = 0.97$; AMVs, $N \approx 8.6\times 10^6$, bias $= -0.02\ \mathrm{m\,s^{-1}}$, $\mathrm{RMSE} \approx 5.1\ \mathrm{m\,s^{-1}}$, $r = 0.96$; sondes, $N \approx 9.7\times 10^5$, bias $= -0.05\ \mathrm{m\,s^{-1}}$, $\mathrm{RMSE} \approx 5.5\ \mathrm{m\,s^{-1}}$, $r = 0.96$; and no Loon comparison because there was no altitude overlap [2312.00190].

These results indicate that Aeolus wind performance was generally characterized by small global biases and high correlations after recommended quality control, with Mie-cloudy winds typically exhibiting lower RMSE than Rayleigh-clear winds in the cited period [2312.00190]. The seasonal analysis also showed gradual degradation in Rayleigh-clear precision and slightly increased bias after mid-2020, explicitly noted as consistent with known signal loss in the Rayleigh channel [2312.00190].

## 4. In-orbit degradation and the problem of signal loss

A defining technical issue in the later mission was the decline of ALADIN’s molecular return signal. Between mid-2019 and 2022, the return signal decreased by over $70\%$, degrading random wind-speed errors from approximately $2\ \mathrm{m\,s^{-1}}$ to approximately $4\ \mathrm{m\,s^{-1}}$ under clear-air conditions [2310.08616]. This degradation directly affected the Rayleigh-clear product, which depended on molecular backscatter.

On the transmitter side, onboard photodiodes recorded the FM A energy falling from approximately $65\ \mathrm{mJ}$ in early 2019 to approximately $40\ \mathrm{mJ}$ by May 2019, prompting a switch to FM B in June 2019 [2310.08616]. FM B initially delivered approximately $67\ \mathrm{mJ}$ and was tuned above $100\ \mathrm{mJ}$ by late 2021 [2310.08616]. Yet the return signal continued to fall even when onboard energy was stable or increasing, leaving unresolved whether the dominant loss was in the transmit path or the receive path [2310.08616].

That uncertainty matters because onboard energy monitoring alone did not measure the energy exiting the full optical train. A common misconception would be to treat photodiode-reported energy as equivalent to effective emitted energy at the telescope exit. The mission data did not support that simplification: subsequent ground-based measurements found true-pulse-energy estimates of approximately $33\ \mathrm{mJ}$ in 2019, below the expected approximately $48\ \mathrm{mJ}$ inferred from onboard photodiode measurements multiplied by emit-path transmission of approximately $0.77$ [2310.08616]. This suggests that the effective optical throughput of the emission path was already lower than implied by internal monitoring.

The degradation problem was therefore not only one of declining wind precision but also one of observability: the instrument’s internal diagnostics were insufficient to localize the dominant loss mechanism unambiguously. That gap motivated independent ground-based monitoring.

## 5. Ground-based observation by the Pierre Auger Observatory

An independent assessment of Aeolus was obtained from the Pierre Auger Observatory in Argentina, located at $35.2^\circ\ \mathrm{S}, 69.3^\circ\ \mathrm{W}$, whose four fluorescence-detector sites each host six ultraviolet telescopes with $13\ \mathrm{m^2}$ mirrors, a $310$–$410\ \mathrm{nm}$ bandpass, $440$-pixel PMT cameras, and a $30^\circ \times 30^\circ$ field of view [2310.08616]. During southern-winter nights from May to August in 2019, 2020, and 2021, low-aerosol clear-sky overpasses of Aeolus were recorded at $50.5\ \mathrm{Hz}$ as linear tracks of scattered ultraviolet light across the fluorescence-detector cameras [2310.08616].

The reconstruction used a monocular analysis per telescope to determine the shower-detector plane and fit the timing along the track with

$$
t_i = t_0 + (R_p/c)\cdot \tan[(\chi_0 - \chi_i)/2],
$$

where $R_p$ is the perpendicular impact distance, $\chi_i$ the pointing angle of pixel $i$, $\chi_0$ the beam inclination in the shower-detector plane, and $t_0$ the time at closest approach [2310.08616]. By intersecting multiple shower-detector planes, when available, or fixing $\chi_0$ to its average, the three-dimensional axis of each laser pulse was recovered and propagated to a reference altitude such as $10\ \mathrm{km}$ to yield ground-track positions [2310.08616].

Comparison with Aeolus Level 1A data revealed a systematic $0.075^\circ$ ($6.8\ \mathrm{km}$) along-track geolocation offset, traced to a mis-assignment of time-scale flags in the CFI software used by the Level 1A processor [2310.08616]. After correction in Processor v7.12, the residual difference between Auger and Aeolus was $0.8\ \mathrm{km}$, with pointing accuracy better than $1\ \mathrm{km}\ (2\sigma)$, well within the Aeolus requirement of $2\ \mathrm{km}\ /(2\sigma)$ [2310.08616]. This was not a minor bookkeeping issue: geolocation errors of this scale affect the spatial fidelity of collocated atmospheric measurements and therefore the interpretation of both validation and assimilation studies.

The same observations were used to estimate transmitted pulse energy at the exit of the Aeolus telescope. The analysis started from

$$
N_{\gamma,i} = E \cdot C_{\mathrm{atm},i} \cdot (\lambda/hc),
$$

with

$$
C_{\mathrm{atm}} = T_{1,R}\cdot T_{1,M}\cdot T_{2,R}\cdot T_{2,M}\cdot [S_R + S_M]\cdot \epsilon.
$$

Here the transmission factors $T_{1,R}$, $T_{1,M}$, $T_{2,R}$, and $T_{2,M}$ described Rayleigh and Mie transmission along the laser path from satellite to scatter point and from scatter point to the fluorescence detector; $S_R$ and $S_M$ were the geometry-weighted scattering cross sections per bin; and $\epsilon$ was the fluorescence-detector optical and quantum efficiency [2310.08616]. Molecular scattering followed

$$
(1/\sigma)d\sigma/d\Omega|_R = [3/(16\pi(1+2\delta))]\cdot[(1+3\delta)+(1-\delta)\cos^2\theta],
$$

while aerosol scattering was approximated by a Henyey–Greenstein phase function with asymmetry $g$ and backscatter ratio $f$ [2310.08616]. Vertical-aerosol-optical-depth profiles were measured by the Observatory’s lidar and central laser facilities within $20\ \mathrm{min}$ of each Aeolus overpass [2310.08616].

A likelihood fit over all bins adjusted $E$ to maximize

$$
L = \prod_i \sum_k [\mathrm{Poisson}(k|N_{\mathrm{exp},i}) \otimes \mathrm{Gaussian}(k|N_{\mathrm{obs},i})],
$$

thereby folding photo-electron Poisson fluctuations with PMT gain spread [2310.08616]. After applying a simulation-derived bias correction of at most $3.7\%$, the mean energies per overpass were

$$
E(3\ \mathrm{Aug}\ 2019) = 33.1^{+1.9}_{-0.8}\ \mathrm{mJ},
$$

$$
E(27\ \mathrm{Jun}\ 2020) = 23.7^{+1.7}_{-0.6}\ \mathrm{mJ},
$$

$$
E(17\ \mathrm{Jul}\ 2021) = 17.3^{+0.9}_{-0.4}\ \mathrm{mJ},
$$

with quoted uncertainties statistical only; a global $13\%$ systematic fluorescence-detector-calibration uncertainty canceled in relative comparisons [2310.08616].

Normalized to the first overpass, the Auger measurements showed a decline of $-28\%$ in 2020 and $-48\%$ in 2021, matching the ALADIN Rayleigh-channel return-signal drops of $-34\%$ and $-53\%$ [2310.08616]. Because the Auger-measured decline on the emit path mirrored the in-orbit receiver decline, the dominant loss was shown to occur before atmospheric transmission, in the transmit optics unique to FM B; this interpretation was later confirmed when swapping back to FM A in November 2022 restored the signal by a factor of $2.2$ despite lower onboard energy of approximately $50\ \mathrm{mJ}$ [2310.08616].

## 6. Scientific legacy, validation infrastructure, and future missions

Aeolus left two linked legacies: an observational one, centered on global HLOS wind profiling from space, and a methodological one, centered on validation, collocation, and external calibration. Within SAWC, Aeolus became a reference case for reproducible, multi-platform wind intercomparison. The framework’s regional and seasonal analyses showed that in the Northern Hemisphere and Tropics, Rayleigh-clear biases were generally $|\mathrm{bias}| < 1\ \mathrm{m\,s^{-1}}$, with $\mathrm{SD}_{\mathrm{Diff}} \approx 6$–$7\ \mathrm{m\,s^{-1}}$ and $r > 0.92$, while Mie-cloudy biases were $|\mathrm{bias}| < 0.5\ \mathrm{m\,s^{-1}}$, with $\mathrm{SD}_{\mathrm{Diff}} \approx 5$–$6\ \mathrm{m\,s^{-1}}$ and $r > 0.95$ [2312.00190]. In the Southern Hemisphere, Rayleigh-clear $\mathrm{SD}_{\mathrm{Diff}}$ rose to $7$–$8\ \mathrm{m\,s^{-1}}$ in jet regions, and Mie-cloudy biases grew to approximately $1\ \mathrm{m\,s^{-1}}$ in places for $p < 400\ \mathrm{hPa}$, with $\mathrm{SD}_{\mathrm{Diff}}$ up to $7$–$8\ \mathrm{m\,s^{-1}}$ [2312.00190].

The Aeolus record also captured external perturbations to the observing system. In aircraft-Aeolus collocations, counts dropped by more than $50\%$ during the pandemic seasons MAM and JJA relative to the pre-pandemic SON and DJF seasons, and $\mathrm{SD}_{\mathrm{Diff}}$ in the upper troposphere increased by $1$–$2\ \mathrm{m\,s^{-1}}$ where collocation counts fell [2312.00190]. This was interpreted in SAWC as evidence that reduced data density degrades precision and likely affects NWP [2312.00190]. A plausible implication is that Aeolus was not only a satellite mission but also a probe of the resilience of the broader global wind observing architecture.

The Pierre Auger study established three specific capabilities: an independent emit-path energy monitor, a subkilometer geolocation benchmark that exposed a $6.8\ \mathrm{km}$ offset, and a demonstration of the feasibility of monitoring space lasers from ground ultraviolet telescopes [2310.08616]. The work therefore set a precedent for monitoring future space lasers and opened new possibilities for calibration of cosmic-ray observatories [2310.08616]. The same study explicitly identified future Doppler-wind lidar missions such as EarthCARE and Aeolus-2 as beneficiaries of these methods [2310.08616].

Taken together, these results place Aeolus at the intersection of atmospheric remote sensing, instrument calibration, and observing-system design. Its principal accomplishment was to demonstrate global Doppler wind lidar from orbit. Its broader significance lies in showing that such missions require not only advanced onboard optics and retrieval processors, but also independent external benchmarks for geolocation, emitted-energy tracking, and cross-platform statistical validation [2310.08616][2312.00190].

Source: https://www.emergentmind.com/topics/aeolus-satellite