A global correction method for cosine response error of marine spectral irradiance sensor

Abstract

The present application relates to a kind of marine spectral irradiance sensor cosine response error global correction method, belong to marine optical detection technical field.The method is established by laboratory multi-angle multi-band calibration spectral segmentation cosine correction function, and obtains vertical incidence basic calibration coefficient, temperature drift correction coefficient and diffuse light average cosine correction factor;When measuring underwater, multiple-source parameters are synchronously collected, actual incidence zenith angle is calculated according to attitude and position time and the cosine deviation compensation factor of hardware inherent is called;The proportion of direct light component is obtained and is respectively modified using the cosine deviation compensation factor of hardware inherent and the diffuse light average cosine correction factor;Fusion each correction factor is calculated after correction spectral irradiance, and posture angle is dynamically self-adapting updated by sliding window filtering.The method fuses multi-dimensional correction parameters, and realizes the systematic suppression of cosine error under underwater dynamic working condition.

Description

Technical Field

[0001] This invention belongs to the field of marine optical detection technology, specifically relating to a method for global correction of cosine response error of a marine spectral irradiance sensor. Background Technology

[0002] Spectral irradiance is a core measurement parameter in ocean optical and sea color remote sensing, and its measurement accuracy directly affects the authenticity verification of sea color satellite data products and the reliability of marine biogeochemical model inversion. Underwater irradiance measurements typically rely on sensors equipped with PTFE (polytetrafluoroethylene) diffuse cosine correctors, ideally whose output signals satisfy Lambert's cosine law. However, once the sensor is deployed for field ocean measurements, its measurement accuracy is affected by multiple dynamic factors such as ocean current disturbances, attitude tilt, and temperature changes, leading to significant deviations between measured and true values.

[0003] To achieve cosine response correction for irradiance sensors, Chinese invention patent CN104807616B discloses a Spectralon diffuse reflector correction method. This method, implemented in a laboratory environment, involves illuminating a Spectralon diffuse reflector from different zenith angles using a light source, collecting multi-angle reflectance spectral data, and establishing a general cosine function expression for the detector's received zenith angle and correction coefficients based on Lambert's cosine law. The drawback of this method is that the correction process is completed in a static laboratory environment. Once the sensor is deployed underwater for on-site measurements, ocean current disturbances causing sensor tilt will unpredictably alter the actual incident zenith angle, rendering the laboratory-calibrated angle-correction coefficient mapping invalid and unable to compensate for measurement errors caused by dynamic underwater tilting.

[0004] To address the shortcomings of the existing technologies, this application provides a global correction method for the cosine response error of a marine spectral irradiance sensor, aiming to solve the technical problem that existing laboratory static calibration methods cannot adapt to underwater dynamic tilting conditions and are difficult to compensate for the cosine error introduced by attitude changes in real time. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a global correction method for the cosine response error of a marine spectral irradiance sensor. This method integrates laboratory multi-angle, multi-band static calibration with real-time underwater attitude calculation. By coupling the pitch / roll dual-axis tilt angle with the solar zenith angle, the actual incident zenith angle of the light is obtained. The cosine correction function fitted piecewise is dynamically retrieved to achieve real-time compensation for the inherent cosine deviation of the hardware under underwater dynamic tilting conditions. This solves the technical problem that existing laboratory static calibration methods cannot adapt to underwater dynamic attitude.

[0006] This invention provides a method for global correction of cosine response error of a marine spectral irradiance sensor, comprising the following steps: Calibration Modeling: The sensor is illuminated from multiple zenith angles using a standard light source. The entire observation spectrum is traversed, and the original output signals and true values ​​of standard irradiance at each wavelength and angle are collected. The cosine response deviation between the actual response and the ideal response at each wavelength and angle is calculated. A cosine correction function of zenith angle-deviation is constructed segmented according to the spectral interval to obtain the angle fitting coefficient matrix corresponding to different wavelengths, and the basic calibration coefficients of vertical incidence are obtained. Isothermal calibration is performed at different temperatures to obtain the temperature drift correction coefficients. The mean value of the cosine response deviation at each wavelength and angle is calculated as the average cosine correction factor for diffuse light. Data acquisition: During underwater measurement, the raw spectral output signal, sensor pitch angle, sensor roll angle, seawater temperature, measurement depth, observation time, and observation location are acquired simultaneously. Attitude compensation: The equivalent tilt zenith angle of the sensor is calculated based on the pitch and roll angles. The solar zenith angle is calculated based on the latitude and longitude of the observation location and the observation time. The actual incident zenith angle of the light relative to the cosine receiving surface of the sensor is obtained based on the equivalent tilt zenith angle and the solar zenith angle. The actual incident zenith angle is input into the cosine correction function of the corresponding spectral range, and the corresponding wavelength and the hardware inherent cosine deviation compensation factor under the corresponding actual incident zenith angle are output. Light field partitioning and correction: Obtain the proportion of direct light component, and split the downlink light field into direct light component and diffuse light component according to the proportion of direct light component. The direct light component is corrected by calling the hardware inherent cosine deviation compensation factor, and the diffuse light component is compensated by calling the diffuse light average cosine correction factor. Fusion calculation: The basic calibration coefficients, hardware inherent cosine deviation compensation factor, temperature drift correction factor and diffuse light average cosine correction factor are fused to calculate the corrected spectral irradiance; Dynamic filtering: The equivalent tilt zenith angle of multiple consecutive frames is collected using a sliding window. The instantaneous value of the current frame is replaced by the mean value within the window, and the actual incident zenith angle is recalculated based on the updated equivalent tilt zenith angle. The tilt angle fluctuation amplitude within the window is calculated. When the fluctuation amplitude does not exceed the threshold, the historical cache is completely updated with the hardware-inherent cosine deviation compensation factor calculated in the current frame. When the fluctuation amplitude exceeds the threshold, the output is weighted and smoothed with the hardware-inherent cosine deviation compensation factor calculated in the current frame and the historical cache.

[0007] This technical solution integrates laboratory multi-angle, multi-band static calibration with underwater real-time attitude calculation. By coupling the pitch / roll dual-axis tilt angle with the solar zenith angle, the actual incident zenith angle of light is obtained. The cosine correction function of piecewise fitting is dynamically retrieved to achieve real-time compensation for the inherent cosine deviation of the hardware under underwater dynamic tilting conditions. This solves the technical problem that existing laboratory static calibration methods cannot adapt to underwater dynamic attitude.

[0008] In some embodiments, during the calibration modeling step, the cosine response deviation at each wavelength and angle is calculated according to equation (1), which is expressed as: (1); In equation (1), wavelength Zenith angle The cosine response deviation is below. wavelength Zenith angle The original output signal below, For vertically incident basic calibration coefficients, wavelength The true value of standard irradiance; Based on equation (2), construct the cosine correction function for zenith angle-deviation. The expression for equation (2) is: (2); In equation (2), wavelength Zenith angle The cosine response deviation correction function is given below. Indicates wavelength Located in the Within each spectral range, For the first The fitting coefficients corresponding to each spectral interval, then This is the fitting coefficient matrix.

[0009] This technical solution uses Equation (1) to normalize the original output signal of the sensor by the vertical incident basic calibration coefficient and compare it with the ideal cosine response to quantitatively obtain the cosine response deviation at each wavelength and angle; and uses Equation (2) to fit the deviation with the angle using a quadratic polynomial to compress the discrete calibration data into three fitting coefficients, which greatly reduces the local storage overhead of the sensor and ensures that the deviation value at any angle can be continuously calculated.

[0010] In some embodiments, during the attitude compensation step, the actual incident zenith angle is calculated by adding the sensor's equivalent tilt zenith angle to the solar zenith angle.

[0011] This technical solution obtains the actual incident zenith angle by directly superimposing the sensor's equivalent tilted zenith angle with the solar zenith angle. The calculation method is simple and efficient, suitable for real-time operation of embedded MCUs, and does not require iteration or matrix transformation, thus meeting the real-time requirements of underwater high-frequency sampling algorithms.

[0012] In some embodiments, the light field partitioning and correction steps include obtaining the surface direct light ratio, which is the proportion of direct light irradiance in the total surface irradiance below the water surface; obtaining the direct light attenuation coefficient and the diffuse light attenuation coefficient; and calculating the direct light component ratio at the current water depth based on the surface direct light ratio, the direct light attenuation coefficient, the diffuse light attenuation coefficient, and the measured water depth.

[0013] This technical solution obtains the proportion of direct light, the attenuation coefficient of direct light, and the attenuation coefficient of diffuse light on the surface. Combined with the measurement of water depth, it calculates the proportion of direct light components at the current water depth. The underwater light field is divided into direct and diffuse parts according to the physical source, providing a basis for weight allocation for subsequent differentiated correction and avoiding the inherent bias introduced by using a unified correction model.

[0014] In some embodiments, the method for calculating the proportion of direct light component at the current water depth includes: The total downlink optical field reaching the underwater sensor is represented as the sum of the direct light component and the diffuse light component: (3); In equation (3), water depth Total downward spectral irradiance; water depth Direct irradiance component; water depth The diffuse irradiance component at the location; water depth The direct irradiance component is obtained from the direct irradiance of the subsurface layer below the water surface, which decreases exponentially with water depth: (4); In equation (4), The direct irradiance of the subsurface layer of water. The attenuation coefficient for the direct beam; water depth The diffuse irradiance component is obtained from the diffuse irradiance of the subsurface layer below the water surface, which decays exponentially with water depth: (5); In equation (5), water depth The diffuse irradiance component at that location, This refers to the diffuse irradiance of the subsurface layer. The diffuse beam attenuation coefficient; The direct light attenuation coefficient is calculated from the water absorption coefficient, the water scattering coefficient, and the underwater refracted solar zenith angle according to equation (6): (6); In equation (6), The water absorption coefficient; The water scattering coefficient; The zenith angle of the sun is refracted underwater; Based on the proportion of direct light at the surface, the attenuation coefficient of direct light, the attenuation coefficient of diffuse light, and the water depth, the proportion of direct light component at the current water depth is calculated according to equation (7): (7); In equation (7), water depth The proportion of direct light component, The percentage of direct sunlight on the surface.

[0015] This technical solution models direct light and diffuse light according to the exponential decay law through equations (3) to (7). It uses the proportion of direct light at the surface, the absorption coefficient of water, the scattering coefficient and the diffuse decay coefficient to derive the proportion of direct light at any water depth. The physical model is clear, does not rely on additional hardware measurement, and is suitable for profile observation scenarios where the direct / diffuse ratio changes continuously in deep water and weak light environment.

[0016] In some embodiments, the method for obtaining the temperature drift correction coefficient in the calibration modeling step includes: placing the sensor in a constant temperature environment and measuring the reference temperature under the condition that the incident irradiance remains constant. The response signal below and multiple calibration temperatures The response signal below The temperature drift correction factor for each wavelength at each calibration temperature is calculated according to equation (8): (8); In equation (8), wavelength ,temperature Temperature drift correction factor; Reference temperature The response signal below; For temperature The response signal below.

[0017] This technical solution obtains the spectrally resolved temperature drift correction coefficient by measuring the ratio of the response signal at each temperature point to the reference temperature under constant incident irradiance. It can eliminate the dark current drift of the detector and the diffuse transmittance change of the cosine corrector caused by water temperature changes wavelength by wavelength, and avoid the correction mismatch of a single temperature coefficient in the full spectrum range.

[0018] In some embodiments, the corrected spectral irradiance is calculated according to equation (9), which is expressed as: (9); In equation (9), This is the corrected spectral irradiance. This is the actual incident zenith angle. This is the average cosine correction factor for diffuse light.

[0019] This technical solution distributes the original signal to the direct correction channel and the diffuse correction channel according to the proportion of the direct light component. Each channel is multiplied by the corresponding cosine correction factor and temperature correction coefficient, and then the results are added together and divided by the basic calibration coefficient to restore the absolute irradiance. This incorporates the five major error sources—hardware cosine deviation, attitude tilt, spectral difference, temperature drift, and light field distribution difference—into the same solution framework, thereby achieving joint suppression of multiple errors.

[0020] In some embodiments, during the dynamic filtering step, the equivalent tilted zenith angles of multiple consecutive frames are stored in a sliding window, and the mean value within the window is calculated according to equation (10). The mean value within the window is then used to replace the instantaneous value of the current frame in the calculation of the actual incident zenith angle. (10); In equation (10), The equivalent tilt zenith angle after filtering. The length of the sliding window; For the first in the window Frame equivalent tilt zenith angle; The actual incident zenith angle is recalculated based on the updated equivalent tilt zenith angle. The cosine correction function for the corresponding wavelength is then retrieved again using the updated actual incident zenith angle as input to obtain the hardware-inherent cosine deviation compensation factor for the current frame.

[0021] This technical solution smooths the spike noise of the equivalent tilted zenith angle in multiple consecutive frames by using sliding window mean filtering. The mean value within the window is used to replace the instantaneous value in the calculation of the actual incident zenith angle. Then, the cosine correction function is re-called to obtain the compensation factor for the current frame, which effectively suppresses the jump in irradiance calculation caused by the instantaneous impact of ocean currents.

[0022] In some embodiments, during the dynamic filtering step, the hardware-inherent cosine deviation compensation factor is adaptively updated based on the amplitude of the inclination angle fluctuation within the window. The difference between the maximum and minimum values ​​of the equivalent tilt zenith angle within the window is used as the tilt angle fluctuation range within the window. Set attitude fluctuation threshold ; when When the current is determined to be a gentle ocean current condition, the historical cache is completely updated with the compensation factor calculated in the current frame. when When the current is identified as a severe ocean current condition, the compensation factor calculated in the current frame is weighted and smoothed with the historical cache, and the historical cache is updated synchronously. (11); In equation (11), This is the attitude fluctuation threshold. This is the hardware-inherent cosine deviation compensation factor of the filtered output. This is a compensation factor for the inherent cosine bias of the historical hardware. The weighting coefficient is between 0 and 1.

[0023] This technical solution uses the maximum difference in inclination angle within the window as the fluctuation amplitude to determine the sea state. When the current is gentle, the compensation factor is updated directly, and when the current is strong, the transition is weighted and smoothed with the historical cache. This ensures the speed of correction response while preventing the data curve from breaking due to sudden changes in correction value under severe shaking, thus taking into account both dynamic response capability and output stability.

[0024] In some embodiments, the hardware-inherent cosine deviation compensation factor of the adaptive output is used. Substituting into equation (12), we can calculate the corrected spectral irradiance: (12); In equation (12), This is the corrected spectral irradiance; The raw spectral signal acquired; This is the temperature drift correction factor; For vertically incident basic calibration coefficients.

[0025] This technical solution substitutes the compensation factor after adaptive filtering into the fusion calculation formula, so that the smoothing result of the dynamic filtering stage directly affects the final irradiance output, forming a closed loop of "attitude smoothing → angle update → factor update → irradiance calculation", ensuring the temporal consistency of continuous profile observation data.

[0026] Based on the above scheme, the global correction method for the cosine response error of the marine spectral irradiance sensor in this embodiment of the invention has the following beneficial effects: First, a piecewise cosine correction function for the spectrum is established through multi-angle and multi-band static calibration in the laboratory. Combined with underwater real-time pitch / roll attitude calculation, the actual incident zenith angle of the light is obtained, so that the hardware's inherent cosine deviation compensation factor can be updated in real time with dynamic attitude. This solves the problem that the existing laboratory static calibration method cannot adapt to underwater dynamic tilting conditions.

[0027] Second, by using a three-stream radiation transmission model, the underwater optical field is divided into three streams according to the propagation direction: downstream direct, downstream diffuse, and upstream diffuse. This decomposes the total downstream optical field into direct and diffuse components, and statistically analyzes their respective cosine deviation characteristics. The direct light is finely corrected using an angle-dependent piecewise polynomial, while the diffuse light is compensated using a statistical mean constant. This reduces the computational load of the embedded MCU while ensuring correction accuracy. Furthermore, a spectral resolution temperature drift correction coefficient is superimposed, achieving joint suppression of hardware cosine deviation, attitude tilt, optical field distribution differences, and temperature drift.

[0028] Third, by smoothing attitude spike noise through sliding window mean filtering and adaptively switching the compensation factor update strategy according to the inclination fluctuation amplitude within the window, the algorithm effectively suppresses the jump in irradiance calculation caused by ocean current impact. The algorithm is integrated into the sensor MCU, has short computation time, requires no mechanical leveling structure, and is suitable for various ocean observation platforms such as buoys and profilers. It significantly reduces cosine error across the entire incident angle range, meeting the requirements for verifying the authenticity of ocean color satellites and high-precision inversion of inherent ocean optical quantities. Attached Figure Description

[0029] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings: Figure 1 This is a flowchart of the global correction method for the cosine response error of the marine spectral irradiance sensor according to the present invention. Detailed Implementation

[0030] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0031] like Figure 1 As shown, this embodiment provides a method for global correction of cosine response error of a marine spectral irradiance sensor, applicable to underwater spectral irradiance sensors equipped with PTFE cosine correctors. The method comprises two parts: laboratory preprocessing and underwater real-time correction. The entire method includes six steps: calibration modeling, data acquisition, attitude compensation, light field partitioning and correction, fusion calculation, and dynamic filtering.

[0032] I. Calibration and Modeling Steps During the laboratory pretreatment stage, the ocean spectral irradiance sensor to be measured is fixed on a precision angle turntable. A standard integrating sphere or other metrologically calibrated uniform radiation source provides the true value of the known standard spectral irradiance.

[0033] S1-1, perform static cosine calibration with multiple incident angles and multiple wavelengths. Control the angle turntable to change the incident zenith angle, and sequentially set multiple angle points such as 0°, 10°, 20°...80°; at each angle point, traverse all observation spectrum channels of the sensor (350~900nm) and collect the raw output electrical signals at each angle. Simultaneously collect the true value of standard irradiance .

[0034] The angle calibration interval is set at 10°, with 0° to 70° as the core fitting range. Additional calibration points (such as 75°) are added between 70° and 80° to improve the accuracy of correction at large incident angles. The angle calibration interval and the additional calibration points are set in the same way. Calibration points can be added appropriately in the large incident angle range according to the characteristics of different sensors to improve the characterization accuracy of the subsequent fitting model in the large angle range.

[0035] Under vertical incidence conditions ( Based on the sensor's raw output signal Obtain the calibration coefficients of the vertically incident base. .

[0036] S1-2, calculate the original cosine response deviation coefficients for each wavelength and angle. Using the ideal cosine response corresponding to Lambert's cosine law as a benchmark, calculate the cosine response deviations for different wavelengths and incident angles according to equation (1): (1); In equation (1), wavelength Zenith angle The cosine response deviation is below. wavelength Zenith angle The original output signal below, For vertically incident basic calibration coefficients, wavelength The true value of standard irradiance; Equation (1) characterizes the sensor's performance at wavelength λ and zenith angle. The degree of deviation of the actual normalized response from the ideal cosine response. A positive value indicates that the measured response is higher than the theoretical value. A negative value indicates that the measured response is lower than the theoretical value.

[0037] S1-3, construct the cosine correction function for angle-error segmented according to spectral intervals. Divide the observed spectral range into multiple spectral intervals according to ocean color characteristics: ultraviolet 350~400nm, blue light 400~500nm, green light 500~600nm, and red light 600~900nm. Construct the cosine correction function for zenith angle-deviation according to equation (2), the expression of equation (2) is: (2); In equation (2), wavelength Zenith angle The cosine response deviation correction function is given below. Indicates wavelength Located in the Within each spectral range, For the first The fitting coefficients corresponding to each spectral interval, then This is the fitting coefficient matrix. The fitting coefficients corresponding to each spectral interval are written to the sensor's local storage unit, serving as the basic parameters for calling the correction function during subsequent underwater measurements. This segmented strategy allows each band to independently describe the shape of its deviation curve, avoiding the inconsistency problem of long and short band corrections caused by using a single correction coefficient.

[0038] S1-4, Temperature calibration is performed under constant incident irradiance conditions. The sensor is placed in a high-low temperature controlled environment, with multiple temperature measurement points covering the actual operating temperature range of the ocean (-2℃ to 35℃, recommended temperature interval of 5℃). Each temperature point is held at a constant temperature for at least 60 minutes to allow the optical components, detector, and circuit board to reach thermal equilibrium. A standard reference temperature is selected. (Usually 20℃, laboratory calibration reference temperature), acquire the raw response signal across the entire band at the reference temperature. At each of the remaining steady-state temperature points, the raw spectral responses at each wavelength were simultaneously acquired. The spectral resolution temperature drift correction factor is calculated according to equation (8): (8); In equation (8), wavelength ,temperature Temperature drift correction factor; Reference temperature The response signal below; For temperature The response signal below.

[0039] Temperature drift correction factor is used to normalize the measurement signal at any temperature to the reference temperature base, eliminate the gain deviation caused by water temperature changes, and reduce the influence of detector dark current, cosine corrector diffuse transmission characteristics, spectrometer response gain, and signal conditioning circuit temperature drift on the measurement results.

[0040] S1-5, Calculate the mean value of the cosine response deviation across the entire incident angle and the entire wavelength range, and use it as the average cosine correction factor for diffuse light. It is used for the correction of diffuse light components.

[0041] II. Data Acquisition Steps During underwater measurements at the marine site, the sensor simultaneously acquires raw spectral signals at various wavelengths. Real-time pitch angle of the sensor and roll angle Seawater temperature Measuring water depth Meanwhile, the sensor's built-in GPS positioning module and RTC real-time clock provide the observation time (UTC standard time) and observation location (longitude). ,latitude If the GPS signal is lost, the RTC local clock will complete the time parameters.

[0042] In addition, the proportion of surface direct light is obtained for estimating the direct and diffuse light components. Water absorption coefficient Water scattering coefficient and diffuse light attenuation coefficient The aforementioned coefficients can be obtained through inversion of radiance observation data carried on the same platform, historical observation data, or estimation using bio-optical models.

[0043] III. Attitude Compensation Steps S3-1, calculate the equivalent tilt zenith angle of the sensor according to equation (13): (13); In equation (13), The equivalent tilt zenith angle for the sensor. The sensor's pitch angle, This refers to the sensor's roll angle. and It is used to describe the attitude changes of an underwater platform caused by ocean current disturbances, buoy swaying, or profile motion.

[0044] S3-2, Calculate the solar zenith angle: First, calculate the yearly sequence based on the current year, month, and day. (i.e., the first in a year) (366 days in a leap year, 365 days in a common year), calculate the solar declination angle. Calculate the solar declination angle according to formula (14) : (14); Then, based on the observation time, observation longitude, and mean time difference, the local true solar time is calculated according to formula (15). : (15); In equation (15), For hours, For minutes, To observe latitude, Mean time difference. Mean time difference Based on date (corrected for Earth's elliptical orbit, unit: min).

[0045] Then calculate the solar hour angle according to formula (16). : (16); Finally, calculate the solar zenith angle according to formula (17). : (17); S3-3, calculate the actual incident zenith angle according to equation (18): (18); In equation (16), This is the actual incident zenith angle. The solar zenith angle is calculated based on the observation time and latitude / longitude.

[0046] The actual incident zenith angle Substitute the cosine response deviation correction function for the corresponding spectral range According to formula (2), the hardware inherent cosine deviation compensation factor is output.

[0047] The hardware-inherent cosine bias compensation factor characterizes the inherent cosine response bias caused by the sensor's cosine corrector, detector, and optical structure.

[0048] IV. Light Field Partitioning and Correction Steps The total downlink optical field reaching the underwater sensor is formed by the superposition of the direct sunlight component and the diffuse light component from the water. Direct sunlight has a strong directionality, and its cosine response error is strongly correlated with the actual incident zenith angle, requiring fine angle-dependent correction. Diffuse light is formed by multiple scatterings from the water, resulting in a more dispersed angular distribution, which can be simplified and corrected using statistical averaging. This implementation method applies differentiated corrections to the two types of optical fields.

[0049] Total irradiance of the subsurface layer Including direct light irradiance and diffuse irradiance Then we have: Obtain the percentage of direct sunlight on the surface. That is, the total surface irradiance below the water surface Medium direct light irradiance The proportions are as follows: (19); If we express the direct surface light component and the diffuse surface light component separately, then we have: (20). The light field reaching the underwater irradiance sensor consists of the direct sunlight component. With water diffuse light component Superposition composition: (3); In equation (3), water depth Total downward spectral irradiance; water depth Direct irradiance component; water depth The diffuse irradiance component.

[0050] water depth The direct irradiance component is obtained from the direct irradiance of the subsurface layer below the water surface, which decreases exponentially with water depth: (4); In equation (4), The direct irradiance of the subsurface layer of water. This is the attenuation coefficient for the direct beam.

[0051] water depth The diffuse irradiance component is obtained from the diffuse irradiance of the subsurface layer below the water surface, which decays exponentially with water depth: (5); In equation (5), water depth The diffuse irradiance component at that location, This refers to the diffuse irradiance of the subsurface layer. is the attenuation coefficient of the diffuse beam.

[0052] The direct light attenuation coefficient is calculated from the water absorption coefficient, the water scattering coefficient, and the underwater refracted solar zenith angle according to equation (6): (6); In equation (6), The water absorption coefficient; The water scattering coefficient; This represents the zenith angle of the sun refracted underwater. The water absorption coefficient is also included. and scattering coefficient The inherent optical parameters of the water body can be obtained through on-site measurement instruments for inherent optical quantities (such as absorption attenuation meters and backscatterers), or through inversion from historical observation data, regional bio-optical models, and water color satellite remote sensing data; underwater refracted solar zenith angle. It can be determined by the solar zenith angle in the air. It was obtained by conversion using Snell's law of refraction.

[0053] water depth Direct light component and water depth Substituting the diffuse light component into equation (20), we get: (twenty one); In equation (21), dividing the two equations eliminates the total surface irradiance term, yielding the water depth. The ratio of direct light to diffuse light is shown in equation (22): (twenty two); Furthermore, a model for estimating the proportion of direct light components at any water depth is derived, and the proportion of direct light components at the current water depth is calculated according to equation (7). : (7); In equation (7), water depth The proportion of direct light component, The percentage of direct sunlight on the surface.

[0054] This percentage represents the proportion of the direct light component in the total downward optical field, with the remaining portion ( () represents the proportion of diffuse light components.

[0055] After obtaining the proportion of the direct light component, the downlink optical field is split into a direct light component and a diffuse light component based on this proportion. The direct light component utilizes the hardware's inherent cosine deviation compensation factor. Perform angle-related corrections; apply the diffuse light component to the diffuse light mean cosine correction factor. (The statistical mean of the deviations in the cosine response across the entire incident angle and the entire wavelength band) is used for simplified compensation, i.e., the deviation caused by water diffusion is... .

[0056] V. Fusion Solution Steps The calibration coefficient of the vertically incident base Hardware-inherent cosine deviation compensation factor, temperature drift correction coefficient and the average cosine correction factor for diffuse light The corrected spectral irradiance is obtained by fusion and calculation according to equation (9): (9); In equation (9), This is the corrected spectral irradiance. This is the actual incident zenith angle. This is the average cosine correction factor for diffuse light. That is, the hardware-inherent cosine deviation compensation factor obtained based on the corresponding wavelength and the actual incident zenith angle.

[0057] The original signal is processed by equation (9). According to and The proportions are allocated to the direct and diffuse correction channels, each multiplied by its corresponding temperature and cosine correction factors, then summed and divided by the vertical incidence base calibration factor. The result is restored to an absolute irradiance value. This solution eliminates the influence of inherent hardware cosine bias, dynamic attitude tilt, spectral band differences, temperature drift, and differences in the angular distribution of direct and diffuse light on the measurement results.

[0058] VI. Dynamic Filtering Steps During continuous underwater profiling, ocean currents can cause instantaneous jumps in sensor pitch and roll angles, resulting in spike noise in the equivalent tilt zenith angle calculated from a single tilt angle. To avoid this noise causing jumps in irradiance calculations when directly substituted into the cosine model, dynamic adaptive filtering iterative optimization is performed. In this stage, a sliding window mean filter is used to smooth the attitude angle sequence, and adaptive update logic is designed to cope with different sea conditions.

[0059] First, store the equivalent tilt zenith angle of multiple consecutive frames into a file of length [length missing]. The sliding window is initialized with all values ​​set to 0. The mean value within the window is calculated according to equation (10): (10); Average within the window Replace the current frame instantaneous value and recalculate the actual incident zenith angle based on the updated equivalent tilt zenith angle; use the updated actual incident zenith angle as input to re-retriev the cosine correction function for the corresponding wavelength to obtain the hardware-inherent cosine deviation compensation factor for the current frame. .

[0060] Then, the difference between the maximum and minimum values ​​of the equivalent tilt zenith angle within the window is used as the tilt angle fluctuation amplitude within the window. Based on the magnitude of the fluctuation, two operating conditions are handled: when At this point, the condition is determined to be a gentle ocean current. In this case, water disturbance is minimal and attitude is stable. The compensation factor calculated in the current frame is directly used as the compensation factor in the filtered output, and the historical buffer is updated. when At this time, it is determined to be a severe ocean current condition. At this time, the instantaneous swaying noise is large, so the new factor calculated in the current frame is not used directly. Instead, the compensation factor calculated in the current frame is weighted and smoothed with the historical cache and output, and the historical cache is updated synchronously. Calculated according to formula (11): (11); In equation (11), This is the attitude fluctuation threshold. This is the hardware-inherent cosine deviation compensation factor of the filtered output. This is a compensation factor for the inherent cosine bias of the historical hardware. The weighting coefficient is between 0 and 1. In this embodiment, a typical value is... =0.2. Weighted smooth transition allows most of the historical experience to be retained even during violent fluctuations, slow iterative convergence, and prevents the irradiance curve from breaking due to abrupt changes in the correction value.

[0061] The hardware-inherent cosine deviation compensation factor output after adaptive filtering. Substituting into the fusion solution formula (12), the corrected spectral irradiance is calculated: (12); In equation (12), This is the corrected spectral irradiance; The raw spectral signal acquired; This is the temperature drift correction factor; For vertically incident basic calibration coefficients.

[0062] For each new set of observation data, the process of enqueuing, mean calculation, fluctuation discrimination, and adaptive update of correction factors is executed cyclically, dynamically smoothing attitude noise throughout the process, and adapting to continuous observation throughout the entire process of descent and ascent.

[0063] By coordinating the six steps of calibration modeling, data acquisition, attitude compensation, light field partitioning and correction, fusion calculation, and dynamic filtering, this implementation method simultaneously compensates for five major error sources: inherent hardware cosine deviation, underwater dual-axis attitude tilt, spectral band differences, temperature drift, and differences in the angle distribution of direct / diffuse light. Before the sensor leaves the factory, the calibration modeling stage is completed, and the fitting coefficients, basic calibration coefficients, temperature drift correction coefficients, and diffuse light average cosine correction factors are permanently stored in the sensor's local storage unit. During underwater measurements, the five stages of data acquisition, attitude compensation, light field partitioning and correction, fusion calculation, and dynamic filtering are executed sequentially to achieve precise suppression of cosine errors under full incident angle, full spectral band, and dynamic underwater conditions.

[0064] Through the description of several embodiments of the global correction method for the cosine response error of the marine spectral irradiance sensor of the present invention, it can be seen that the embodiments of the global correction method for the cosine response error of the marine spectral irradiance sensor of the present invention have at least one or more of the following advantages: 1. The global correction method for cosine response error of marine spectral irradiance sensor provided by this invention integrates laboratory multi-angle and multi-band static calibration with underwater real-time attitude calculation. By coupling the pitch / roll dual-axis tilt angle with the solar zenith angle, the actual incident zenith angle of light is obtained. The piecewise fitted cosine correction function is dynamically retrieved to achieve real-time compensation for the inherent cosine deviation of the hardware under underwater dynamic tilting conditions. This solves the technical problem that the existing laboratory static calibration method cannot adapt to underwater dynamic attitude.

[0065] 2. The global correction method for cosine response error of marine spectral irradiance sensor provided by the present invention decomposes the downlink optical field into direct light component and diffuse light component through a three-flow radiative transfer model. The direct light component is finely corrected wavelength by wavelength using a piecewise polynomial model, while the diffuse light component is compensated using a statistical mean constant. This method reduces the computational load of the embedded MCU while ensuring the correction accuracy. Furthermore, it superimposes a spectral resolution temperature drift correction coefficient to achieve joint suppression of multiple error sources.

[0066] 3. The global correction method for cosine response error of marine spectral irradiance sensor provided by the present invention smooths attitude spike noise through sliding window mean filtering and adaptively switches the correction factor update strategy according to the dip angle fluctuation amplitude within the window. This effectively suppresses the jump in irradiance calculation caused by ocean current impact. The algorithm is lightweight and integrated into the sensor MCU, requiring no mechanical leveling structure, and is compatible with various underwater observation platforms such as buoys and profilers.

[0067] Finally, it should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0068] The above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them; although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications can still be made to the specific implementation of the present invention or equivalent substitutions can be made to some technical features without departing from the spirit of the technical solutions of the present invention, and all such modifications and substitutions should be covered within the scope of the technical solutions claimed in the present invention.

Claims

1. A method for global correction of cosine response error of a marine spectral irradiance sensor, characterized in that, Includes the following steps: Calibration Modeling: The sensor is illuminated from multiple zenith angles using a standard light source. The entire observation spectrum is traversed, and the original output signals and true values ​​of standard irradiance at each wavelength and angle are collected. The cosine response deviation between the actual response and the ideal response at each wavelength and angle is calculated. A cosine correction function of zenith angle-deviation is constructed segmented according to the spectral interval to obtain the angle fitting coefficient matrix corresponding to different wavelengths, and the basic calibration coefficients of vertical incidence are obtained. Isothermal calibration is performed at different temperatures to obtain the temperature drift correction coefficients. The mean value of the cosine response deviation at each wavelength and angle is used as the average cosine correction factor for diffuse light. Data acquisition: During underwater measurement, the raw spectral output signal, sensor pitch angle, sensor roll angle, seawater temperature, measurement depth, observation time, and observation location are acquired simultaneously. Attitude compensation: The equivalent tilt zenith angle of the sensor is calculated based on the pitch and roll angles. The solar zenith angle is calculated based on the latitude and longitude of the observation location and the observation time. The actual incident zenith angle of the light relative to the cosine receiving surface of the sensor is obtained based on the equivalent tilt zenith angle and the solar zenith angle. The actual incident zenith angle is input into the cosine correction function of the corresponding spectral range, and the corresponding wavelength and the hardware inherent cosine deviation compensation factor under the corresponding actual incident zenith angle are output. Light field partitioning and correction: Obtain the proportion of direct light component, and split the downlink light field into direct light component and diffuse light component according to the proportion of direct light component. The direct light component is corrected by calling the hardware inherent cosine deviation compensation factor, and the diffuse light component is compensated by calling the diffuse light average cosine correction factor. Fusion calculation: The basic calibration coefficients, hardware inherent cosine deviation compensation factor, temperature drift correction factor and diffuse light average cosine correction factor are fused to calculate the corrected spectral irradiance; Dynamic filtering: The equivalent tilt zenith angle of multiple consecutive frames is collected using a sliding window. The instantaneous value of the current frame is replaced by the mean value within the window, and the actual incident zenith angle is recalculated based on the updated equivalent tilt zenith angle. The tilt angle fluctuation amplitude within the window is calculated. When the fluctuation amplitude does not exceed the threshold, the historical cache is completely updated with the hardware-inherent cosine deviation compensation factor calculated in the current frame. When the fluctuation amplitude exceeds the threshold, the output is weighted and smoothed with the hardware-inherent cosine deviation compensation factor calculated in the current frame and the historical cache.

2. The method for global correction of cosine response error of marine spectral irradiance sensor according to claim 1, characterized in that, In the calibration modeling step, the cosine response deviation at each wavelength and angle is calculated according to equation (1). The expression of equation (1) is: （1）； In equation (1), wavelength Zenith angle The cosine response deviation is below. wavelength Zenith angle The original output signal below, For vertically incident basic calibration coefficients, wavelength The true value of standard irradiance; Based on equation (2), construct the cosine correction function for zenith angle-deviation. The expression for equation (2) is: （2）； In equation (2), wavelength Zenith angle The cosine response deviation correction function is given below. Indicates wavelength Located in the Within each spectral range, For the first The fitting coefficients corresponding to each spectral interval, then This is the fitting coefficient matrix.

3. The method for global correction of cosine response error of marine spectral irradiance sensor according to claim 1, characterized in that, In the attitude compensation step, the actual incident zenith angle is calculated by adding the sensor's equivalent tilt zenith angle to the solar zenith angle.

4. The method for global correction of cosine response error of marine spectral irradiance sensor according to claim 1, characterized in that, In the light field partitioning and correction steps, the proportion of direct surface light is obtained, which is the proportion of direct light irradiance in the total irradiance of the surface layer below the water surface; the direct light attenuation coefficient and the diffuse light attenuation coefficient are obtained. The proportion of direct light at the current water depth is calculated based on the proportion of direct light at the surface, the attenuation coefficient of direct light, the attenuation coefficient of diffuse light, and the measured water depth.

5. The method for global correction of cosine response error of marine spectral irradiance sensor according to claim 4, characterized in that, Methods for calculating the proportion of direct light at the current water depth include: The total downlink optical field reaching the underwater sensor is represented as the sum of the direct light component and the diffuse light component: （3）； In equation (3), water depth Total downward spectral irradiance; water depth Direct irradiance component; water depth The diffuse irradiance component at the location; water depth The direct irradiance component is obtained from the direct irradiance of the subsurface layer below the water surface, which decreases exponentially with water depth: （4）； In equation (4), The direct irradiance of the subsurface layer of water. The attenuation coefficient for the direct beam; water depth The diffuse irradiance component is obtained from the diffuse irradiance of the subsurface layer below the water surface, which decays exponentially with water depth: （5）； In equation (5), water depth The diffuse irradiance component at that location, This refers to the diffuse irradiance of the subsurface layer. The diffuse beam attenuation coefficient; The direct light attenuation coefficient is calculated from the water absorption coefficient, the water scattering coefficient, and the underwater refracted solar zenith angle according to equation (6): （6）； In equation (6), The water absorption coefficient; The water scattering coefficient; The zenith angle of the sun is refracted underwater; Based on the proportion of direct light at the surface, the attenuation coefficient of direct light, the attenuation coefficient of diffuse light, and the water depth, the proportion of direct light component at the current water depth is calculated according to equation (7): （7）； In equation (7), water depth The proportion of direct light component, The percentage of direct sunlight on the surface.

6. The method for global correction of cosine response error of marine spectral irradiance sensor according to claim 1, characterized in that, In the calibration modeling step, the method for obtaining the temperature drift correction coefficient includes: placing the sensor in a constant temperature environment and measuring the reference temperature under the condition that the incident irradiance remains constant. The response signal below and multiple calibration temperatures The response signal below The temperature drift correction factor for each wavelength at each calibration temperature is calculated according to equation (8): （8）； In equation (8), wavelength ,temperature Temperature drift correction factor; Reference temperature The response signal below; For temperature The response signal below.

7. The method for global correction of cosine response error of marine spectral irradiance sensor according to claim 1, characterized in that, In the fusion solution step, the corrected spectral irradiance is calculated according to equation (9), which is expressed as follows: （9）； In equation (9), This is the corrected spectral irradiance. This is the actual incident zenith angle. This is the average cosine correction factor for diffuse light.

8. The method for global correction of cosine response error of marine spectral irradiance sensor according to claim 1, characterized in that, In the dynamic filtering step, the equivalent tilt zenith angle of multiple consecutive frames is stored in a sliding window, and the mean value within the window is calculated according to equation (10). The mean value within the window is used to replace the instantaneous value of the current frame in the calculation of the actual incident zenith angle. （10）； In equation (10), The equivalent tilt zenith angle after filtering. The length of the sliding window; For the first in the window Frame equivalent tilt zenith angle; The actual incident zenith angle is recalculated based on the updated equivalent tilt zenith angle. Using the updated actual incident zenith angle as input, the cosine correction function for the corresponding wavelength is re-retrieved to obtain the hardware-inherent cosine deviation compensation factor for the current frame. .

9. The method for global correction of cosine response error of marine spectral irradiance sensor according to claim 8, characterized in that, In the dynamic filtering step, the hardware-inherent cosine deviation compensation factor is adaptively updated based on the fluctuation amplitude of the tilt angle within the window. The difference between the maximum and minimum values ​​of the equivalent tilt zenith angle within the window is used as the tilt angle fluctuation range within the window. Set attitude fluctuation threshold ; when When the current is determined to be a gentle ocean current condition, the historical cache is completely updated with the compensation factor calculated in the current frame. when When the current is determined to be a severe ocean current condition, the compensation factor calculated in the current frame is weighted and smoothed with the historical cache, and the historical cache is updated synchronously. （11）； In equation (11), This is the attitude fluctuation threshold. This is the hardware-inherent cosine deviation compensation factor of the filtered output. This is a compensation factor for the inherent cosine bias of the historical hardware. The weighting coefficient is between 0 and 1.

10. The method for global correction of cosine response error of a marine spectral irradiance sensor according to claim 9, characterized in that, The hardware-inherent cosine deviation compensation factor of the adaptive output. Substituting into equation (12), we can calculate the corrected spectral irradiance: （12）； In equation (12), This is the corrected spectral irradiance; The raw spectral signal acquired; This is the temperature drift correction factor; For vertically incident basic calibration coefficients.

Images