Noise identification and data reconstruction method for raw data of airborne eddy covariance flux observation system

Abstract

The present invention relates to the field of earth observation technology, and provides a method for noise identification and data reconstruction of raw data of an airborne eddy-correlated flux observation system, comprising: obtaining raw observation data of an airborne eddy-correlated flux observation system; performing time-frequency conversion on the raw observation data to obtain an observation power spectrum, comparing the observation power spectrum with a standard power spectrum, and identifying the type of noise in the raw observation data based on the comparison result; and filtering and reconstructing the raw observation data based on the noise type. The method provided by the present invention identifies noise based on analyzing the presentation form of the raw observation data in the frequency domain and possible noise sources. On this basis, the raw observation data is subjected to noise filtering and data reconstruction, providing technical support for obtaining reliable material and energy flux observations at a regional scale. The method can improve the reliability of the airborne eddy-correlated flux observation system in practical applications.

Description

Technical Field

[0001] The present invention relates to the field of earth observation technology, and in particular to a method for raw data noise identification and data reconstruction of an airborne eddy covariance flux observation system. Background Art

[0002] The airborne eddy covariance flux observation system mounted on the aircraft platform actively and rapidly performs high-density, spatially continuous sampling of turbulent three-dimensional wind vectors, air temperature and target gas concentrations (such as water vapor, carbon dioxide and methane) on the flight path, and then realizes spatially continuous observation of energy, moisture and target gas fluxes based on the eddy covariance flux observation theory.

[0003] At present, the existing processing methods for possible errors in airborne eddy covariance observation data mainly include outlier identification and gap interpolation of the original data sequence. This method can only eliminate relatively independent and obvious abnormal points in the observation sequence, but cannot identify other types of noise, thus affecting the accuracy of flux observations. Summary of the invention

[0004] The present invention provides a method for raw data noise identification and data reconstruction of an airborne eddy covariance flux observation system, which is used to solve the defect in the prior art that the noise in the airborne eddy covariance observation data leads to poor flux observation accuracy.

[0005] The present invention provides a method for raw data noise identification and data reconstruction of an airborne eddy covariance flux observation system, comprising:

[0006] Obtain the original observation data of the airborne eddy covariance flux observation system;

[0007] Performing time-frequency conversion on the original observation data to obtain an observation power spectrum, comparing the observation power spectrum with a standard power spectrum, and identifying the noise type in the original observation data based on the comparison result;

[0008] Based on the noise type, noise filtering and data reconstruction are performed on the original observation data.

[0009] According to the method for raw data noise identification and data reconstruction of an airborne eddy covariance flux observation system provided by the present invention, the observation power spectrum is compared with the standard power spectrum, and the noise type in the raw observation data is identified based on the comparison result, including:

[0010] Comparing the observed power spectrum with the standard power spectrum, and determining the frequency band interval where the noise in the original observed data appears based on the comparison result;

[0011] Based on the frequency band interval where the noise appears, the noise type in the original observation data is identified.

[0012] According to the method for raw data noise identification and data reconstruction of an airborne eddy covariance flux observation system provided by the present invention, the noise type in the raw observation data is identified based on the frequency band interval where the noise occurs, including:

[0013] If the frequency band in which the noise occurs is at the end of the high frequency, the noise type is white noise; otherwise, the noise type is colored noise.

[0014] According to the method for noise identification and data reconstruction of raw data of an airborne eddy covariance flux observation system provided by the present invention, based on the noise type, the raw observation data is subjected to noise filtering and data reconstruction, comprising:

[0015] If the noise type is white noise, the original observation data is filtered out and the data is reconstructed by using a zero-phase shift mean filter;

[0016] If the noise type is colored noise, the noise effect generated by the colored noise is quantified in the frequency domain for the frequency band where the colored noise appears, and the noise effect is filtered out during data reconstruction to obtain reconstructed observation data.

[0017] According to the method for raw data noise identification and data reconstruction of an airborne eddy covariance flux observation system provided by the present invention, the noise influence generated by the colored noise is quantified in the frequency domain, and the noise influence is filtered out during data reconstruction to obtain reconstructed observation data, including:

[0018] For the frequency band interval where the colored noise appears, calculating the ratio between the observed power spectrum mean and the standard power spectrum mean;

[0019] Based on the ratio, the reconstructed observation data is obtained by the following formula:

[0020]

[0021] In the formula, To reconstruct the data sequence of the observed data, is the discrete Fourier transform coefficient of the original observation data, is the data sequence length of the original observation data, [b1, b2] is the frequency band interval where the colored noise appears, is the ratio between the mean of the observed power spectrum and the mean of the standard power spectrum, is the lowest frequency of discrete Fourier transform, is the Nyquist frequency.

[0022] According to the method for raw data noise identification and data reconstruction of an airborne eddy covariance flux observation system provided by the present invention, the raw observation data is subjected to time-frequency conversion to obtain an observation power spectrum, including:

[0023] Performing discrete Fourier transform on the original observation data to obtain an original power spectrum;

[0024] Performing interval averaging processing on the original power spectrum to obtain a stable power spectrum;

[0025] The stable power spectrum is normalized to obtain a normalized power spectrum, and the normalized power spectrum is used as the observed power spectrum.

[0026] According to the original data noise identification and data reconstruction method of the airborne eddy covariance flux observation system provided by the present invention, the original observation data includes at least one of three-dimensional wind speed, air temperature and target gas concentration.

[0027] The present invention also provides a device for identifying noise in raw data and reconstructing data of an airborne eddy covariance flux observation system, comprising:

[0028] A data acquisition unit, used to acquire raw observation data of an airborne eddy covariance flux observation system;

[0029] A noise identification unit, configured to perform time-frequency conversion on the original observation data to obtain an observation power spectrum, compare the observation power spectrum with a standard power spectrum, and identify the noise type in the original observation data based on the comparison result;

[0030] A data reconstruction unit is used to perform noise filtering and data reconstruction on the original observation data based on the noise type.

[0031] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, it implements the raw data noise identification and data reconstruction method of the airborne eddy correlation flux observation system as described in any one of the above-mentioned methods.

[0032] The present invention also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for raw data noise identification and data reconstruction of an airborne eddy covariance flux observation system as described in any one of the above is implemented.

[0033] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method for raw data noise identification and data reconstruction of an airborne eddy covariance flux observation system as described in any one of the above is implemented.

[0034] The method for identifying noise in raw data and reconstructing data of an airborne eddy covariance flux observation system provided by the present invention identifies noise in the frequency domain by analyzing the presentation form of raw observation data in the frequency domain and possible noise sources. Based on the noise type identified, the raw observation data is subjected to noise filtering and data reconstruction to filter out the noise in the raw observation data and reconstruct a high-quality data sequence, thereby providing technical support for obtaining reliable material and energy flux observations at a regional scale. The method can improve the reliability of the airborne eddy covariance flux observation system in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0036] Figure 1 It is one of the flow charts of the method for raw data noise identification and data reconstruction of the airborne eddy covariance flux observation system provided by the present invention.

[0037] Figure 2 This is a schematic diagram of white noise recognition provided by the present invention.

[0038] Figure 3 It is a schematic diagram of colored noise recognition provided by the present invention.

[0039] Figure 4 This is the second flow chart of the method for raw data noise identification and data reconstruction of the airborne eddy covariance flux observation system provided by the present invention.

[0040] Figure 5 It is a schematic diagram of original observation data and reconstructed observation data provided by the present invention.

[0041] Figure 6 It is a structural schematic diagram of the raw data noise identification and data reconstruction device of the airborne eddy covariance flux observation system provided by the present invention.

[0042] Figure 7 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] The airborne eddy covariance flux observation system can be used to study the exchange of matter and energy between the surface and the atmosphere. It has been used in many large-scale field observation experiments and is currently the only means to directly obtain regional-scale flux observations.

[0045] In the eddy covariance flux observation system, three-dimensional wind speed, air temperature and target gas concentration are the three most important parameters for obtaining material and energy flux. However, due to the high integration of the airborne eddy covariance flux observation system, the high-frequency, high-quality three-dimensional wind speed, air temperature and target gas concentration observations that meet the flux calculation requirements rely on the collaborative work of multiple sensors. On the other hand, affected by the dynamic characteristics of the fuselage, the natural frequency of the fuselage and its components, abnormal jitter and the interaction between the fuselage and turbulence will bring noise to the observation of three-dimensional wind speed, air temperature and target gas concentration. For example, the rotation of the aircraft propeller affects the flow field around the fuselage, interferes with the airborne observation signal, and affects its further application. At present, the observation of high-frequency, high-quality three-dimensional wind speed, air temperature and target gas concentration still faces many challenges, and observation noise is ubiquitous. Error identification and data reconstruction of the high-frequency three-dimensional wind speed, air temperature and target gas concentration data obtained by the airborne eddy covariance flux observation system are crucial for subsequent flux calculations.

[0046] At present, the noise processing methods for high-frequency three-dimensional wind speed, air temperature and target gas concentration of airborne eddy covariance flux observation systems are mainly limited to the identification of wild points and simple interpolation of vacant positions. In terms of wild point identification, some simple rules are usually specified. For example: points beyond the physical range are defined as wild points; when the absolute value of the mean of the current data and a certain segment of data is greater than N times (4-6 times for flux) of the standard deviation of the segment of data, the current point is considered a wild point, and this rule is applied to the sliding window to identify wild points in the entire segment of data. For the identified wild points, linear interpolation of adjacent points is usually used.

[0047] In view of the possible error problems in airborne eddy covariance observation data, the existing processing methods mainly include outlier identification and gap interpolation of the original data sequence, which can only remove relatively independent and obvious abnormal points in the observation sequence, and cannot process the white noise and colored noise that persist in the signal. However, white noise and colored noise related to factors such as instrument undercoordination and fuselage dynamic characteristics are common and critical errors in airborne eddy covariance flux observation systems.

[0048] In view of the above problems, an embodiment of the present invention proposes a method for noise identification and data reconstruction of raw data of an airborne eddy-correlated flux observation system. In the method, the raw observation data is subjected to time-frequency conversion to obtain an observation power spectrum, and the observation power spectrum is compared with the standard power spectrum. Based on the comparison result, the noise type in the raw observation data is identified; based on the noise type, the raw observation data is reconstructed. The method identifies the noise based on analyzing the presentation form of the raw observation data in the frequency domain and the possible sources of noise. Based on the identification of the noise type, the raw observation data is subjected to noise filtering and data reconstruction to filter the noise in the raw observation data and reconstruct a high-quality data sequence, thereby providing technical support for obtaining reliable material and energy flux observations at a regional scale. The method can improve the reliability of the airborne eddy-correlated flux observation system in practical applications.

[0049] The embodiment of the present invention can be applied to the scenario where the original observation data needs to be noise identified and data reconstructed in the airborne eddy covariance flux observation system. The execution subject of the method can be an electronic device such as a terminal device, a computer, a server, a server cluster, or a specially designed noise identification and data reconstruction device, or a noise identification and data reconstruction device set in the electronic device, which can be implemented by software, hardware, or a combination of the two.

[0050] Figure 1 FIG. 1 is one of the flow charts of the method for raw data noise identification and data reconstruction of the airborne eddy covariance flux observation system provided by the present invention, such as Figure 1 As shown, the method comprises the following steps:

[0051] Step 110, obtaining the original observation data of the airborne eddy covariance flux observation system.

[0052] Specifically, the airborne eddy covariance flux observation system can be carried on an aircraft platform. The airborne eddy covariance flux observation system can actively and quickly perform high-density, spatially continuous sampling of turbulent three-dimensional wind vectors, air temperature and target gas concentrations (such as water vapor, carbon dioxide and methane) on the flight path through high-precision sensors such as three-dimensional ultrasonic anemometers and CO2 / H2O analyzers to obtain original observation data.

[0053] Here, the original observation data may include at least one of three-dimensional wind speed, air temperature, and target gas concentration.

[0054] Step 120, performing time-frequency conversion on the original observation data to obtain an observation power spectrum, comparing the observation power spectrum with the standard power spectrum, and identifying the noise type in the original observation data based on the comparison result;

[0055] Step 130: based on the noise type, perform noise filtering and data reconstruction on the original observation data.

[0056] Specifically, considering the possible error problems in the original observation data, such as instrument under-coordination and white noise and colored noise related to factors such as fuselage dynamic characteristics, they are common and critical errors in the airborne eddy covariance flux observation system.

[0057] In the related technology, by identifying outliers and filling gaps in the original observation data sequence, only relatively independent and obvious abnormal points in the observation sequence can be eliminated, but the white noise and colored noise that persist in the signal cannot be processed, thus affecting the accuracy of flux observation.

[0058] In this step, the original observation data is first converted to a time-frequency format to obtain the observed power spectrum. Time-frequency conversion refers to the conversion between time domain signals and frequency domain signals. The discrete Fourier transform (DFT) can be used to transform the sampling of the time domain signal into the sampling in the discrete time Fourier transform (DTFT) frequency domain. In practical applications, the fast Fourier transform (FFT) can be used to improve computational efficiency.

[0059] Discrete Fourier transform and its inverse transform provide a method to convert observation sequences between the spatial domain and the frequency domain, which can be used to separate signal components and synthesize signals. For spatial domain observation data, after the original signal is subjected to discrete Fourier transform, the white noise and colored noise it contains, as a "signal" component, will also be presented in the frequency domain. Based on Kolmogorov's locally uniform isotropic turbulence theory, at large Reynolds values, it can be considered that the random characteristics of turbulent motion are isotropic, and there is an inertial sub-region scale space. Within this range, turbulent motion is determined by inertial forces, and the distribution of energy between turbulent eddies of different scales follows the k^(-5 / 3) power law rule, where k is the wave number or time frequency. The power spectrum of boundary layer meteorological variables usually satisfies this power law rule, and this feature can be used as a reference for evaluating the quality of observations. The standard power spectrum here can be the power spectrum of boundary layer meteorological variables.

[0060] The embodiment of the present invention uses time-frequency conversion to classify and identify noise in the frequency domain, compares the observed power spectrum with the standard power spectrum, and identifies the noise type in the original observed data based on the comparison result. Based on the noise type, the corresponding filter is designed to reduce the noise and reconstruct the data sequence without the influence of the noise.

[0061] The method provided in the embodiment of the present invention identifies noise in the frequency domain based on analyzing the presentation form of the original observation data in the frequency domain and possible noise sources. Based on the identified noise type, the original observation data is subjected to noise filtering and data reconstruction to filter out the noise in the original observation data and reconstruct a high-quality data sequence, thereby providing technical support for obtaining reliable material and energy flux observations at a regional scale. This method can improve the reliability of the airborne eddy covariance flux observation system in practical applications.

[0062] Based on the above embodiment, in step 120, the original observation data is subjected to time-frequency conversion to obtain an observation power spectrum, which specifically includes:

[0063] Step 121, performing discrete Fourier transform on the original observation data to obtain the original power spectrum;

[0064] Step 122, performing interval averaging processing on the original power spectrum to obtain a stable power spectrum;

[0065] Step 123, normalize the stable power spectrum to obtain a normalized power spectrum, and use the normalized power spectrum as the observed power spectrum.

[0066] Specifically, the original observation data can be a discrete time series signal. The sampling frequency (Fs) and the number of sampling points (N) of the original observation data can be determined first. The original observation data is transformed using the DFT algorithm. DFT is a discrete form of Fourier transform in both the time domain and the frequency domain, which transforms the sampling of the time domain signal into sampling in the discrete time Fourier transform (DTFT) frequency domain. The result of DFT is an array of complex numbers, which contains the amplitude and phase information of the signal at each frequency component. The power spectrum is usually defined as the energy distribution of the signal at each frequency. In the result of DFT, the power at the frequency can be obtained by calculating the square of the modulus of the complex number of each frequency component, that is, the original power spectrum is obtained.

[0067] In step 122, based on the analysis requirements and data characteristics, the intervals on the power spectrum that need to be averaged are determined, and the intervals can be divided based on the frequency range or specific power spectrum characteristics. In each determined interval, the average value of the power of all frequency points is calculated. This can be achieved by adding the power of all frequency points in the interval and then dividing it by the number of frequency points in the interval, thereby obtaining a stable power spectrum that can be used to analyze the frequency characteristics of the data sequence.

[0068] Then, add up the power values ​​of all bins in the stable power spectrum to get the total power in the entire frequency range. Divide the power value of each bin in the stable power spectrum by the total power to get the normalized power value. In this way, the sum of the normalized powers of all bins will be equal to 1.

[0069] The normalized power spectrum reflects the relative distribution of signal energy at different frequencies, thereby obtaining the power distribution corresponding to turbulence of different scales for subsequent analysis and comparison. At the same time, the normalized power spectrum is used as the observed power spectrum.

[0070] Based on any of the above embodiments, in step 120, the observed power spectrum is compared with the standard power spectrum, and the noise type in the original observed data is identified based on the comparison result, including:

[0071] Step 124, comparing the observed power spectrum with the standard power spectrum, and determining the frequency band interval where the noise appears in the original observed data based on the comparison result;

[0072] Step 125: Based on the frequency band where the noise occurs, identify the type of noise in the original observation data.

[0073] Specifically, the standard power spectrum can be used as a reference for noise identification. This standard power spectrum can be a power spectrum of a noise-free signal based on a theoretical model, experimental data, or previous observations, for example, a Kolmogorov theory standard spectrum, that is, a power spectrum line that satisfies the k^(-5 / 3) power law rule.

[0074] At the same time, ensure that the standard power spectrum and the observed power spectrum are consistent in parameters such as frequency resolution and sampling frequency to facilitate accurate comparison.

[0075] The observed power spectrum can be compared with the standard power spectrum one by one in the same frequency range. This is achieved by plotting the power spectrum curves of the two in logarithmic space and displaying them in the same coordinate system. Observe the differences between the two, especially whether there is a significant increase or decrease in energy in certain frequency bands. These differences may indicate the presence of noise. Based on the comparison results, identify the frequency bands in the observed power spectrum where the energy is significantly increased relative to the standard power spectrum. These intervals are likely to be the frequency bands where noise occurs.

[0076] Carefully observe the frequency range where the noise appears, and analyze the power spectrum characteristics of the noise in this range, such as spectrum width, spectrum shape, energy distribution, etc. Try to match the noise characteristics with known noise types, such as white noise, pink noise, periodic noise, etc.

[0077] In some embodiments, based on the frequency band interval where the noise occurs, identifying the noise type in the original observation data includes:

[0078] If the frequency band where the noise occurs is at the high-frequency end, the noise type is white noise; otherwise, the noise type is colored noise.

[0079] Specifically, the noise types in this embodiment mainly include white noise and colored noise. Since white noise mainly manifests as high-frequency oscillations uniformly distributed in the observed data, the situation where the normalized power of the observed data at the high-frequency terminal is greater than the standard power can be used as a basis for determining white noise. Figure 2 : is a schematic diagram of white noise recognition provided by the present invention, such as Figure 2 As shown, if the frequency band where the noise occurs is at the high-frequency end, the noise type is white noise.

[0080] Since colored noise mainly manifests itself as additional oscillations in a certain mid-frequency range superimposed on the data signal, the scenario in which the normalized power of the observed data in a certain mid-frequency range (not the highest frequency range) is greater than the standardized power can be used as a basis for determining colored noise. Figure 3 is a schematic diagram of colored noise recognition provided by the present invention, such as Figure 3 As shown, if the frequency band where the noise occurs is not at the end of the high frequency, such as a mid-frequency band, the noise type is colored noise.

[0081] Based on any of the above embodiments, step 130 performs noise filtering and data reconstruction on the original observation data based on the noise type, specifically including:

[0082] Step 131, if the noise type is white noise, the original observation data is filtered out and the data is reconstructed by using a zero-phase shift mean filter;

[0083] Step 132: If the noise type is colored noise, the noise effect generated by the colored noise is quantified in the frequency domain for the frequency band where the colored noise appears, and the noise effect is filtered out during data reconstruction to obtain reconstructed observation data.

[0084] Specifically, for white noise, since it is a high-frequency error, the original observation data sequence is processed by zero-phase shift mean filtering to reduce the high-frequency error. Zero-phase shift mean filtering is a low-pass filter. Compared with general low-pass filters, its characteristic is that after filtering the data, the phase characteristics of the data signal can be kept unchanged. This feature is very important for subsequent turbulent flux calculations. For airborne eddy-correlated flux observation systems, white noise is common in high-frequency air temperature observations, mainly caused by insufficient bandwidth of the instrument.

[0085] For the original observation data sequence with colored noise, first determine the frequency band where the colored noise appears, which will be the focus of subsequent processing. In the colored noise frequency band, calculate its power spectral density (PSD) in detail. The calculation of PSD can help understand the energy distribution of noise at different frequencies, so as to quantify its impact more accurately.

[0086] According to the PSD of colored noise, the degree of its influence on the original observed signal can be evaluated. This can be achieved by comparing the relative size of noise power and signal power. If the noise power is significantly higher than the signal power in a certain frequency band, the signal in this frequency band may be seriously interfered.

[0087] The designed filter is then applied to the original observation data to remove the effects of colored noise in the frequency domain. This can be achieved by applying the filter in the frequency domain and finally converting the signal back to the time domain through IFFT (Inverse Fast Fourier Transform). The noise effects of the reconstructed observation data in the colored noise frequency band have been significantly reduced or eliminated.

[0088] Based on any of the above embodiments, the noise effect generated by the colored noise is quantified in the frequency domain, and the noise effect is filtered out during data reconstruction to obtain reconstructed observation data, including:

[0089] For the frequency band where colored noise appears, calculate the ratio between the mean of the observed power spectrum and the mean of the standard power spectrum;

[0090] Based on the ratio, the reconstructed observation data is obtained by the following formula:

[0091]

[0092] In the formula, To reconstruct the data sequence of the observed data, is the discrete Fourier transform coefficient of the original observation data, is the data sequence length of the original observation data, [b1, b2] is the frequency band interval where the colored noise appears, is the ratio between the mean of the observed power spectrum and the mean of the standard power spectrum, is the lowest frequency of discrete Fourier transform, is the Nyquist frequency.

[0093] Specifically, for the airborne eddy covariance flux observation system, colored noise is common in high-frequency three-dimensional wind speed observations and target gas concentration observations. Its main influencing factors include: when measuring wind speed, the time consistency between the inertial navigation system and the sensors of the relative wind speed measurement system is poor; the natural frequency of the fuselage interferes with the measured flow field.

[0094] For a data sequence with colored noise, the correction method includes the following steps:

[0095] 1) Compare the observed power spectrum (normalized power spectrum) and the standardized power spectrum of the original observed data, and find the spectral interval where the power of the original observed data is greater than the standardized power, recorded as [b1, b2].

[0096] 2) For this spectrum segment, calculate the ratio of the observed power spectrum mean of the original observation data to the standardized power mean based on formula (1): .

[0097] (1)

[0098] in, For frequency The value of the observed power spectrum is For frequency The value of the standard power spectrum at .

[0099] 3) Correct the discrete Fourier transform components corresponding to the [b1, b2] frequency band, and perform inverse discrete Fourier transform based on formula (2) to reconstruct the data sequence after filtering out the influence of colored noise.

[0100] (2)

[0101] in, To reconstruct the data sequence of the observed data, is the discrete Fourier transform coefficient of the original observation data, is the data sequence length of the original observation data, [b1, b2] is the frequency band interval where the colored noise appears, is the ratio between the mean of the observed power spectrum and the mean of the standard power spectrum, is the lowest frequency of discrete Fourier transform, is the Nyquist frequency.

[0102] Based on any of the above embodiments, Figure 4 FIG. 2 is a flow chart of the method for raw data noise identification and data reconstruction of the airborne eddy covariance flux observation system provided by the present invention. Figure 4 As shown, the method includes:

[0103] S1, obtain the original observation data of the airborne eddy covariance flux observation system.

[0104] S2, performing time-frequency conversion on the original observation data to obtain the observation power spectrum, which specifically includes: performing discrete Fourier transform on the original observation data to obtain the original power spectrum; performing interval averaging processing on the original power spectrum to obtain a stable power spectrum; performing normalization processing on the stable power spectrum to obtain a normalized power spectrum, and using the normalized power spectrum as the observation power spectrum.

[0105] S3, comparing the observed power spectrum with the standard power spectrum, and identifying the noise type in the original observed data based on the comparison result. Specifically including:

[0106] The observed power spectrum is compared with the standard power spectrum, and the frequency band interval where the noise appears in the original observed data is determined based on the comparison result; if the frequency band interval where the noise appears is at the high-frequency end, the noise type is white noise; otherwise, the noise type is colored noise.

[0107] S4, based on the noise type, performs noise filtering and data reconstruction on the original observation data. Specifically includes:

[0108] If the noise type is white noise, the original observation data is filtered out and reconstructed by using zero-phase shift mean filtering;

[0109] If the noise type is colored noise, the noise impact caused by the colored noise is quantified in the frequency domain for the frequency band where the colored noise appears, and the noise impact is filtered out during data reconstruction to obtain reconstructed observation data.

[0110] Turbulent motion in a flow field can usually be regarded as the superposition of turbulent motion of different scales. Turbulent motion of different scales corresponds to different frequency characteristics. Therefore, high-frequency sampling in the spatial domain of a turbulent field can be interpreted as the superposition of signals of different periods. The use of discrete Fourier transform can effectively separate the turbulent motion of different scales contained in long time series or wide spatial scale turbulence signals, and simplify the data expression with a finite frequency sequence. On the other hand, the inverse discrete Fourier transform can reconstruct the frequency domain components into time domain or space domain signals, and there is no information loss in the conversion process. Therefore, the present invention uses discrete Fourier transform to convert the original observation data into the frequency domain, identify the frequency of noise, correct the original power spectrum based on the prior knowledge of the power spectrum of boundary layer meteorological variables, and then use the inverse Fourier transform to reconstruct the time domain or space domain data sequence, which can effectively filter out the influence of noise.

[0111] The present invention targets the data noise commonly existing in high-frequency three-dimensional wind speed, air temperature and target gas concentration, utilizes frequency domain and space domain data analysis methods, and combines boundary layer meteorological prior knowledge to perform noise identification and data reconstruction, thereby improving the reliability of airborne eddy covariance flux observation technology in practical applications.

[0112] For different types of noise, different filtering methods are set to remove noise. For white noise, the zero-phase shift mean filtering method is used to remove it, which has the characteristics of good effect and ease of use. For colored noise, the noise frequency band is identified in the frequency domain, and the noise impact is quantified using prior knowledge, and removed when reconstructing the data sequence.

[0113] Figure 5 is a schematic diagram of the original observation data and the reconstructed observation data provided by the present invention, such as Figure 5 As shown, Figure 5(a) is a schematic diagram of the normalized power spectrum of the original air temperature observation data and the normalized power spectrum of the reconstructed observation data. The horizontal axis is the frequency (Hz) and the vertical axis is the normalized power value. Figure 5 (b) is a schematic diagram of the original observation data of air temperature, with the horizontal axis being the observation sequence number and the vertical axis being the temperature value (K). Figure 5 (C) is a schematic diagram of air temperature reconstruction observation data, with the horizontal axis being the observation sequence number and the vertical axis being the temperature value (K). Compared with the original observation data schematic diagram (b), the reconstructed observation data schematic diagram (C) has a narrower data variation range and less white noise.

[0114] Figure 5 Middle (d) is a schematic diagram of the normalized power spectrum of the original wind speed observation data and the normalized power spectrum of the reconstructed observation data. The horizontal axis is the frequency (Hz) and the vertical axis is the normalized power value. Figure 5 (e) is a schematic diagram of the original wind speed observation data, with the horizontal axis being the observation sequence number and the vertical axis being the wind speed value (m / s). Figure 5 (f) is a schematic diagram of wind speed reconstruction observation data, with the horizontal axis being the observation sequence number and the vertical axis being the wind speed value (m / s).

[0115] Based on any of the above embodiments, Figure 6 : is a schematic diagram of the structure of the original data noise identification and data reconstruction device of the airborne eddy covariance flux observation system provided by the present invention, such as Figure 6 As shown, the device comprises:

[0116] A data acquisition unit 610 is used to acquire raw observation data of an airborne eddy covariance flux observation system;

[0117] A noise identification unit 620 is used to perform time-frequency conversion on the original observation data to obtain an observation power spectrum, compare the observation power spectrum with a standard power spectrum, and identify the noise type in the original observation data based on the comparison result;

[0118] The data reconstruction unit 630 is used to perform noise filtering and data reconstruction on the original observation data based on the noise type.

[0119] The device provided by the embodiment of the present invention identifies noise in the frequency domain based on analyzing the presentation form of the original observation data in the frequency domain and possible noise sources. Based on the identified noise type, the original observation data is subjected to noise filtering and data reconstruction to filter out the noise in the original observation data and reconstruct a high-quality data sequence, thereby providing technical support for obtaining reliable material and energy flux observations at a regional scale. This method can improve the reliability of the airborne eddy covariance flux observation system in practical applications.

[0120] Based on any of the above embodiments, the noise recognition unit is specifically used for:

[0121] Comparing the observed power spectrum with the standard power spectrum, and determining the frequency band interval where the noise in the original observed data appears based on the comparison result;

[0122] Based on the frequency band interval where the noise appears, the noise type in the original observation data is identified.

[0123] Based on any of the above embodiments, the noise recognition unit is specifically used for:

[0124] If the frequency band in which the noise occurs is at the end of the high frequency, the noise type is white noise; otherwise, the noise type is colored noise.

[0125] Based on any of the above embodiments, the data reconstruction unit is specifically used for:

[0126] If the noise type is white noise, the original observation data is filtered out and the data is reconstructed by using a zero-phase shift mean filter;

[0127] If the noise type is colored noise, the noise effect generated by the colored noise is quantified in the frequency domain for the frequency band where the colored noise appears, and the noise effect is filtered out during data reconstruction to obtain reconstructed observation data.

[0128] Based on any of the above embodiments, the data reconstruction unit is specifically used for:

[0129] For the frequency band interval where the colored noise appears, calculating the ratio between the observed power spectrum mean and the standard power spectrum mean;

[0130] Based on the ratio, the reconstructed observation data is obtained by the following formula:

[0131]

[0132] In the formula, To reconstruct the data sequence of the observed data, is the discrete Fourier transform coefficient of the original observation data, is the data sequence length of the original observation data, [b1, b2] is the frequency band interval where the colored noise appears, is the ratio between the mean of the observed power spectrum and the mean of the standard power spectrum, is the lowest frequency of discrete Fourier transform, is the Nyquist frequency.

[0133] Based on any of the above embodiments, the noise recognition unit is specifically used for:

[0134] Performing discrete Fourier transform on the original observation data to obtain an original power spectrum;

[0135] Performing interval averaging processing on the original power spectrum to obtain a stable power spectrum;

[0136] The stable power spectrum is normalized to obtain a normalized power spectrum, and the normalized power spectrum is used as the observed power spectrum.

[0137] Based on any of the above embodiments, the original observation data includes at least one of three-dimensional wind speed, air temperature and target gas concentration.

[0138] Figure 7 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 7 As shown, the electronic device may include: a processor 710, a communication interface 720, a memory 730 and a communication bus 740, wherein the processor 710, the communication interface 720 and the memory 730 communicate with each other through the communication bus 740. The processor 710 may call the logic instructions in the memory 730 to execute the original data noise identification and data reconstruction method of the airborne eddy correlation flux observation system, the method comprising: obtaining the original observation data of the airborne eddy correlation flux observation system; performing time-frequency conversion on the original observation data to obtain an observation power spectrum, comparing the observation power spectrum with the standard power spectrum, and identifying the noise type in the original observation data based on the comparison result; based on the noise type, performing noise filtering and data reconstruction on the original observation data.

[0139] In addition, the logic instructions in the above-mentioned memory 730 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0140] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the original data noise identification and data reconstruction method of the airborne eddy-correlated flux observation system provided by the above-mentioned methods. The method includes: acquiring the original observation data of the airborne eddy-correlated flux observation system; performing time-frequency conversion on the original observation data to obtain an observation power spectrum, comparing the observation power spectrum with a standard power spectrum, and identifying the noise type in the original observation data based on the comparison result; based on the noise type, performing noise filtering and data reconstruction on the original observation data.

[0141] On the other hand, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the method for noise identification and data reconstruction of raw data of an airborne eddy-correlated flux observation system provided by the above-mentioned methods, the method comprising: acquiring raw observation data of an airborne eddy-correlated flux observation system; performing time-frequency conversion on the raw observation data to obtain an observation power spectrum, comparing the observation power spectrum with a standard power spectrum, and identifying the type of noise in the raw observation data based on the comparison result; and performing noise filtering and data reconstruction on the raw observation data based on the noise type.

[0142] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0143] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0144] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for raw data noise identification and data reconstruction of an airborne eddy covariance flux observation system, characterized in that: include: Obtain the original observation data of the airborne eddy covariance flux observation system; Performing time-frequency conversion on the original observation data to obtain an observation power spectrum, comparing the observation power spectrum with a standard power spectrum, and identifying the noise type in the original observation data based on the comparison result; Based on the noise type, performing noise filtering and data reconstruction on the original observation data; The performing noise filtering and data reconstruction on the original observation data based on the noise type includes: If the noise type is colored noise, then for the frequency band interval where the colored noise appears, the noise effect generated by the colored noise is quantified in the frequency domain, and the noise effect is filtered out during data reconstruction to obtain reconstructed observation data; The noise influence generated by the colored noise is quantified in the frequency domain, and the noise influence is filtered out during data reconstruction to obtain reconstructed observation data, including: For the frequency band interval where the colored noise appears, calculating the ratio between the observed power spectrum mean and the standard power spectrum mean; Based on the ratio, the reconstructed observation data is obtained by the following formula: In the formula, S new (n) is the data sequence of reconstructed observation data, F(f) is the discrete Fourier transform coefficient of the original observation data, N is the data sequence length of the original observation data, [b1, b2] is the frequency band where colored noise appears, a is the ratio between the mean of the observed power spectrum and the mean of the standard power spectrum, fl is the lowest frequency of discrete Fourier transform, and f0 is the Nyquist frequency.

2. The method for raw data noise identification and data reconstruction of an airborne eddy covariance flux observation system according to claim 1, characterized in that: The step of comparing the observed power spectrum with the standard power spectrum and identifying the noise type in the original observed data based on the comparison result includes: Comparing the observed power spectrum with the standard power spectrum, and determining the frequency band interval where the noise in the original observed data appears based on the comparison result; Based on the frequency band interval where the noise appears, the noise type in the original observation data is identified.

3. The method for raw data noise identification and data reconstruction of an airborne eddy covariance flux observation system according to claim 2 is characterized in that: The identifying the noise type in the original observation data based on the frequency band interval where the noise occurs includes: If the frequency band in which the noise occurs is at the end of the high frequency, the noise type is white noise; otherwise, the noise type is colored noise.

4. The method for raw data noise identification and data reconstruction of an airborne eddy covariance flux observation system according to claim 1, characterized in that: The performing noise filtering and data reconstruction on the original observation data based on the noise type includes: If the noise type is white noise, the original observation data is filtered out and the data is reconstructed by using a zero-phase shift mean filter.

5. The method for raw data noise identification and data reconstruction of an airborne eddy covariance flux observation system according to any one of claims 1 to 4, characterized in that: The performing time-frequency conversion on the original observation data to obtain an observation power spectrum includes: Performing discrete Fourier transform on the original observation data to obtain an original power spectrum; Performing interval averaging processing on the original power spectrum to obtain a stable power spectrum; The stable power spectrum is normalized to obtain a normalized power spectrum, and the normalized power spectrum is used as the observed power spectrum.

6. The method for raw data noise identification and data reconstruction of an airborne eddy covariance flux observation system according to any one of claims 1 to 4, characterized in that: The original observation data includes at least one of three-dimensional wind speed, air temperature and target gas concentration.

7. A device for identifying noise in raw data and reconstructing data of an airborne eddy covariance flux observation system, characterized in that: include: A data acquisition unit, used to acquire raw observation data of an airborne eddy covariance flux observation system; A noise identification unit, configured to perform time-frequency conversion on the original observation data to obtain an observation power spectrum, compare the observation power spectrum with a standard power spectrum, and identify the noise type in the original observation data based on the comparison result; A data reconstruction unit, used for performing noise filtering and data reconstruction on the original observation data based on the noise type; The data reconstruction unit is specifically used for: If the noise type is colored noise, then for the frequency band interval where the colored noise appears, the noise effect generated by the colored noise is quantified in the frequency domain, and the noise effect is filtered out during data reconstruction to obtain reconstructed observation data; The data reconstruction unit is specifically used for: For the frequency band interval where the colored noise appears, calculating the ratio between the observed power spectrum mean and the standard power spectrum mean; Based on the ratio, the reconstructed observation data is obtained by the following formula: In the formula, S new (n) is the data sequence of reconstructed observation data, F(f) is the discrete Fourier transform coefficient of the original observation data, N is the data sequence length of the original observation data, [b1, b2] is the frequency band where colored noise appears, a is the ratio between the mean of the observed power spectrum and the mean of the standard power spectrum, fl is the lowest frequency of discrete Fourier transform, and f0 is the Nyquist frequency.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the raw data noise identification and data reconstruction method of the airborne eddy covariance flux observation system as described in any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for raw data noise identification and data reconstruction of an airborne eddy covariance flux observation system as claimed in any one of claims 1 to 6 is implemented.

Images