Wildfire monitoring using polarization sky images of neutral points

Abstract

Devices, systems, and methods for the detection of aerosol-generating events, such as wildfires are described. One example method for detecting a presence of an aerosol-generating event includes determining a current position of a neutral point in a sky based on polarimetric measurements conducted by a ground-based polarimeter at one or more measurement wavelengths and at one or more time instances. The method further includes obtaining information corresponding to baseline position of the neutral point, where the baseline position associated with a known atmospheric condition, and comparing the current neutral point position to the baseline neutral point position to determine a presence of the aerosol-generating event.

Description

PCT Application Attorney Docket No: 044974.8148. WO00(UA25-200)WILDFIRE MONITORING USING POLARIZATION SKY IMAGES OF NEUTRAL POINTS CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This patent document claims priority to and benefits of U. S. Provisional Patent Application No. 63 / 760,873 entitled ‘WILDFIRE MONITORING USING POLARIZATION SKY IMAGES OF NEUTRAL POINTS” filed on February 20, 2025. The entire content of the aforementioned patent application is incorporated by reference as part of the disclosure of this patent document.TECHNICAL FIELD

[0002] This patent document relates to devices, systems, and methods for the detection of events, such as wildfires, based on measurements of the sky.BACKGROUND

[0003] Common techniques for detecting wildfires rely on humans in watchtowers, satellite imagery, and strategically placed sensors or cameras. These techniques have shown the ability to detect wildfires within 24 hours or as short as 1 minute after ignition. However, there are several disadvantages to these methods. Watch towers and satellites are limited in spatial coverage and require the wildfire to be large enough for detection at large distances. Sensors and cameras offer increased spatial coverage but require that the fire be within the field of view of the camera or the coverage of the sensor.SUMMARY

[0004] Devices, systems, and methods for the detection of events that affect atmospheric turbidity, including wildfires and other aerosol generating events are described. In some aspects, the present technology uses changes in the position or trajectory of neutral points in order to detect the presence of a fire or smoke. Notably, the disclosed technology is able to detect the presence of a wildfire without requiring that smoke be within a field of view of a camera.

[0005] One example method for detecting the presence of an aerosol-generating event includes determining a current position of a neutral point in a sky’ based on polarimetric measurements conducted by a ground-based polarimeter at one or more measurement wavelengths and at one or more time instances; obtaining information corresponding to aPCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)baseline position of the neutral point, where the baseline position is associated with a known atmospheric condition; and comparing the current neutral point position to the baseline neutral point position to determine a presence of the aerosol-generating event, such as a wildfire.BRIEF DESCRIPTION OF THE DRAWINGS

[0006] FIGS. 1A and IB show data plots of the scattering phase functions for Rayleigh and Mie (m = 1, (na= 1.5)) plotted over scattering angle, 6.

[0007] FIGS. 2A-2C show example contour maps of polarization orientation relative to the Sun or Anti-Sun.

[0008] FIG. 2D shows a schematic illustration of the four neutral points.

[0009] FIG. 3A shows a schematic illustration of the polarization orientation of sunlight at two positions in the sky.

[0010] FIG. 3B shows a data plot showing the angle and degree of linear polarization at various points in the sky.

[0011] FIGS. 4A and 4B show data plots of the degree of linear polarization (DoLP) and angle of linear polarization (AoLP).

[0012] FIGS. 5A and 5B show data plots showing the angle and degree of linear polarization at various points in the sky and the positions of neutral points.

[0013] FIG. 6 shows a photograph of an example ULTRASIP system.

[0014] FIGS. 7A and 7B show data plots showing the 0 and U Stokes parameters, respectively, normalized by the total irradiance.

[0015] FIGS. 8A and 8B show data plots showing weighted least-squares (WLS) regression to determine the altitude and azimuth of the neutral point.

[0016] FIGS. 9A and 9B show data plots showing the log(DoLP) and AoLP [°], respectively, of the Babinet neutral point.

[0017] FIG. 10 shows data plots showing Sun and Babinet neutral point positions recorded over 10 hours.

[0018] FIG. 11 A shows a data plot comparing the estimated Babinet neutral point azimuth and Sun azimuth angles.PCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)

[0019] FIG. 11B shows a data plot showing the variation in altitude difference between the Sun position and Babinet position relative to time.

[0020] FIG. 12A shows a data plot showing the altitude difference between the positions of the Sun and the Babinet neutral point measured over three different days.

[0021] FIG. 12B shows a zoomed in view of the data plot shown in FIG. 12A.

[0022] FIGS. 13A and 13B show data plots showing example radiative transfer simulations of DoLP at 0.355 pm under smoky and dusty conditions, respectively.

[0023] FIG. 14 shows a schematic illustrating the BNP location, 6sa, solar zenith angle (SZA), 6s, and the behavior of h and 7|| in the vicinity of the neutral points.

[0024] FIG. 15 shows data plots showing BNP displacement 5 as a function of wavelength for representative aerosol regimes at 6s = 40° and 70°.

[0025] FIG. 16 shows data plots showing BNP displacement as a function of AOD, at 355, 440, and 550 nm for representative aerosol regimes.

[0026] FIG. 17 shows data plots showing AOD, SSA, FMF, and AE at Granada, El Arenosillo, and Evora during the Saharan dust outbreak.

[0027] FIG. 18 shows data plots showing daily simulated BNP displacement for solar zeniths 6s = 40 and 70° for Granada, El Arenosillo, and Evora.

[0028] FIG. 19 shows data plots showing AOD, SSA, FMF, and AE at Monterey, Fresno, and Table Mountain.

[0029] FIG. 20 shows data plots showing daily simulated BNP displacement for solar zeniths 6s = 40° and 70° at Monterey, Fresno, and Table Mountain.

[0030] FIG. 21 shows a data plot of simulated BNP displacement for simulated values of Hm.

[0031] FIG. 22 illustrates a set of operations that can be carried out for detecting the presence of an aerosol-generating event in accordance with an example embodiment.DETAILED DESCRIPTION

[0032] The positions of the four sky polarization neutral points are changed by the presence of aerosols (e.g. smoke, dust) anywhere within the sky dome. So, monitoring theirPCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)positions offers a promising tool for wildfire detection. The disclosed technology, among other features and benefits, uses the neutral points as optical markers to detect the presence of wildfire smoke without the smoke needing to be within the field of view of a camera. In addition, given that neutral points, such as the Babinet point substantially track the sun and are within a relatively small view angle (e.g., 10 to 35 degrees) from the sun, the disclosed embodiments can use neutral points as optical markers for detecting wildfire smoke without having to scan the entire sky.

[0033] Example embodiments can be implemented on a processor capable of providing near real-time data on aerosol content or atmospheric turbidity. Such an application can alert users to significant changes in turbidity within a specified radius — such as the arrival of wildfire smoke.Introduction

[0034] Understanding the propagation of sunlight through Earth’s atmosphere is challenging due to rapid changes in weather conditions and the dynamics of atmospheric particulates, such as aerosols. Those familiar with atmospheric scattering know that unpolarized sunlight becomes polarized upon interacting with the Earth's atmosphere. This produces a pattern of polarization that is tangential to the Sun, except for the presence of four polarization singularities, known as neutral points. The recognition of the sensitivity of the sky polarization signature to changes in atmospheric conditions has led to renewed efforts to incorporate polarized measurements of Earth’s atmosphere, particularly in the study of aerosols and clouds which alter the clear sky polarization signature.

[0035] Light propagates as a transverse wave, with the electric field vector oscillating in a plane perpendicular to the direction of propagation. This transverse nature of light can be explained by Maxwell’s equations. In the case of a plane wave traveling along the z-direction, the charge-free condition ∇·E=0 and ∇·B=0 implies that 5E’z / 5z=0, which means that changes in the electric field occur in the directions perpendicular (along x or y) to the propagation direction (z). thus confirming the transverse nature of light. Similarly, the magnetic field is transverse, perpendicular to both the direction of propagation and the electric field.

[0036] Within the transverse plane, at any fixed point in time, the electric field can be decomposed into orthogonal components, often denoted as E == E.. This composite E vector traces out an ellipse known as the polarization ellipse. This ellipse represents the path followed by the tip of the electric field vector as it oscillates over time, and is referred to as thePCT Application Attorney Docket No: 044974.8148. WO00(UA25-200)polarization of the wave. The orientation and shape of this ellipse are determined by the relative magnitudes and phases of Ex̂and Eŷ, which can be derived from the initial conditions of the plane wave equation, ensuring that the real part of the field oscillates in both time and space. The polarization state of a wave can be linear= 0° or 180°), circular = |E. |, - = 90° or -90°), or elliptical, with all other cases falling into the latter category'. The orientation of electric field oscillations can have no preferential orientation (i.e., randomly polarized), oscillate in many orientations with a tendency towards a given orientation (i.e., partially polarized), or only oscillate in a specific orientation (i.e., polarized). Sunlight does not have a preferential polarization state and is considered randomly polarized. The interactions of Sunlight with Earth's atmosphere create polarized skylight through the process of Rayleigh scattering.

[0037] Rayleigh scattering occurs when the incident wavelength (Z) is much larger than the diameter of the scattering object (d). When Sunlight interacts with the dielectric air molecules the air molecule becomes a radiating dipole source. The intensity of the scattered radiation (7) is given by Equation 1.8^4a2r / ?2- I 'I(l + cos2#)T47?2I^2+2 )(1)

[0038] Here, Io is the incident intensity in Wm~2, R is the distance from the scattering particle to the point of observation in m, a is the polarizability of the scattering particle in m3, n is the refractive index of the medium surrounding the particle, 0 is the scattering angle, the angle between the incident light and direction of observation. Thus, Equation 1 shows that / is inversely proportional to X4and decreases as R’2. In the atmosphere, the X and R'2relationships result in diffuse skylight which appears predominantly blue during daylight hours and more red and orange during sunrise and sunset. Rayleigh scattering is anisotropic, meaning the intensity and polarization of the scattered light vary over the scattering angle.

[0039] FIGS. 1A and IB show data plots of the scattering phase functions for Rayleigh and Mie (m = 1, (na= 1.5)) plotted over scattering angle, 0. 1 is shown as the outermost data trace (102), I\ as the innermost data trace (106), and Z± as the middle circular data trace (104).PCT Application Attorney Docket No: 044974.8148. WO00(UA25-200)

[0040] From Equation 1 the variation in intensity follows J a (1 + cos2(#)). Plotting / over 0 yields the plot shown in FIG. 1A, known as the scattering phase function. The polarization state is defined from the scattering plane, the plane containing the direction of incident light, and the plane containing the direction of observation. Then, the intensity can be broken into two components, i and I\\. The parallel component is proportional to cos2(rij and the perpendicular component is independent of the scattering angle, as shown in FIG. I A. When / i = I\\, the scattered light is randomly polarized at Q =0 or 180°. When J|| drops to zero, the scattered light is fully perpendicularly polarized, at 6 =90 or 270°. The scattered light is partially polarized at all other values of 0, FIG. 1A.

[0041] FIGS. 2A-2C show example contour maps of polarization orientation relative to the Sun or Anti-Sun. FIG. 2A shows a contour map corresponding to one polarization singularity, while FIG. 2B shows a contour map around two singularities. FIG. 2C shows a hypothetical scenario where there is an absence of singularities.

[0042] FIG. 2D shows a schematic illustration of the four neutral points.

[0043] FIG. 3A shows a schematic illustration of the polarization orientation of sunlight at two positions in the sky, as viewed from an example position on the ground.

[0044] FIG. 3B shows a data plot showing the angle and degree of linear polarization at various points in the sky.

[0045] The intensify and polarization of Rayleigh scattering within the atmosphere were formulated into the Rayleigh sky model by Lord Rayleigh in 1871. The Rayleigh sky model demonstrates that the polarization signature of the sky follows Rayleigh scattering expectations, with fully polarized light that is perpendicular to the scattering plane 90° from the Sun (the plane that contains the illumination and view vectors, as shown in FIG. 3A), as shown in FIGS. 2C and 3B. However, this model did not include four polarization singularities that disrupt the expectation of partially polarized light above and below' the Sun and above and below the antiSun.

[0046] FIGS. 4A and 4B show- data plots of the degree of linear polarization (DoLP) and angle of linear polarization (AoLP), respectively, as determined from all-sky polarimetry. In each plot, the neutral point is marked with a small circle.PCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)

[0047] A full description of sky polarization must extend beyond the Rayleigh model to include neutral points, four points of unpolarized light that disrupt the otherwise continuous polarization signature of the sky. At the neutral points, the degree of linear polarization (DoLP) is zero and the angle of linear polarization (AoLP) precesses 180° around the neutral point, as shown in FIGS. 4A and 4B. A polarization contribution of equal magnitude that is parallel to the scattering plane creates neutral points at distinct sky' positions.

[0048] FIGS. 5A and 5B show data plots showing the angle and degree of linear polarization at various points in the sky and the formation of neutral points as a result of the contributions of single and multiple scattering. In FIG. 5A, the polarization contributions of single and multiple scattering are shown as separate lines, while in FIG. 5B the contributions are combined into a single line.

[0049] Incoherent addition of orthogonal polarization states creates un-polarized light at the neutral points, as shown in FIGS. 5 A and 5B. Early studies concluded that multiple scattering of polarized skylight can generate parallel polarized light. The exact position of the neutral points depends on the ratio between multiple and single scattering. Surface reflections and scattering by aerosols (such as haze and smoke) can modify this ratio by contributing parallel or perpendicular polarization, thereby influencing the position of neutral points in the sky. This dependence on scattering means the neutral point positions will vary with different atmospheric conditions.

[0050] These singularities or neutral points are named Babinet (above the Sun), Brewster (below the Sun), Arago (above the anti-Sun), and The Fourth (below the anti-Sun); as shown in FIG. 2D. At these neutral points, the skylight is randomly polarized. These singularities are the result of multiple scattering of polarized skylight. Referring back to FIG. 1 A, 7j| is predominantly forward and backscattered. When the polarized skylight is multiply scattered in the atmosphere,the 7|| component of this new scattered radiation, which will be denoted with a prime,, is sent back towards the Sun and anti-Sun. Between the Sun and the neutral pointis greater than h, resulting in a net parallel polarization state, until canceling out at the neutral point, before Ii again dominates and the polarization signature returns to Rayleigh expectations, as shown schematically in FIGS. 2A and 2B. The scattering angles over which Z is greater than Z±, i.e. the position of the neutral point from the Sun, will depend on wavelength and atmosphericPCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)conditions. In general, it has been observed that the locations of these neutral points are between 15-35° above or below the Sun.

[0051] When particulates are suspended in the atmosphere, otherwise known as aerosols, they will scatter light according to Rayleigh scattering when ( / .»d) or according to Mie scattering when ( ~ d), or larger, but not yet converging to geometric optics approximations. Mie scattering is ty pically reserved for approximating aerosols as spherical scattering objects. The scattering phase function from a Mie scattering event is far more complex and involves directly solving Maxwell’s equations with solutions varying due to the index of refraction of the scattering object (wa) and the size parameter, m = itdl'. This differs from the scattering phase function of Rayleigh scattering, Equation 1, which is not dependent on the index of refraction of the scatter or the size of the scatterer. The lack of these dependencies arises from the fact that Rayleigh scattering is an approximation of Mie scattering. The Mie scattering phase function relationship to X is less pronounced and does not have a simple power-law proportion. There are multiple solutions for Mie scattering phase functions, but the scattered intensity is typically highest in forward scattering, as shown in FIG. IB. So, the intensity of the polarized scattered radiation is weaker compared to Rayleigh scattering. When the skylight undergoes Mie scattering in addition to multiple scattering, the intensity of will increase or decrease, affecting the position of the neutral point.

[0052] Aerosols vary widely in size and index of refraction, so each type of aerosol will scatter light differently. In the case of wildfire smoke, the predominant aerosols are black carbon and organic aerosols ranging in size from 1-1000 nm. Since UV and visible wavelengths are on the same order of these particle sizes, Mie’s solution or the Rayleigh approximation can be used to describe the scattering. Until now. the effects of smoke aerosols (or any other similar aerosols) on the position of the neutral point have not been systematically studied.

[0053] The Aerosol Robotic Network (AERONET) is currently the gold standard for ground-based measurements of atmospheric turbidity, providing decades of invaluable data on aerosol properties. However, AERONET has limitations, particularly in detecting scattering changes outside its line of sight and processing cloudy data, as its inversion algorithm struggles with cloud contamination. Neutral point observations at the very least can complement AERONET by offering a more flexible and rapid means of estimating atmospheric turbidity, especially when detailed aerosol properties are not required. Unlike AERONET’s reliance onPCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)direct line-of-sight measurements, neutral point positions respond to the broader polarization signature of the sky, making them effective markers of atmospheric changes. This capability is especially relevant for wildfire smoke detection. Smoke injected into the atmosphere during a wildfire can alter the polarization pattern of the sky even if the plume is not directly visible to the instrument. Neutral points are predicted to be sensitive to changes in atmospheric turbidity since the sky location where orthogonal polarization states are equal defines the neutral points. By monitoring diurnal and spectral variations in their positions, the disclosed embodiments make it possible to attribute their movements to changes in atmospheric turbidity. This makes neutral point observations a valuable tool for detecting wildfire smoke that can replace or augment traditional remote sensing methods, particularly in rapidly evolving scenarios like wildfire smoke events.

[0054] To use neutral points as optical markers of atmospheric turbidity, a highly precise measurement of the diurnal movements of the neutral point, such as the Babinet neutral point, is necessary. Typically, the position of the Babinet neutral point is reported as an angular deviation from the Sun. Previous work by others has examined the diurnal dependence of the Babinet and Arago neutral points over visible wavelengths. Their findings revealed that the azimuth angles of the neutral points coincided with the positions of the Sun and anti-Sun, while the altitude differences changed throughout the day. These position estimations are consistent with other studies, which report Babinet and Arago neutral point deviations ranging from 10° to 35° throughout the day.

[0055] Additional observations by others have identified variations in the angular deviation of the Arago neutral point from the anti-Sun when observing different wavelengths of light, with a maximum difference in angular deviation of 6.9° across the red (730 ±65 nm), green (600 ±65 nm), and blue (470 ±65 nm) spectral ranges. The largest angular deviation was reported for the blue wavelength. Similar results are reported for the Babinet and Brewster neutral points. The larger angular deviation in the blue wavelength was explained by the fact that polarization experiences strong dispersion at shorter wavelengths due to multiple scattering, which reduces the degree of polarization and shifts the locations of the neutral points. Additionally, in the ultraviolet and blue ranges, the regions of parallel polarization surrounding the Sun and anti-Sun were observed to be more extensive, leading to greater angular deviations between the neutral points. These findings, paired with reduced surface reflections in the ultraviolet and ultraviolet variations in wildfire smoke, suggest that ultraviolet wavelengths are especially suitable forPCT Application Attorney Docket No: 044974.8148. WO00(UA25-200)neutral point measurements in the context of aerosol detection. Moreover, AoLP is known to be more invariant to cloud cover in the ultraviolet range.

[0056] In some example configurations, neutral point measurements utilize fish-eye lenses to capture neutral point images, yielding valuable insights into the relationship between neutral point positions, observed wavelengths, and atmospheric conditions. However, these approaches can be constrained by limited spatial resolution and the reliance on complex algorithms to pinpoint neutral point positions.

[0057] An additional example system and method for neutral point measurements includes the use of a ground-based Ultraviolet Linear Stokes Imaging Polarimeter (ULTRASIP), specifically designed for Sun-tracking. ULTRASIP provides position estimation with high spatial resolution and low uncertainty. ULTRASIP enables imaging of the Babinet neutral point at a center wavelength of 355 nm with an instantaneous field of view (IFOV) of 7.2 arcseconds per pixel. In comparison, fish-eye systems with similar sensor sizes typically exhibit an IFOV of 360 arcseconds.Examples of Babinet Neutral Point Estimation Using ULTRASIP

[0058] FIG. 6 shows a photograph of an example ULTRASIP deployed in a forest. ULTRASIP is portable and designed to sustain prolonged outdoor use.

[0059] ULTRASIP is a division-of-time (DoT) Stokes imaging polarimeter that measures polarized flux (P). The Stokes parameters are derived from the time-average of the polarizationellipse1, where T is the total averaging time. Then the Stokes parameters of units W m2are defined in a vector, S, as described in Equation 2.IC|2+|E,| X" r 1A( p +p lei2-KI2cos(2x)cos(2< / >) P10 ~P 52X 2|Ej|E,|costy,) cos(2x) sin (20) P ~P sin(2x) j PW ~P JLHCj 2|E,||E,|s,n(tf.,),(2)

[0060] Here, So is the total intensity, S1 indicates the prevalence of x-component intensity over y-component intensity, S2 is the counterclockwise diagonal prevalence over the clockwise diagonal intensity, and S3 refers to the intensity of right-handed circular (RHC) polarizationPCT Application Attorney Docket No: 044974.8148. WO00(UA25-200)prevalence over the left-hand circular (LHC) polarization. In the context of atmospheric polarization, the latter is considered negligible, and hence discussions on atmospheric polarization only consider the linear Stokes parameters (Si, &).

[0061] Within a linear Stokes polarimetric imaging system, the Stokes parameters are determined by analyzing flux images taken at various polarizer orientations, (0 < ^ < 360), and the resulting flux measurements are denoted as P<f,. To obtain the complete linear Stokes vector, at least three linearly independent measurements are required. For optimal polarization measurements, commonly chosen values are <j> = [0, 45, 90, 135], This selection is based on considerations of geometric interpretation and hardware practicality. Each measured flux image p = I / .’ I2p I2p _ = E E* captures different aspects of the electric field such that0 1 v|’901 ' I ’43 x yand P135=Ex*Ey. The Stokes parameters are used to define a polarization state with two quantities: degree of linear polarization (DoLP) and angle of linear polarization (AoLP).

[0062] DoLP represents the ratio of polarized light to total intensity', or the randomness of the polarization state, with values ranging from 0 (randomly polarized) to 1 (fully polarized).cos (2x1 DoLP is calculated from the Stokes parameters using the formula DoLP — - — --.AoLP describes the orientation of the polarization state and depends on the choice of the reference plane. AoLP is defined as counter-clockwise looking into the beam. In sky' polarization measurements, the common reference planes are the meridian plane (spanning the zenith and the view vector) and the scattering plane (spanning the Sun illumination vector and the view vector).1 i AoLP = — arctan —2 I S I AoLP is determined from the Stokes parameters using the formula1 7and is equal to. Typically, the range of AoLP is constrained to 0° < AoLP < 180°, as AoLP remains unchanged under 180° rotations.

[0063] In one example, ULTRASIP measures P by rotating a Moxtek ultra-broadband wire grid linear polarizer. The polarizer is followed by a fused silica 78mm focal length lens paired with the SONY IMX487 12-bit back-illuminated CMOS sensor housed in an Alvium camera (1800 U-812 UV).

[0064] CMOS sensors individually convert light to each pixel, offering faster processing than a charge-coupled device type sensor (CCD) but sacrificing uniformity, sensitivity, andPCT Application Attorney Docket No: 044974.8148. WO00(UA25-200)increasing fixed-pattern noise. However, pixel-level processing reduces blooming and increases the dynamic range. This makes CMOS sensors suitable for sky imaging, as they can handle high light levels without blooming, ensuring clearer images. Additionally, their fast readout speed is advantageous for tracking dynamic atmospheric features. The SONY IMX487 sensor is fabricated for optimum quantum efficiency (QE) in the UV, specified with a maximum QE in the UV of 50.75% at 400 nm and QE above 30% from 220 nm to 400 nm. A 355 nm ± 10 nm hard-coated bandpass filter is placed in front of the sensor. Then, a 25 mm lens with a variable aperture is mounted in front of the camera and sensor. The lens elements are made from fused silica, which is highly transmissive in the UV, with transmittance above 85% across the UV spectrum. When used with the Alvium camera, the field of view (FOV) is 17.68°. or 0.0062° per pixel. The internal iris on the lens allows for -numbers ( / #) ranging from / 2.8 to / 16. ULTRASIP has a full FOV of 20808 arcseconds with an IFOV of 7.2 arcseconds operating over a 20 nm spectral range centered at 355 nm. An ultraviolet wavelength was chosen for measurements since contributions from surface scattering and cloud cover are reduced in the ultraviolet wavelengths compared to visible wavelengths. The optical system is mounted on a Moog Quickset QPT-20 pan and tilt motor with 0.25° repeatability. The pan and tilt motor enables autonomous pointing within the sky dome.

[0065] The polarized flux measurements are converted to linear Stokes parameters according to Equation 3.Q P ~P (3)1^7 P ~P where the subscripts on the flux P denote the orientation of the polarizer's transmission axis. I, 0, and U correspond to So, Si, and S? in Equation 2, respectively. In some examples, ULTRASIP takes measurements with the linear polarizer at four discrete positions: 0°, 45°, 90° and 135°. These four measurements are made in approximately 4-6 seconds. Since the neutral point continuously moves in the sky. some temporal artifacts are inevitable. In some further examples, the polarizer may be continuously rotated while collecting video rate images to enhance the speed of data collection.

[0066] For a DoT polarimeter that measures linear polarization states, ^=[0, 45, 90, 135]° the output fluxes are given by, P = IPS = [Ro, P45, P90, i35]t Here, IFis a 4x3 matrix, where thePCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)rows are the response of each polarizer orientation. The rows are not linearly independent because only the first three linear Stokes components of Equation 2 are measured. The IT-matrix of a polarimeter is estimated by polarimetric calibration. The quality of W is affected by: the spectrally -dependent extinction ratio of polarizers, total attenuation of the system, stray light and surface scattering, and field-dependent polarization aberrations.

[0067] To interpret the linear Stokes parameters accurately, it is important to select a reference plane and establish consistent rotation conventions. ULTRASIP is deployed with the instrument reference plane aligned to the solar principal plane, the plane containing the Sun illumination vector and zenith vector. AoLP is defined counterclockwise looking into the beam. Azimuths are reported from cardinal South, with clockwise rotations as positive deviations, and altitudes are reported from the horizon.

[0068] In some examples, to begin data collection, ULTRASIP is homed to the Sun position, (as determined, for example, using the SunCalc Python library), and verified with a solar quad cell and pinhole alignment system. Then, ULTRASIP samples between the Sun and zenith in steps of 2°. The exposure times may vary from 100 ps to 1 second, depending on the time of day and the deviation of altitude from the Sun. The complete step-and-stare scanning sequence takes 2-6 minutes, depending on exposure times and Sun zenith angle, during which the Sun’s position is continually updated to ensure that ULTRASIP’s azimuth position moves with the Sun. The image containing the neutral point is then identified from the set of collected images. This sequence highlights one advantageous aspect of the present technology', which is that the neutral point location and information associated with that location (such as the occurrence of a wildfire or other aerosol-generating event) can be determined by scanning only a selected portion of the sky (e.g., between the Sun and the zenith).

[0069] In a theoretical scenario with infinitesimal polarimetric and spatial resolution, ULTRASIP would measure Q = 0 and U = 0 at the neutral point. However, in practice, the variance of DoLP measurements tends to increase as the signal depolarizes; when the DoLP falls below the polarimetric resolution, the AoLP image is characterized by high-frequency variations. Rather than relying on the highly variable non-Gaussian DoLP and AoLP distributions, the linear Stokes parameters are used to estimate the position of the Babinet neutral point.PCT Application Attorney Docket No: 044974.8148. WO00(UA25-200)

[0070] FIGS. 7A and 7B are example data plots showing the Q and U Stokes parameters, respectively, normalized by the total irradiance, (7), for a measurement taken at 16:16:31 UTC on 24 September 2024. 0 and U are normalized by the total irradiance to account for the nonuniformity of the system response. For the Q / I image (FIG. 7A) the gradient is greatest along the image columns and minimal along the rows; showing variance to altitude deviations and invariance to azimuth deviations. The opposite is found for the U / I image (FIG. 7B). This aligns with the AoLP distribution expectations that depict an AoLP of 90° between the Sun and the Babinet neutral point, an AoLP of 0° above the Babinet neutral point, and an AoLP of 45° and 135° on either side of the Babinet neutral point.

[0071] The estimation technique used to determine the position of the Babinet neutral point is detailed here using a measurement taken at 16:16:31 UTC on 24 September 2024. The flux images are used to calculate the linear Stokes parameters according to Equation 3.

[0072] The invariance of Q to azimuth changes allows for the analysis of Q along the image rows, while U ’s invariance to altitude changes enables analysis along the image columns. This dimensionality reduction is achieved by calculating the row average of Q u)anc^ the column average of U (q, ) as defined in Equation 4.- _ 1 UcuL= — / — (4)' L h I

[0073] In this averaging operation, rQbecomes exclusively dependent on variations in altitude (alt), whereasis influenced solely by changes in azimuth (az). This allows for the individual determination of the altitude and azimuth of the neutral point.

[0074] FIGS. 8A and 8B show data plots showing weighted least-squares (WLS) regression to determine the altitude and azimuth of the neutral point. ULTRASIP measurements are shown with the corresponding w eighted fit-line overlaid.

[0075] In FIG. 8A the dependence of altitude on the averaged parameter 7Qis shown, which follows the equation alt = m070+ bQ,(7alt= 8.64, with an estimated uncertainty of 2.52 arcseconds. In FIG. 8B, the azimuth dependence on cvis shown, which follows the equation az = + b,., cro;= 2.16, with an estimated uncertainty of 2.16 arcseconds. The intercepts,PCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)corresponding to rQ= 0 and <2^ =0, provide estimates of the Babinet neutral azimuth at -61.1821° ± 2.52 arcseconds and altitude at 52.3693° ± 2.16 arcseconds.

[0076] A weighted least-squares regression (WLS) is performed on equations alt = moro+bQand az = mucu+bu, allowing for a linear fit that accounts for the standard deviation of the measurement parameters. These linear expressions provide an empirical method for predicting the altitude and azimuth at which the neutral flux condition — where orthogonal polarizations are equal — occurs. The x-intercepts of these regression lines, corresponding to the conditions 7Q= 0 and cv= 0, yield estimates of the altitude and azimuth of the Babinet neutral point, respectively. The results presented will report the estimation uncertainties of the regression of the altitude (o« / t) and azimuth (oaz), which are provided output from the WLS algorithm, as the Babinet neutral point position uncertainties. The WLS fitting approach, illustrated in FIGS. 8A and 8B. provides an estimated Babinet neutral point azimuth of -61.1821° ± 2.52 arcseconds and altitude of 52.3693° ± 2.16 arcseconds. Prior to implementing the WLS, outlier data points near the edge of the image were omitted.

[0077] FIGS. 9A and 9B show example data plots showing the log(DoLP) and AoLP [°], respectively, surrounding the Babinet neutral point observed on 24 September 2024, 16:16:31 UTC, with a 355 nm center wavelength. The variance increases approaching the neutral point, corresponding with a decrease in the DoLP towards zero and a rapid variation in the AoLP. The Sun was at an (Azimuth, Altitude) = (-61.2139°, 36.3711°). The circle at the center of the plots denotes the estimated position of the neutral point at (Azimuth ±Altitude ± Gait) = (-61.1821° ± 2.52 arcseconds, 52.3693° ± 2.16 arcseconds). The AoLP and DoLP images provide a visual assessment of minimal DoLP and rapidly precessing AoLP around the Babinet neutral point.

[0078] The WLS estimation technique was applied to a dataset comprising 10 hours of observations, as detailed below. A series of Babinet neutral point measurements were performed on the roof of the University of Arizona Meinel Optics Building in Tucson, Arizona, USA (32.2313° N, 110.9471° W), from 14:50 UTC to 01:03 UTC, on 24 September 2024. FIG. 10 shows data plots showing Sun and Babinet neutral point positions recorded over 10 hours. The Sun is marked by larger circles and the Babinet neutral point as smaller circles, time is reported in universal time co-ordinates (UTC). In the top plot, the Babinet point and Sun azimuths are reported from cardinal South with clockwise rotations as positive degrees. The bottom plotPCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)shows the change in altitude, where the altitude difference between the Sun and the Babinet neutral point changes over the course of the day. In this example, the Sun position was determined by the GPS position and time of day was provided by the Python SunCalc library'. The Babinet neutral point position was calculated from each measurement using the techniques described above. The azimuth angles of the Sun and the Babinet neutral point coincide during the day, with the maximum absolute difference of 1.6988° attributed to instrument alignment and increased estimation uncertainties.

[0079] FIG. 11 A shows a data plot comparing the estimated Babinet neutral point azimuth and Sun azimuth angles, where the dashed line indicates the ideal 1: 1 relationship. The minimum azimuth difference is 0.0034° and the maximum difference is 1.6988° across the observed range.

[0080] FIG. 11B shows a data plot showing the variation in altitude difference between the Sun position and Babinet position relative to time (UTC), spanning a range of 9.4275° to 25.4579° over the day with a minimum difference when the Sun reached peak altitude and maximum difference when the Sun was at a minimum altitude.

[0081] The altitude of the Babinet neutral point, like the Sun, changes over the day, but is always higher than the Sun, as shown in FIG. 10. The nonmonotonic altitude difference between the Sun and the Babinet neutral point is shown in FIGS. 10 and 1 IB. The Babinet neutral point is further above the Sun during the hours closest to sunrise and sunset, and closest to the Sun during midday. The maximum altitude difference between the Sun and Babinet neutral point was 25.4579° at 00:39:58 UTC. The minimum difference in altitude was 9.4275° at 19:25:19 UTC. Gaps in the reported trajectories occur when no measurements were taken.

[0082] As the above example measurements show-, highly precise measurements of the neutral point position can be made, which provides new opportunities for atmospheric polarization studies. For example, ground-based ultraviolet polarimetry may be used to image and estimate the position of the Babinet neutral point with sub-pixel resolution. In the examples show n, the estimated Babinet neutral point positions have estimation uncertainties ranging from 2.16 to 5.40 arcseconds for altitude and 2.52 to 21.96 arcseconds for azimuth. Observations over 10 hours demonstrated how the Babinet neutral point moves with the Sun staying at the same azimuth position, within 1.6988°, and with an altitude deviation ranging from 9.4275° to 25.4579°. The minimum altitude deviation between the Sun and the Babinet neutral point occursPCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)when the Sun is at its peak altitude and the maximum altitude deviation occurs during sunset. These findings match previous observations of the Babinet neutral point.

[0083] Table 1 tabulates <saz and Gait of the WLS corresponding to each of the measurements. For all 61 measurements, the WLS uncertainties range from a max(o., / () = 5.40 arcseconds to a min(oaft) = 2.16 arcseconds and max(oa,-) = 21.96 arcseconds to min(oaz) = 2.52 arcseconds. The minimum estimation uncertainties of the WLS regression are smaller than the IFOV of ULTRASIP demonstrating the estimation technique’s ability to estimate the neutral point position at a sub-pixel resolution. The estimation uncertainties are highest when the angular deviation between the Babinet neutral point and the Sun is lowest, see Table 1. This is expected since as Babinet is closer to the Sun the imaging task becomes more susceptible to stray sunlight. By mapping the diurnal dependence on the position of the Babinet neutral point at one wavelength, research can continue toward attributing changes in the Babinet neutral point position to changes in atmospheric turbidity.Table 1Timestamp (UTC) Gaz Galt Timestamp (UTC) Gaz Galt 14:50:13 3.96 3.60 15:00:53 3.24 2.52 15:07:59 2.88 2.52 15:15:41 3.24 2.16 15:23:16 3.24 2.16 15:38:33 2.88 2.16 15:46:03 2.88 2.16 15:52:05 2.88 2.16 16:16:31 2.52 2.16 16:42:02 2.52 2.52 17:26:44 2.88 2.52 18:05:00 5.04 2.88 18:12:00 5.04 2.88 18:16:54 5.04 2.88 18:22:35 5.04 3.24 18:29:43 5.04 3.24 18:40:34 6.12 3.24 18:45:22 6.12 3.60 18:53:45 6.84 5.40 19:02:13 18.36 5.04 19:08:08 18.36 3.60 19:13:54 17.28 3.60 19: 17:53 17.64 3.60 19:25:19 21.96 2.88 19:31:02 18.36 2.88 19:36:14 17.28 3.60 19:41:52 15.12 2.88 19:46:36 13.68 3.24 19:49:43 15.12 3.24 19:52:38 13.68 3.24 20:43:18 4.68 2.52 20:47:21 5.40 3.96 20:52:35 4.68 4.32 20:57:51 4.32 2.88PCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)Timestamp (UTC) (5az Galt Timestamp (UTC) (J az O<r / f 21:02:59 3.24 2.52 21:08:25 3.60 2.88 21:13:06 2.88 2.52 21:19:57 2.88 2.52 21:24:25 2.88 2.88 21:33:31 4.32 2.88 21:41:09 3.60 3.60 21:46:30 2.88 2.88 21:52:27 3.24 2.88 22:02:32 3.24 2.16 22:17:58 3.24 2.88 22:24:10 3.60 2.52 22:33:36 3.24 2.52 22:46:56 2.88 2.16 22:52:05 2.88 2.52 23:00:26 3.24 2.88 23:11:59 3.24 2.16 23:21:12 3.24 2.52 23:31:58 3.60 2.52 23:43:44 3.24 2.52 23:53:50 3.24 2.52 0:05:06 3.60 2.52 0:15:51 3.60 2.52 0:28:21 3.96 2.88 0:39:58 3.60 2.88 0:51:41 3.96 3.24 1:03:55 5.40 4.32Examples of Changes in Neutral Point Location Over Time

[0084] FIG. 12A shows a data plot showing the altitude difference between the positions of the Sun and the Babinet neutral point measured over three different days. The measurements were performed on September 24, 2024, February 5, 2025, and February 6, 2025. These measurements illustrate that the position of the neutral points relative to the Sun can change over time.

[0085] FIG. 12B shows a zoomed in view of the data plot shown in FIG. 12 A, focusing on data points collected on February 5 and February 6. While the altitude differences measured on these days are more similar to each other compared to the measurements from September 24, this plot indicates that there are measurable day-to-day changes in the position of the Babinet neutral point relative to the sun.Example Simulations Illustrating Neutral Point Sensitivity to Aerosols

[0086] The Earth's atmosphere is an optically turbid medium composed of molecules, aerosols, clouds, and trace gases. Among these components, aerosols are a primary contributorPCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)to atmospheric turbidity, modifying both the magnitude, angular structure, and partial polarization of scattered radiation. Through aerosol-radiation and aerosol-cloud interactions, aerosols alter surface irradiance, influence regional weather and climate, and degrade air quality with direct consequences for human health. Fine particulate aerosols, particularly PM2.5, are of concern because their small size allows them to remain suspended in the atmosphere while also posing significant health risks. Wildfire events are a major source of PM2.5, injecting large quantities of fine-mode carbonaceous aerosols that substantially increase atmospheric turbidity over regional to continental scales. In contrast, coarse particulate aerosols, such as PM10 are primarily associated with dust storms, which introduce large mineral particles that degrade visibility and alter atmospheric scattering properties. Together, wildfire, smoke and dust storms represent aerosol perturbations that substantially increase atmospheric turbidity and modify radiative-transfer conditions. These air-quality hazards have motivated dedicated satellite missions, such as NASA’s Multi-Angle Imager for Aerosols (MAIA), which aims to improve global monitoring of aerosol properties and quantify their impacts on human health. These airquality hazards motivate the need for continuous monitoring of atmospheric turbidity.

[0087] From a remote-sensing perspective, atmospheric turbidity is an optical property that describes the strength and angular structure of scattering and absorption, shaping the sky light radiance and polarization field. The polarization pattern of the sky and location of neutral points is sensitive to aerosol microphysics, providing complementary constraints to intensity -only remote sensing for atmospheric characterization and climate studies. Additionally, the locations of polarization neutral points are not fixed, but vary’ with observation wavelength and solar geometry, reflecting the combined influence of spectral scattering properties, solarobserver geometry, and aerosol scattering and absorption processes. As aerosol optical depth (AOD) increases, the Babinet point (BNP) generally departs from its molecular-scattering reference location; the direction of this displacement, toward or away from the Sun, is determined by how aerosol loading perturbs the relative angular responses of the parallel and perpendicular intensity in the vicinity of the neutral point.

[0088] To assess how sensitive the BNP location is to aerosol microphysical variations, we performed a physics-based sensitivity study using polarized radiative-transfer simulations.

[0089] FIGS. 13A and 13B are example data plots showing radiative transfer simulations of DoLP at 0.355 pm under smoky and dusty conditions, respectively, for different Sun altitude positions. Each panel is labeled with the positions of the neutral points (indicated by circles) andPCT Application Attorney Docket No: 044974.8148. WO00(UA25-200)the Sun (indicated by a square). The simulations indicate that the relative positions of the neutral points are dependent on the type of aerosols present in the atmosphere.

[0090] Some simulations were conducted with the Generalized Retrieval of Atmosphere and Surface Properties (GRASP) forward model in the sky principal plane at wavelengths of 355, 440, and 550 nm. Aerosol scenarios for the sensitivity study were constructed by combining four canonical aerosol models that represent distinct fine- and coarse-mode microphysical characteristics. Within this framework, we systematically varied aerosol optical parameters to quantify their individual and combined impacts on BNP location. By explicitly controlling aerosol loading, fine mode fraction (FMF), and absorption characteristics, these idealized simulations enable a structured sensitivity analysis of the physical mechanisms responsible for polarization neutral point shifts. To complement the controlled simulations, we conducted GRASP simulations driven by aerosol microphysical properties retrieved from AERONET inversion products, focusing on two representative real-world events: a Saharan dust outbreak in 2007 and a major California wildfire in 2020. The dust storm and wildfire event act as case studies that provide observationally constrained examples of how BNP locations arise under realistic atmospheric conditions.

[0091] To investigate how aerosol microphysical properties influence the location of the BNP in a controlled and physically interpretable manner, we constructed a modeling framework that decouples aerosol size, absorption, and contributions from fine and coarse modes. A set of representative aerosol configurations were defined, and their impacts on BNP location GBO were systematically examined.

[0092] FIG. 14 shows a schematic illustrating how' the BNP location, 0BO, and solar zenith angle (SZA), ds, are defined in the simulations described below, as well as illustrating the behavior of / ± and 7|| in the vicinity of the neutral points.

[0093] In some examples, the simulation workflow begins with the definition of candidate aerosol modes that span the primary combinations of particle size and absorptive properties. These candidates are combined into canonical fine-coarse mode pairings, and for each pairing, the aerosol volume particle size distribution (VPSD) is constructed through FMF-controlled bi-modal mixing, enabling a continuous transition between fine-dominated and coarse-dominated regimes while preserving the intrinsic properties of each mode. The resulting aerosol mixturesPCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)are then used as inputs to polarized radiative-transfer simulations to evaluate the corresponding BNP response.

[0094] Aerosol microphysical properties are represented within a modal framework, in which each aerosol mode corresponds to a physically distinct particle population described by a consistent set of microphysical parameters. For each mode, these parameters include the VPSD, the wavelength-dependent complex refractive index, and particle shape.

[0095] The VPSD provides a physically meaningful description of how particles of different sizes contribute to bulk optical properties and serves as the primary descriptor of aerosol size variability. In several widely used radiative-transfer and atmospheric-optics frameworks, aerosol microphysics is specified through VPSD, and it is commonly parameterized as lognormal modes. The bimodal formulation partitions the particle size distribution into a fine mode and a coarse mode. The fine mode typically represents particles with radii smaller than 0.6 pm, primarily originating from combustion processes and secondary aerosol formation, whereas the coarse mode corresponds to particles larger than 0.6 pm and is often associated with natural sources such as mineral dust, sea salt, or volcanic ash.

[0096] In addition to the size distribution, the vertical structure of the aerosol layer may also be specified, as it directly influences the radiative-transfer geometry and scattering contributions. In this study, the aerosol vertical concentration profile is parameterized as a Gaussian distribution with respect to altitude above the ground surface, which may be described by h (the altitude above ground level), Hm(the mean aerosol layer height), and σH(the vertical standard deviation describing the layer thickness).

[0097] Aerosol loading is parameterized by the AOD, τ(λ). Aerosol absorption is commonly characterized by the single-scattering albedo (SSA), ω₀(λ), defined as the scattering-to-extinction ratio. The magnitude of ω₀(λ) is primarily controlled by the imaginary part of the complex refractive index, κ(λ), together with particle size; these factors jointly regulate absorption strength and scattering efficiency.

[0098] While particle size governs the overall scattering regime — often summarized by the asymmetry parameter g(λ), which describes the angular structure of the phase function — it does not uniquely determine aerosol absorption. Aerosols with similar size distributions may nevertheless exhibit substantially different absorptive behavior depending on composition and mixing state.PCT Application Attorney Docket No: 044974.8148. WO00(UA25-200)

[0099] Particle-shape effects are incorporated through a sphericity factor, where = 1 denotes spherical particles and, < I denotes nonspherical particles; this parameterization is particularly important for representing coarse-mode scattering.

[0100] The aerosol microphysical and optical parameters adopted in this study are taken from the canonical aerosol property parameterizations in which aerosol optical characteristics are originally expressed as functions of AOD and wavelength. In this simulation, these canonical parameterizations are evaluated at a reference AOD of τ = 0.5, from which the representative mode parameters are derived. The resulting fine and coarse aerosol mode parameters, including the lognormal size distribution parameters and complex refractive indices, are listed in Table 2 and are used consistently throughout the forward radiative-transfer calculations.Table2Model H_m (m) σ_H (m) r (μm) σ ζ m = n + κi g ω₀ Absorbing / Smoke (fine) 3000 500 0.1388 0.4227 99.9% 1.51+0.02i 0.6803 0.9025Non- 1000 500 0.1817 0.4405 99.9% 1.42+0.00625i 0.7396 0.9628 absorbing / Urban- industrial(fine)Absorbing / Dust (coarse) 3000 500 2.2000 0.5740 0.1% 470 nm: 1.51+0.00228i 0.7798 0.8945550 nm:1.51+0.00200 / Non1000 500 2.78 0.73 99.9% 1.37+0.0001 / 0.7925 0.9716 absorbing / Oceanic(coarse)

[0101] GRASP is a highly versatile open-source radiative transfer framework that has been widely applied to a variety of remote-sensing measurements. Originally developed for the inversion of aerosol microphysical properties from AERONET multi-channel, multi-angular observations, the GRASP framework has since been substantially extended to support diversePCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)passive and active, ground-based and satellite observation geometries. In the examples discussed below, we use the open-source GRASP model in forward-simulation mode as a physics-based engine for polarized radiative-transfer calculations. GRASP couples flexible aerosol and surface parameterizations with vector radiative transfer to simulate sky polarization fields, from which neutral point locations can be estimated. GRASP has been extensively validated against AERONET and POLDER / PARASOL observations and is widely used in NASA and international remote-sensing applications.

[0102] For each fine-coarse pairing (p, q) and each FMF grid point f„, the reconstructed VPSD is used as input to the GRASP radiative transfer model in forward mode. In these simulations, the aerosol system is fully specified by the FMF-controlled redistribution between the fine and coarse physical modes, while the total optical depth τtotalis held fixed. As a result, variations in the simulated radiative and polarimetric fields arise solely from changes in the fine-coarse partitioning and the associated microphysical and optical properties, rather than from changes in aerosol loading. GRASP computes multi-spectral, multi-angular polarized radiance by¬ solving the vector radiative transfer equation for the specified aerosol state, surface properties, and viewing geometry. For each simulation, the flux-normalized (dimensionless) full-sky Stokes parameters (I, Q, U) are reported as functions of viewing zenith angle and relative azimuth angle at the prescribed solar zenith angle 6s and wavelength, rather than being expressed in absolute radiance units. From these quantities. DoLP is calculated, enabling characterization of the angular structure of the sky polarization field. In addition to the polarized radiance. GRASP products provide bulk aerosol optical properties, including the SSA (coo), asymmetry parameter g, and the Angstrom exponent (AE), corresponding to each FMF-controlled aerosol realization. Here, AE quantifies the spectral slope of AOD (AE(λ₁ / λ₂) = −ln[τ(λ₁) / τ(λ₂)] / ln(λ₁ / λ₂)), describing how AOD varies with wavelength and thereby providing an indicator of effective particle size (larger AE ty pically indicates smaller particles.)

[0103] Polarization neutral points lie in the solar principal plane, defined by the Sun and the zenith. Although the azimuth of this plane varies with solar position, the BNP remains constrained to it, so its location can be uniquely specified by its zenith angle. To characterize how the BNP responds to different aerosol conditions, we parameterize its location using the BNP zenith angle, θBa, which is defined as shown in FIG. 15.

[0104] θBais estimated by finding the zero crossing of the Stokes parameter Q. Along the principal plane, 0 is sampled every 0.5°, yielding 359 viewing-zenith samples. The algorithmPCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)examines adjacent pairs of points. (θj, Qj) and (θj+1, Qj+1), to detect sign changes that mark the occurrence of a neutral point. For each interval satisfying Qj * Qj+ < 0, the exact zero-crossing angle is computed using linear interpolation. Near NPs, the Stokes parameter Q approaches zero and ty pically has a very small magnitude, on the order of 10−4Over such a narrow angular neighborhood, the variation of Q with 6 is smooth and can be well approximated by a linear function. Given the dense angular sampling along the principal plane (Δθ = 0.5°), the use of linear interpolation to estimate the zero crossing of Q is therefore a robust approach for determining the neutral -point location. While higher-order curvature in Q(6) cannot be entirely excluded, its effect on the inferred zero-crossing angle is expected to be minimal. In the worst case, the uncertainty introduced by the finite angular sampling and linear approximation is bounded by the sampling interval and does not exceed 0.5°. A similar 0 zero-crossing detection approach has also been adopted in the ULTRASIP experimental framework (described above) and has demonstrated robust and stable performance in practical BNP observations.

[0105] To isolate aerosol-induced effects, we compute the BNP displacement relative to a molecular scattering reference at the same solar zenith angle 6s and wavelength as described in Equation 5.δ(λ, θS) = θ̃Ba(λ, θS) − θBa(λ, θS) (5)

[0106] In Equation 5, θ̃Badenotes the BNP zenith angle in the molecular scattering reference simulation, and θBadenotes the corresponding BNP zenith angle in the aerosol simulation. A positive 5 indicates that the BNP zenith angle decreases relative to the molecular reference case, corresponding to a shift of the BNP toward the zenith and away from the Sun. This shift reflects a geometric contraction of the polarization pattern toward the zenith. The 5 value may be used as a metric for quantifying how atmospheric turbidity alters BNP location as a function of wavelength and solar geometry.

[0107] We simulate sky polarization along the principal plane at three wavelengths (X = 355 nm, 440 nm, 550 nm) and two representative SZAs (6s = 40° and 70°) to probe both spectral and scattering geometric dependencies. These wavelengths are selected to span the UV-visible range and to align with both physical sensitivity and observational relevance. The 355 nm channel corresponds to the ultraviolet band used in ULTRASIP and is characterized by strong molecular scattering and reduced surface-reflectance contributions, minimizing surface and cloud influences on the sky polarization pattern. The 440 nm wavelength matches a standard AERONETPCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)observation channel. The 550 nm wavelength lies near the center of the visible solar spectrum and is widely used in aerosol and radiative-transfer studies as a representative visible wavelength. Together, these wavelengths provide sensitivity to aerosol microphysical effects while maintaining consistency with observational and modeling conventions. At wavelengths approaching the near-IR and shortwave-IR, molecular scattering becomes much weaker and DoLP along the principal plane diminishes (with a marked reduction beyond 1 pm). In this regime, surface reflection contributions become relatively more important, increasing sensitivity7to surface BRDF assumptions; therefore, these longer wavelengths are not considered here.

[0108] To ensure physical consistency, all sensitivity results are referenced to a molecular scattering reference. We obtain this reference by configuring GRASP in a near-molecular scattering limit. We set the aerosol volume concentration to an extremely small value (10−6). driving AOD far below the molecular optical depth at all wavelengths so that aerosol scattering can be neglected. All other settings (surface reflectance, site altitude, wavelength sampling, and solar geometry) are kept identical to the aerosol sensitivity-study simulations, so the amount of aerosol scattering is the only controlled difference. This molecular scattering reference provides the θ̃Bavalues used to compute BNP displacements δ and enables consistent comparisons across solar-zenith angles and wavelengths. Table 3 reports the resulting molecular scattering BNP zenith angles θ̃Bafor all θSand λ considered.Table 3A = 355 nm A = 440 nm A = 550 nm es= 40° 25.63 28.21 30.81 es= 70° 45.57 49.86 53.99

[0109] FIG. 15 provides example data plots showing BNP displacement δ as a function of wavelength for representative aerosol regimes at θS= 40° (top row) and 70° (bottom row). Columns correspond to the prescribed control-wavelength aerosol loading τ355= 0.05, 0.1, and 0.2 (left to right). Each line denotes pure aerosol type (Smoke (1502), Urban-Industrial (1504), Oceanic (1506), Dust (1508)). Overall, the magnitude of the BNP displacement exhibits a clear wavelength dependence that is aerosol- and SZA-specific. For the Smoke, Urban-Industrial, and Dust aerosols, increasing wavelength leads to a progressively more sunw ard BNP shift relativePCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)to the molecular scattering reference. In contrast, the Oceanic case behaves differently: at 6s = 70°, increasing wavelength drives the BNP away from the Sun relative to molecular scattering reference, whereas at 6s = 40° this wavelength trend is not monotonic.

[0110] The example GRASP simulations illustrate that the BNP displacement exhibits a pronounced spectral dependence. This wavelength dependence arises through changes in the particle size parameter and the complex refractive index, which together modulate aerosol scattering and polarization. FIG. 15 summarizes 5 as a function of wavelength for representative aerosol regimes under fixed control-wavelength loadings. The wavelength dependence of 5 is not a simple scaling: both the sign and curvature of the response can change with wavelength, reflecting shifts in the effective radiative-transfer balance sampled near the neutral point. At low AOD (τ355= 0.05 ~ 0.2), Smoke (~ 1°), Urban–Industrial (~ 1°), and Dust (~ 5°) become progressively more sunward (more negative δ) as wavelength increases from 355 nm to 550 nm. By contrast, for Oceanic aerosol, the BNP exhibits a distinct wavelength-dependent behavior. At SZA = 40°, δ(λ) displays a U-shaped structure over 355 nm–550 nm, whereas at SZA = 70° the BNP shifts consistently away from the Sun across the same 355 nm–550 nm range for τ355= 0.05–0.2. In both SZA cases, |δ| increases monotonically with AOD, indicating that the magnitude of the departure from θBa(mol)is progressively amplified as aerosol loading increases. At higher aerosol loading (τ = 0.2–1.0), the wavelength response of the BNP becomes aerosoltype dependent and can be distinctly non-monotonic.

[0111] FIG. 16 shows data plots showing BNP displacement, 5, as a function of AOD, r(X), at 355, 440, and 550 nm for representative aerosol regimes. The top row corresponds to θS= 40° and the bottom row to θS= 70°. Each line denotes an aerosol type (Smoke (1602), Urban–Industrial (1604), and Dust (1606)). At 355 nm, increasing AOD produces a systematic sunward shift of the BNP for all aerosol types. At 440 nm, fine mode aerosols (Smoke and Urban–Industrial) exhibit a clear non-monotonic (U-shaped) AOD–δ relationship, whereas Dust remains monotonic, with a reduced rate of change for θS= 40° beyond τ4400.3. At 550 nm, the U-shaped behavior strengthens for fine-mode aerosols, with a minimum near τ550≈0.2, while Dust shows non-monotonic behavior only for θS= 70° and remains weakly monotonic for θS= 40°.

[0112] In FIG. 16, we examine the relationship between 5 and wavelength-dependent AOD for three optically pure aerosol regimes — Smoke, Urban-Industrial, and Dust. In all cases, the fine-PCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)mode AOD decreases with wavelength more rapidly than the coarse-mode AOD. The Oceanic case is not included at these loadings, as such high aerosol optical depths (> 0.2) are rarely achieved under realistic marine conditions

[0113] At 355 nm, all three aerosol regimes exhibit a monotonic sunward displacement of the Babinet neutral point with increasing AOD, indicating a common short-wavelength response under low-to-moderate multiple-scattering conditions. At θS= 70°, the two absorption-dominated aerosol types, Smoke and Dust, show nearly overlapping AOD–δ curves at 355 nm, suggesting a similar effective control by absorption under high solar zenith geometry. As the wavelength increases to 440 nm, the AOD–δ relationship diverges by aerosol type. Fine mode dominated aerosols (Smoke and Urban-Industrial) develop a U-shaped dependence of δ on AOD: increasing loading no longer produces a uniform sunward shift. Instead, the BNP initially moves toward the Sun and then reverses direction (while remaining 5 < 0), a feature particularly evident near τ~0.3 at θS= 70°. In contrast, Dust maintains a predominantly monotonic sunward displacement with increasing AOD, although the rate of change diminishes at θS= 40° for τ4400.3, where 5 approaches approximately 11°.

[0114] At 550 nm. the non-monotonic response becomes even more pronounced for fine-mode aerosols, with a distinct minimum in δ (approximately 3°) near τ550≈0.2, underscoring the increasing role of multiple scattering and absorption in regulating the BNP location. For Dust, a comparable non-monotonic AOD–δ relationship emerges only at high SZA (θS= 70°), whereas at θS= 40° the response remains weakly monotonic.

[0115] In addition to simulating the effect of hypothetical aerosol loadings on the neutral point location, we also investigate how the BNP location is perturbed by two major aerosol producing events: the 2007 Spain dust storm and the 2020 California wildfire. The AERONET aerosol inversion products provided during these events is used as input to constrain the GRASP simulations.

[0116] AERONET is a global network of automated sun photometers that measure direct solar radiance and scan the sky to obtain directional radiances. Each instrument operates with a 1.2° field of view, maintains zenith-pointing accuracy better than 0.1° (occasionally up to 0.3°), and records radiances at eight wavelengths (340, 380, 440, 500, 675, 870, 940, and 1020 nm). Measurements are taken every 5-15 min, yielding AOD, with uncertainties of 0.010-0.021; cloud-affected observations are screened in the Level 1.5 data product. The AERONET inversionPCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)algorithm retrieves aerosol microphysical properties using four wavelengths (440, 675, 870, and 1020 nm). The Level 1.5 inversion provides VPSD, m = n + ZK, and all of which serve as a priori constraints for the GRASP forward simulations. Reported uncertainties include 25-100% for the volume size distribution, ±0.04 for the real part of the refractive index, and 30-50% for the imaginary part, as estimated from sensitivity analyses of the AERONET inversion algorithm. In Version 3, ®o uncertainties are typically below 0.03 under clear-sky conditions. All Level 1.5 and inversion products are publicly available through NASA data repository.

[0117] Since AERONET does not provide mode-separated mor but only mixture-weighted values that aggregate the fine and coarse modes, we applied these bulk values identically to both modes in GRASP while retaining distinct VPSD parameters for each mode. Besides, because AERONET retrievals do not provide aerosol vertical profile information, we prescribe an empirically representative vertical distribution. Dust and biomass-burning smoke commonly occur as transported aerosol layers at altitudes of approximately 2-5 km. Accordingly, a Gaussian profile with Hm= 3000 m and σH= 500 m is prescribed for both modes, representing a t pical elevated trans-ported aerosol layer. The observation geometry, landsurface, and solar-geometry settings are identical to those described previously.

[0118] Between 3-8 September 2007, a major Saharan dust intrusion affected southern Spain, as documented by lidar and AERONET observations in Evora, El Arenosillo, and Granada. The dust storm was characterized by the advection of mineral dust layers between 2-5 km a.s.l., producing strong enhancements in column aerosol load and distinct changes in the optical and polarization properties of the atmosphere. The rich AERONET dataset from this dust storm provides detailed measurements of dust emission, transport, and removal, enabling a process-resolved investigation of how aerosol evolution modulates BNP behavior.

[0119] Under typical Saharan dust conditions, the aerosol population is dominated by non-spherical coarse mineral particles lofted from the Sahara, with only a minor fine-mode contribution The fine mode, which may include hematite-rich and soot-like components, contributes disproportionately to absorption and therefore tends to exhibit lower singlescattering albedo and larger imaginary refractive index. In contrast, the coarse mode — composed primarily of irregular, non-spherical aggregates of Saharan mineral dust (e.g., quartz and feldspar) — is predominantly scattering, while still allowing for weak mineral absorption; accordingly, it typically exhibits a higher single-scattering albedo and a smaller imaginary refractive index. FIG. 17 summarizes the daily evolution of bulk aerosol properties (reported at 440 nm) for 2-PCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)11 September 2007. We interpret BNP displacement by pairing these property changes with the contemporaneous 5 evolution shown in FIG. 18.

[0120] FIG. 17 shows data plots showing daily AERONET retrievals of AOD (panel a), SSA (panel b), FMF (panel c), and AE (panel d) at Granada (circle markers), El Arenosillo (square markers), and Evora (triangle markers) during the Saharan dust outbreak. Vertical lines mark the onset and termination of the dust event.

[0121] FIG. 18 shows data plots showing daily simulated BNP displacement for solar zeniths θS= 40° (1802) and 70° (1804), at AERONET sites arranged by latitude along the dustplume transport direction (columns). Rows correspond to 355, 440, and 550 nm. Vertical lines mark the onset and termination of the dust event. The BNP exhibits displacements of several degrees, reaching up to 7° toward the Sun at 355 nm during peak aerosol loading. The largest displacements occur within the main transport period, with systematically weaker responses at longer wavelengths and a greater dependence on solar zenith.

[0122] Before the Saharan dust arrival (1-3 September), aerosol loading was T440nm«0.2-0.3, and the column was fine-mode dominated (0.7 < FMF440 < 0.9), with AE(440 / 675) «1.0-1.3 and moderately absorbing conditions (0.84 < coo, 440 nm <0.94). During this background period, 5 remains comparatively weak and shows only modest day-to-day variability, with BNP displacements of approximately -2° at 355 nm and approximately -4° at 440 nm.

[0123] At the onset of the Saharan dust intrusion, (4-5 September), τ440 nmincreased while FMF440rapidly decreased, indicating a transition toward a coarse-mode mineral-dust mixture; in the same interval, the 5 time series at 355 nm and 440 nm shows a clear, time-coherent departure from the pre-event behavior. The dust storms peaked on 7 September at Granada and El Arenosillo, when τ440nmexceeded 1.0 and FMF440dropped below 0.45 (FIG. 17); by contrast, Evora (farther northwest) shows a delayed response, with elevated loading persisting and the local peak occurring around 9 September. During this peak-loading, coarse-mode-dominant state, the BNP exhibits the largest displacements, reaching approximately −5° at 355 nm and −6° at 440 nm, towards the Sun (FIG. 18). In contrast, at 550 nm, δ shows a reversal relative to the peak-loading extremum, shifting away from the Sun under high aerosol loading conditions (τ440~ 1.2), consistent with the non-monotonic AOD dependence identified in the sensitivity study discussed previously. Satellite data additionally confirms the presence and progressive northward transport of a dense Saharan dust plume over the region during this period. Elevated AIRS dustPCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)scores coincide spatially with the AERONET sites and temporally with the observed BNP extrema, supporting the interpretation that the large BNP displacements are directly driven by the advection and peak loading of coarse-mode mineral dust in the atmospheric column.

[0124] Following 7 September, AOD decreased while FMF increased to over 0.7 as the dust signal weakened; correspondingly, the magnitude of δ diminished as aerosol loading decreased toward pre-event levels. After 10 September, AERONET retrievals were unavailable due to cloud contamination. Overall, the observed coevolution of aerosol properties and BNP displacement provides a direct consistency check on the dust aerosol results from the sensitivity study: periods of high dust loading, characterized by elevated AOD and reduced FMF, exhibit the largest BNP displacements, whereas periods with lower aerosol loading show correspondingly weaker BNP responses.

[0125] Between 12-23 August 2020, a series of intense biomass burning plumes originating from wildfires across central California affected a broad west-east transect extending from the Pacific coast to the Central Valley. Biomass burning aerosols are dominated by fine-mode particles produced by incomplete combustion, typically with FMF > 0.8. The fine mode, enriched in black and brown carbon, is strongly absorbing with coo = 0.80-0.90 at 440 nm, and refractive-index ranges of n = 1.50–1.55 and κ = 0.02–0.06. The coarse mode (ash and soil dust) contributes only a small fraction of the total column and is generally more weakly absorbing.

[0126] Near-source AERONET sites were excluded in this example because heavy biomass burning smoke loading during the fire period suppressed reliable inversions. Instead, we analyzed observations from three downwind stations — Monterey (coastal), Fresno (inland valley), and Table Mountain (elevated) — which are sufficiently distant to ensure stable retrievals but near enough to ensure the smoke would influence the observations. These locations primarily sampled transported and aged smoke, which would have optical properties that differ from fresh smoke. During transport and photochemical aging, oxidation and coating processes increase the fine-mode single-scattering albedo, shifting the aerosol toward a more scattering-dominated regime. Observations report conditions with coo, 440 nm~0.96-0.98 and refractive indices as low as m ≈ 1.48 + 0.001i. AERONET measurements at these sites documented the temporal evolution of smoke optical characteristics under different boundary-layer conditions.

[0127] FIG. 19 shows data plots showing daily AERONET retrievals of AOD (panel a), SSA (panel b), FMF (panel c), and AE (panel d) at Monterey (circle markers), Fresno (squarePCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)markers), and Table Mountain (triangle markers) before and after the August 2020 California wildfires. Vertical lines mark the onset of the biomass burning event.

[0128] FIG. 20 shows data plots showing daily simulated BNP displacement (5) relative to the Rayleigh reference for solar zeniths 9s = 40° (2002) and 70° (2004) at three California AERONET sites (Monterey, Fresno, and Table Mountain) during 13-23 August 2020. The vertical dashed line marks the ignition date (16 August 2020) of major wildfires in California. After ignition, Monterey and Fresno exhibit the largest negative 5 departures at 355 nm during peak aerosol loading, whereas Table Mountain remains near background with only weak variability. At longer wavelengths (440 nm and 550 nm), 5 shows a nonlinear, threshold-dependent response to AOD, with an initial shift toward the Sun followed by a reversal at high loading; at Monterey, this reversal is most pronounced at 550 nm, where 5 changes from —5° pre-event to +7° at peak loading, while Table Mountain never reaches the reversal regime.

[0129] In FIG. 19 all three sites exhibited low aerosol loadings (T440 nm < 0.3) prior to the fire ignition, and were mostly non-absorbing at Fresno and Monterey (1.0 < AE < 1.3, 0.91 < mo < 0.99), but more moderately absorbing at Table Mountain (AE > 1.25, 0.88 < coo < 0.99), characteristic of clean marine or continental background conditions. In this regime, AOD is very low, and the BNP shifts by 1-2° toward the Sun relative to the molecular scattering reference.

[0130] Following the fire ignition on 16 August, the transported smoke gradually aged during its downwind evolution. Peak aerosol loading at 440 nm differed strongly across the three sites: Monterey reached T44o nm~4, Fresno reached T44o nm~2.3. while Table Mountain showed only a modest increase to T44o nm~0.45. These site-to-site differences are consistent with variations in boundary-layer structure, which regulate aerosol loading and vertical confinement in the observed atmosphere. At 355 nm, Monterey and Fresno, located within or frequently influenced by the planetary boundary layer, exhibited elevated AOD, producing BNP displacements of approximately 12° at Monterey and at Fresno. In contrast, Table Mountain, located at 2200 m and frequently above the boundary layer, sampled substantially lower aerosol loading. These differences are reflected in the aerosol loading and optical properties, resulting in a correspondingly smaller BNP shift of only 1-2° toward the Sun at Table Mountain. The aerosol layer altitude was held fixed in the simulations; thus, the boundary-layer influence is represented through differences in aerosol loading and microphysical properties rather than explicit changes in aerosol vertical distribution.PCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)

[0131] At longer wavelengths (440 nm and 550 nm), 5 exhibits a pronounced nonmonotonic dependence on aerosol loading, analogous to the behavior observed in the dust case study. As AOD increases, 5 initially shifts toward the Sun, but beyond a loading threshold the trend reverses, with further increases in AOD driving the BNP away from the Sun. Consistent with the high FMF and elevated t at Monterey and Fresno, these two sites show the largest-magnitude departures in 5, including intervals where 5 becomes positive. At Monterey, the reversal is particularly clear at 550 nm, where 5 transitions from a pre-event value of approximately —5° to a peak-event displacement of +7°. By contrast, Table Mountain remains in a comparatively low-loading regime (1440 < 0.42). such that aerosol multiple scattering remains weak and δ does not exhibit a non-monotonic dependence on AOD observed at higher-loading sites.

[0132] As explained earlier, the vertical structure of the aerosol layer influences the radiative-transfer geometry and scattering contributions. FIG. 21 shows a data plot of simulated BNP displacement (5) relative to the Rayleigh reference for solar zenith 9s = 40° and a wavelength of 355 nm for simulated values of Hm(the mean aerosol layer height in meters). The different lines correspond to the following AOD values as determined from AERONET retrievals on different dates at Fresno: Aug 14 (2102, τ440=0.13), Aug 15 (2104, τ440=0.16), Aug 16 (2106, τ440=0.19), Aug 17 (2108, τ440=0.31), Aug 18 (2110, τ440=0.79), Aug 19 (2112, τ440=1.87), and Aug 20 (2114, τ440=2.34). AS shown in FIG. 19, AOD increased each day between Aug 16. the date of wildfire ignition, and Aug 20 at the Fresno location. As described previously. 5 becomes more negative (NP moves sunward) with increasing AOD. FIG. 21 shows that 8, and therefore the NP position, is also sensitive to the vertical height distribution of the aerosol content, particularly for higher AOD values. For example, for the AOD values of 1.87 and 2.34, measured on Aug 19 and 20, respectively, the value of 8 becomes systematically less negative for larger values of the mean aerosol height. These results indicate that the NP position can also provide information about the mean aerosol height during an aerosol-generating event. For example, wildfire smoke is typically higher in the atmosphere (-3,000 m) compared to pollution (-1,000 m). The characteristic heights of these two types of aerosols will have distinct effects on the NP position, thereby enabling an observer to distinguish between them.

[0133] This study investigated how aerosol loading and aerosol microphysical properties modify the geometry of the skylight polarization field. We used the GRASP forward model to run a controlled sensitivity study. AOD and modal mixing were prescribed by combining canonical fine and coarse modes through the FMF. All simulations focused on principal-planePCT Application Attorney Docket No: 044974.8148. WO00(UA25-200)sky-polarization patterns. To connect the synthetic results to realistic variability, we analyzed two events. They represent contrasting regimes: a Saharan dust outbreak in September 2007 and a California wildfire smoke episode in August 2020. For each event, aerosol properties from AERONET inversion products were used as inputs to GRASP forward simulations.

[0134] Molecular scattering sets the baseline sky-polarization structure. Increasing aerosol loading generally shifts the BNP away from its molecular scattering location. In most regimes this displacement is sunward, but oceanic-dominated mixtures often shift the BNP away from the Sun. Wavelength exerts a strong influence on BNP displacement, but the response is not universally monotonic from 355 to 550 nm. At low loading (1355 = 0.1), several regimes exhibit larger-magnitude displacements toward 550 nm, whereas at higher loading the wavelength dependence becomes aerosol-type dependent and can be non-monotonic. This behavior is consistent with the rapid reduction of Rayleigh optical depth at longer wavelengths and the corresponding increase in the relative importance of multiple scattering. SZA also modulates BNP displacement by changing the atmospheric path length. The response is not uniform, and it depends on aerosol regime and particle size. We also identify configurations in which the BNP location is nearly invariant with respect to loading.

[0135] The case studies demonstrate coherent, event-scale BNP anomalies with magnitudes of several degrees. These results support BNP displacement as a physically interpretable metric for aerosol monitoring or characterization. This framework also establishes the quantitative information needed to design, specify, and tolerance NP localization polarimeters. By isolating how BNP location responds to aerosol loading, size distribution, absorption, wavelength, and solar geometry under controlled conditions, we identify both high-sensitivity regimes — where small BNP shifts are physically meaningful — and weak-sensitivity or non-monotonic regimes — where localization precision or interpretability is limited. These results directly translate into instrument requirements: the expected dynamic range of BNP displacement defines the required angular resolution and field-of-view placement; the nearinvariant and transition configurations define tolerance bounds on pointing, calibration stability, and sampling density; and the wavelength- and SZA-dependent sensitivities inform optimal spectral band selection and observing geometries for specific aerosol tasks (e.g., fine-mode smoke versus coarse dust). When combined with the event-driven case studies, the framework provides a practical basis for linking polarimeter design choices and error budgets to the physicalPCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)retrievability of aerosol information from NP localization, enabling task-driven instrument optimization.

[0136] The BNP displacement alone, at a single wavelength and solar geometry, may not be sufficient for some applications. The observed weak-sensitivity and non-monotonic regimes arise from changes in the balance between multiple- and single-scattering contributions near the neutral point, which depend strongly on wavelength and solar zenith angle. Because these dependencies differ across aerosol microphysical classes, the same atmospheric state that produces a near-invariant or non-monotonic response at one wavelength or geometry generally exhibits a monotonic and higher-sensitivity response at another. Consequently, the framework directly motivates multi-spectral and multi-geometry NP localization strategies, in which complementary wavelength and solar-zenith sampling can remove the apparent ambiguities and restores sensitivity to aerosol loading and microphysical regime.

[0137] An example method for the detection of wildfires, or other aerosol-generating events, at a given location and time may include first measuring a baseline position of a neutral point, such as the Babinet neutral point. The baseline position may be measured at a single wavelength and SZA, or the baseline neutral point position may be measured at multiple different combinations of wavelength and SZA. Alternatively, a baseline position for the neutral point may be determined by radiative transfer simulations if the typical aerosol microphysical properties of the location are known (e.g., from AERONET retrievals).

[0138] Subsequent measurements of the neutral point position may be compared against the baseline neutral point position in order to detect the presence of an aerosol-generating event. If the measured neutral point position deviates from the baseline position by more than a predetermined first threshold value, this can indicate the presence of an aerosol-generating event. For example, in the Saharan dust event described previously, the BNP position shifted by at least ±1° for most combinations of wavelength and SZA following the dust incursion, compared to the pre-event position. Similarly, for the California wildfire event, the BNP position for the Monterey and Fresno locations shifted by at least ±1-2° following the fire ignition. An appropriate first threshold value may be determined based upon the uncertainty of the measured BNP position, as well as the typical variability of the measured BNP position for a given location. For example, a lower threshold value (i.e., a more sensitive detection method) may be used in embodiments where it is possible to measure the neutral point position with a low amount of uncertainty and where the position of the BNP is fairly stable on a day-to-day basis.PCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)

[0139] For some combinations of wavelength, SZA, and aerosol characteristics (e.g., vertical distribution, AOD, SSA, FMF, ect.) the position of the neutral point can be less sensitive to aerosol loading. Therefore, in some embodiments, multiple measurements of the neutral point position are made at varying wavelengths and / or SZAs and compared against the corresponding baseline neutral point positions (the baseline neutral point positions measured or simulated at the same combination of wavelength and SZA). In this case, a second threshold value can be introduced, wherein if greater than a second threshold number or proportion of the measured neutral point locations differ from the corresponding baseline neutral point positions by the first threshold value or more, then this triggers the detection of an aerosol event.

[0140] In some embodiments, in order to increase the reliability of the detection, neutral point measurements are made across multiple physical locations and compared to their corresponding baseline neutral point measurements. If greater than a third threshold number or proportion of the measured neutral point locations differ from the corresponding baseline neutral point positions by the first threshold value or more, then triggers the detection of an aerosol event.

[0141] FIG. 22 illustrates a set of operations that can be carried out for detecting the presence of an aerosol-generating event in accordance with an example embodiment. At 2202, a current position of a neutral point in a sky is determined based on polarimetric measurements conducted by a ground-based polarimeter at one or more measurement wavelengths and at one or more time instances. At 2204, information corresponding to baseline position of the neutral point is obtained, where the baseline position associated with a known atmospheric condition. At 2206, the current neutral point position is compared to the baseline neutral point position to determine a presence of the aerosol-generating event.

[0142] In some example embodiments, the one or more time instances correspond to one or more sun zenith angles (SZAs) during a single day.

[0143] In some example embodiments, obtaining the information corresponding to the baseline neutral point position comprises measuring a position of the neutral point using the one or more measurement wavelengths during a period of time without any substantial aerosolgenerating events.

[0144] In some example embodiments, obtaining the information corresponding to the baseline neutral point position comprises performing a radiative transfer simulation using knownPCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)aerosol microphysical properties associated with a location of the ground-based polarimeter and corresponding to the one or more measurement wavelengths.

[0145] In some example embodiments, determining the current position of the neutral point comprises (a) measuring a plurality of positions of the neutral point corresponding to a plurality of different sun zenith angles (SZAs), or (b) measuring the position of the neutral point using a plurality of measurement wavelengths, and comparing the current neutral point position comprises comparing the plurality of positions of the neutral point or the position of the neutral point corresponding to the plurality of measurement wavelengths to baseline information.

[0146] In some example embodiments, the plurality of the measurement wavelengths includes at least two of: 355 nm, 440 nm, or 550 nm.

[0147] In some example embodiments, the plurality of different SZAs includes 40° and 70°

[0148] In some example embodiments, comparing the current neutral point position to the baseline neutral point position comprises determining a deviation from the baseline neutral point position, and upon a determination that the deviation exceeds a first threshold, indicating the presence of the aerosol -generating event.

[0149] In some example embodiments, determination of the presence of the aerosolgenerating event includes evaluating a change in neutral point position based on a height of an aerosol profile.

[0150] In some example embodiments, the first threshold is at least 0.5 degree.

[0151] In some example embodiments, the neutral point is Babinet neutral point.

[0152] In some example embodiments, the aerosol-generating event is a wildfire.

[0153] In some example embodiments, the presence of the wildfire is detected without requiring that smoke associated with the wildfire be in the field of view of the ground-based polarimeter.

[0154] In some example embodiments, determining the current location of the neutral point includes pointing the ground-based polarimeter within 10 to 35 degrees with respect to the Sun.PCT Application Attorney Docket No: 044974.8148. WO00(UA25-200)

[0155] In some example embodiments, the current position of the neutral point and the baseline position of the neutral point are associated with an ultraviolet measurement wavelength.

[0156] In another example embodiment, a system for detecting an aerosol generating event, comprises a ground-based polarimeter configured to conduct polarimetric measurements; and a processor coupled to the polarimeter and configured to: determine a current position of a neutral point in a sky based on the polarimetric measurements conducted by the ground-based polarimeter at one or more measurement wavelengths and at one or more time instances; obtain information corresponding to baseline position of the neutral point, the baseline position associated with a known atmospheric condition; and compare the current neutral point position to the baseline neutral point position to determine a presence of the aerosol-generating event.

[0157] In some example embodiments, the ground-based polarimeter includes an ultraviolet linear stokes imaging polarimeter.

[0158] In some example embodiments, the ground-based polarimeter conducts polarimetric measurements at an angle within 10 to 35 degrees with respect to the Sun.

[0159] While this patent document contains many specifics, these should not be construed as limitations on the scope of any invention or of what may be claimed, but rather as descriptions of features that may be specific to particular embodiments of particular inventions. Certain features that are described in this patent document in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable subcombination. Moreover, although features may be described above as acting in certain combinations and even initially claimed as such, one or more features from a claimed combination can in some cases be excised from the combination, and the claimed combination may be directed to a subcombination or variation of a subcombination.

[0160] Similarly, while operations are depicted in the drawings in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed, to achieve desirable results. Moreover, the separation of various system components in the embodiments described in this patent document should not be understood as requiring such separation in all embodiments.PCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)

[0161] It is understood that the various disclosed embodiments may be implemented individually, or collectively, in devices comprised of various optical components, electronics hardware and / or software modules and components. These devices, for example, may comprise a processor, a memory unit, an interface that are communicatively connected to each other. The processor and / or controller can perform various disclosed operations based on execution of program code that is stored on a storage medium. The processor and / or controller can, for example, be in communication with at least one memory and with at least one communication unit that enables the exchange of data and information, directly or indirectly, through the communication link with other entities, devices and networks. The communication unit may provide wired and / or wireless communication capabilities in accordance with one or more communication protocols, and therefore it may comprise the proper transmitter / receiver antennas, circuitry and ports, as well as the encoding / decoding capabilities that may be necessary for proper transmission and / or reception of data and other information.

[0162] Various information and data processing operations described herein may be implemented in one embodiment by a computer program product, embodied in a computer-readable medium, including computer-executable instructions, such as program code, executed by computers in networked environments. A computer-readable medium may include removable and non-removable storage devices including, but not limited to, Read Only Memory (ROM), Random Access Memory (RAM), compact discs (CDs), digital versatile discs (DVD), etc. Therefore, the computer-readable media that is described in the present application comprises non-transitory storage media. The instructions may be stored on memory of a local processing device, or may be stored in a remote location, such as a remote server, a cloud sever, or other networked devices and environments. Generally, program modules may include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. Computer-executable instructions, associated data structures, and program modules represent examples of program code for executing steps of the methods disclosed herein. The particular sequence of such executable instructions or associated data structures represents examples of corresponding acts for implementing the functions described in such steps or processes.

[0163] Only a few implementations and examples are described and other implementations, enhancements and variations can be made based on what is described and illustrated in this patent document.

Claims

PCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)CLAIMSWhat is claimed is:

1. A method for detecting the presence of an aerosol -generating event, comprising:determining a current position of a neutral point in a sky based on polarimetric measurements conducted by a ground-based polarimeter at one or more measurement wavelengths and at one or more time instances;obtaining information corresponding to a baseline position of the neutral point, the baseline position associated with a known atmospheric condition; and comparing the current neutral point position to the baseline neutral point position to determine a presence of the aerosol-generating event.

2. The method of claim 1, wherein the one or more time instances correspond to one or more sun zenith angles (SZAs) during a single day.

3. The method of claim 1, wherein obtaining the information corresponding to the baseline neutral point position comprises measuring a position of the neutral point using the one or more measurement wavelengths during a period of time without any substantial aerosol-generating events.

4. The method of claim 1, wherein obtaining the information corresponding to the baseline neutral point position comprises performing a radiative transfer simulation using known aerosol microphysical properties associated with a location of the ground-based polarimeter and corresponding to the one or more measurement w avelengths.

5. The method of claim 1, wherein:determining the current position of the neutral point comprises (a) measuring a plurality of positions of the neutral point corresponding to a plurality of different sun zenith angles (SZAs), or (b) measuring the position of the neutral point using a plurality of measurement wavelengths, andPCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)comparing the current neutral point position comprises comparing the plurality of positions of the neutral point or the position of the neutral point corresponding to the plurality of measurement wavelengths to baseline information.

6. The method of claim 5, wherein the plurality of the measurement wavelengths includes at least two of: 355 nm, 440 nm, or 550 nm.

7. The method of claim 5, wherein the plurality of different SZAs includes 40° and 70°.

8. The method of claim 1, wherein comparing the current neutral point position to the baseline neutral point position comprises determining a deviation from the baseline neutral point position, and upon a determination that the deviation exceeds a first threshold, indicating the presence of the aerosol-generating event.

9. The method of claim 1, wherein determination of the presence of the aerosol-generating event includes evaluating a change in neutral point position based on a height of an aerosol profile.

10. The method of claim 1, wherein the neutral point is Babinet neutral point.

11. The method of any of claims 1 to 10, wherein the aerosol-generating event is a wildfire.

12. The method of claim 11, wherein the presence of the wildfire is detected without requiring that smoke associated with the wildfire be in the field of view of the ground- based polarimeter.

13. The method of claim 1, wherein determining the current location of the neutral point includes pointing the ground-based polarimeter within 10 to 35 degrees with respect to the Sun.

14. The method of claim 1. wherein the current position of the neutral point and the baseline position of the neutral point are associated with an ultraviolet measurement wavelength.PCT Application Attorney Docket No: 044974.8148. WOOO (UA25-200)15. A system for detecting an aerosol generating event, comprising:a ground-based polarimeter configured to conduct polarimetric measurements; and a processor coupled to the polarimeter and configured to:determine a current position of a neutral point in a sky based on the polarimetric measurements conducted by the ground-based polarimeter at one or more measurement wavelengths and at one or more time instances;obtain information corresponding to baseline position of the neutral point, the baseline position associated with a known atmospheric condition; and compare the current neutral point position to the baseline neutral point position to determine a presence of the aerosol-generating event.

16. The system of claim 15, wherein the ground-based polarimeter includes an ultraviolet linear stokes imaging polarimeter.

17. The system of claim 15, wherein the one or more time instances correspond to one or more sun zenith angles (SZAs) during a single day.

18. The system of claim 15, wherein the neutral point is a Babinet neutral point.

19. The system of claim 15, wherein the ground-based polarimeter conducts polarimetric measurements at an angle within 10 to 35 degrees with respect to the Sun.

20. The system of any one of claims 15-19, wherein the aerosol-generating event is a wildfire.

21. The sy stem of claim 20, wherein the presence of the wildfire is detected without requiring that smoke associated with the wildfire be in the field of view of the ground- based polarimeter.

Images