Fast registration method for forward-looking and backward-looking observations of space-borne cloud radar beam conical spiral scanning

Abstract

The application discloses a kind of spaceborne cloud radar beam conical spiral scanning forward-looking and rearview observation fast registration method, belong to meteorological radar signal processing and spaceborne observation registration technical field.The method includes: according to satellite orbit parameter and scanning geometry, the ground intersection of each pulse beam footprint is calculated, and footprint point set is constructed;According to the interval of scanning circle, the potential overlapping observation is traversed, and the candidate point pair is generated in combination with spatial distance gate;Matching cost function is constructed with footprint distance as main item and fusion observation consistency constraint, and effective registration point pair is obtained under one-to-one correspondence constraint;Registration point pair obtained by different scanning circle interval and different distance gate is accumulated and deduplicated, and global registration point set is formed.The present application makes full use of multiple overlapping observations of spiral trajectory, significantly increases the same sample, improves the reliability and precision of registration under the condition of high-speed spaceborne platform, and can provide high-quality input for subsequent three-dimensional observation data fusion, data quality control and data assimilation.

Description

Technical Field

[0001] This invention belongs to the field of meteorological radar signal processing and satellite remote sensing measurement technology, and in particular relates to a method for determining the registration and matching points for the consistency of forward and backward observations of a spaceborne conical scanning Doppler cloud radar system. This method can be used to support the fusion and assimilation of three-dimensional observation data of cloud precipitation systems. Background Technology

[0002] When a spaceborne Doppler cloud radar operates in a conical scanning configuration, the beam forms a spiral observation trajectory on the Earth's surface. Due to the high-speed orbital movement of the satellite platform and the continuous changes in scanning geometry, there are orbital displacements and scanning phase differences between different scanning circles. This causes forward-looking and backward-looking observations to exhibit asynchronous characteristics in space and time, and the complex observation geometry makes it difficult to establish a stable correspondence for the same spatial location.

[0003] Existing registration methods often rely on simplified correspondences based on time or azimuth, or only utilize overlapping observations at finite intervals for local matching. When platform motion effects are significant or the scan circle intervals are large, mismatches, missed matches, and insufficient effective samples are prone to occur, thus affecting the quality of multi-circle observation fusion and the stability and accuracy of subsequent 3D wind field inversion.

[0004] In recent years, to address the critical weakness of observing wind fields within clouds, major international space agencies have accelerated the development of new types of spaceborne Doppler radar missions. The European Space Agency (ESA) has selected WIVERN and is promoting its engineering development. This type of spaceborne conical scanning Doppler radar possesses near all-weather, cloud-penetrating observation capabilities and wide-area coverage, and is considered an important development direction for global three-dimensional wind field detection. Meanwhile, my country is also accelerating the development of a new generation of spaceborne active microwave detection payloads and observation systems, striving to achieve key breakthroughs in the observation capabilities of three-dimensional cloud precipitation structure and wind fields within clouds, driving the development of high-end remote sensing equipment and data service systems, and meeting the needs of new-type productivity.

[0005] However, the spatiotemporal asynchrony of foresight and backsight observations and the coupling effect of cross-circle displacement under the conical scanning system make it difficult to reliably determine the matching points of overlapping observations within the interval of multiple scanning circles, thus restricting the effect of observation fusion and three-dimensional wind field inversion. Therefore, there is an urgent need for a method that can systematically search for overlapping observations within the interval of multiple scanning circles and introduce consistency constraints to achieve reliable registration point determination, so as to improve the matching accuracy and engineering applicability.

[0006] Therefore, a method is needed that can systematically search all overlapping observations within a multi-scan interval and reliably determine registration points through consistency constraints. Summary of the Invention

[0007] Purpose of the invention: The purpose of this invention is to provide a rapid registration method for forward and backward observations of spaceborne cloud radar beam conic spiral scanning. By using an incremental traversal matching strategy, candidate gating, multi-target matching cost, and consistency constraint solution, reliable one-to-one registration point determination of forward and backward observations is achieved, and a global registration point set is formed, providing high-quality co-location observation samples for subsequent cloud precipitation system three-dimensional observation data fusion and data assimilation processing.

[0008] Technical solution: The present invention provides a rapid registration method for forward-looking and backward-looking observations using a satellite-borne cloud radar beam conic spiral scanning system, comprising the following steps:

[0009] S1. Acquire observation data from the spaceborne conical scanning radar within multiple consecutive scanning circles. The observation data includes radial Doppler velocity. Reflectivity Angle of incidence Azimuth Slope distance Satellite platform altitude and along-rail speed Where the radial Doppler velocity is the radial velocity after platform motion compensation has been completed;

[0010] S2. Divide the continuous observation into multiple scanning circles according to the azimuth circumference point to obtain the first... A set of pulse indices for each scan cycle ;

[0011] S3. Based on satellite orbit parameters and radar scanning geometry, calculate the location coordinates of the footprint in the geographic coordinate system for each scan circle, each range gate, and each pulse, and construct a footprint point set. ;

[0012] S4. Perform incremental traversal matching with the scan circle interval as the independent variable. Starting from the adjacent scan circle, increase the interval for subsequent scan circle pairs and generate candidate point pairs until the termination condition is met, so as to cover all cross-registration points that can overlap.

[0013] S5. Construct a matching cost function for candidate point pairs. The matching cost function includes a footprint spatial distance term and at least one of reflectivity, radial velocity, incident angle and time consistency terms.

[0014] S6. Solve for the set of registered point pairs under matching consistency constraints and distance threshold constraints. The matching consistency constraints include at least one of the following: one-to-one correspondence constraint, bidirectional consistency constraint, global optimum or near-optimal constraint, and iterative convergence consistency constraint.

[0015] S7. The obtained registration point pairs are classified and marked as foresight and backsight according to the azimuth symbol or equivalent sight criterion. The registration point pairs obtained by different scanning circle intervals, different scanning circles and different distance gates are accumulated and deduplicated to obtain the global registration point set.

[0016] Furthermore, in step S2, dividing the continuous observation into multiple scanning circles according to the azimuth angle loop point specifically satisfies: for the discrete azimuth angle sequence calculate ,when Time determination Here is the point of return, where This is a preset threshold.

[0017] Furthermore, step S3 specifically includes: the first Each scanning circle, pulse within the circle Distance Gate The plane coordinates of the footprint are:

[0018]

[0019] in , The radar range sampling interval or range gate width is determined by the system bandwidth and sampling configuration. This represents the relative time within the scan circle compared to the start time of the scan circle. For the first Within the first scanning circle The incident angle of each pulse, For the first Within the first scanning circle The azimuth angle of each pulse.

[0020] Furthermore, in step S4, the termination condition includes at least one of the following:

[0021] a. The scanning circle interval reaches the preset upper limit. ;

[0022] b. The number of newly registered point pairs that pass the threshold constraint after a continuous preset number of interval increment matching is zero.

[0023] c. The overlap width criterion estimated by the displacement along the track is no longer satisfied.

[0024] Furthermore, in step S4, spatial gating is applied after the candidate point pairs are generated, satisfying:

[0025] ,

[0026] in, and The scan circle number to be matched. Number the distance gates; Indicates the first The pulse index used for matching within each scan cycle. Indicates the first The pulse index used for matching within each scan cycle;

[0027]

[0028] in , These are the coordinate components of the footprint point in the planar coordinate system. Indicates transpose. Describing the Euclidean norm, The threshold radius.

[0029] Furthermore, in step S5, the matching cost function is:

[0030] ,

[0031] in, These are the weighting coefficients. This is the normalization scale.

[0032] Furthermore, in step S6, the one-to-one correspondence constraint determines the set of registration point pairs through global optimization, satisfying:

[0033] ,

[0034] ,

[0035] ,

[0036] ,

[0037] in, For scanning circle With scanning circle At the distance of the door The following set of candidate point pairs is generated by spatial gating. For candidate point pairs The matching cost, This indicates that the candidate point pair was selected as a valid registration point pair.

[0038] Furthermore, in step S6, the bidirectional consistency constraint includes: forward matching mapping. Reverse matching mapping satisfy:

[0039]

[0040] in, This represents the bidirectional consistency distance threshold, used to limit the upper limit of the resubmission error after forward matching and then reverse matching;

[0041] And the iterative convergence consistency constraint satisfies: when

[0042]

[0043] Stop iteration when, where For the first The set of registration point pairs in the next iteration This represents the iteration convergence threshold.

[0044] Furthermore, in step S6, the threshold filtering for the distance threshold constraint includes at least one of the following:

[0045] a. Distance threshold , This represents the distance threshold, used to limit the upper limit of the footprint space distance between candidate point pairs;

[0046] b. Cost threshold , Indicates candidate point pairs The matching cost, This represents the cost threshold, used to remove point pairs with excessive matching costs.

[0047] c. Observational consistency threshold or , Indicates the reflectivity consistency threshold. These represent the radial velocity consistency thresholds, used to limit the upper limit of differences between registration point pairs in reflectivity and radial velocity observations.

[0048] Furthermore, in step S7, the accumulation and deduplication satisfy the requirements of the fixed... Union of the registration sets obtained at different intervals:

[0049]

[0050] in, For the scan circle number, Number the distance gate. Scan circle interval, Indicates scanning circle With scanning circle At the distance of the door The resulting set of valid registration point pairs consists of observation index pairs. ,in , ; It represents the union of sets.

[0051] The set of globally registered point pairs formed by all scan circles and the distance gate is:

[0052]

[0053] When the same observation index participates in multiple registration point pairs, the cost of preservation is retained. The smallest registration point pair, or a limit on the number of registration point pairs that the same observation index can participate in at most a preset number of times.

[0054] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages:

[0055] 1. This invention establishes a footprint localization model based on scanning geometry to obtain the footprint point set of each pulse in the geographic coordinate system; with adjacent scanning circles as the starting interval, incremental traversal matching is performed on subsequent scanning circle pairs to cover all overlapping observation intersections; candidate point pairs are generated through spatial gating to construct multi-target matching cost, and effective registration point pairs are obtained by global optimization or iterative consistency solution under one-to-one correspondence and other consistency constraints; registration point pairs obtained by different intervals and different distance gates are accumulated and deduplicated to form a global registration point set.

[0056] 2. This invention, through an incremental traversal matching strategy, can cover overlapping observations across multiple scan intervals, effectively mining the multiple intersection information generated by the spiral trajectory and significantly increasing the number of effective registration samples; by jointly solving the multi-target matching cost and consistency constraints, it improves the reliability and anti-misregistration capability of registration; by using gating and spatial indexing to reduce computational complexity, it is suitable for large-scale data processing; and it provides a basis for subsequent fusion of three-dimensional observation data of cloud precipitation systems and data assimilation processing of co-located observation samples. Attached Figure Description

[0057] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0058] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0059] like Figure 1 As shown, the present invention provides a rapid registration method for forward-looking and backward-looking observations using a beam conic spiral scanning method for spaceborne cloud radar, comprising the following steps:

[0060] S1. Acquire observation data from the spaceborne conical scanning radar within multiple consecutive scanning circles. The observation data includes radial Doppler velocity. Reflectivity Angle of incidence Azimuth Slope distance Satellite platform altitude and along-rail speed Where the radial Doppler velocity is the radial velocity after platform motion compensation has been completed;

[0061] S2. Divide the continuous observation into multiple scanning circles according to the azimuth circumference point to obtain the first... A set of pulse indices for each scan cycle ;

[0062] S3. Based on satellite orbit parameters and radar scanning geometry, calculate the location coordinates of the footprint in the geographic coordinate system for each scan circle, each range gate, and each pulse, and construct a footprint point set. ;

[0063] S4. Perform incremental traversal matching with the scan circle interval as the independent variable. Starting from the adjacent scan circle, increase the interval for subsequent scan circle pairs and generate candidate point pairs until the termination condition is met, so as to cover all cross-registration points that can overlap.

[0064] S5. Construct a matching cost function for candidate point pairs. The matching cost function includes a footprint spatial distance term and at least one of reflectivity, radial velocity, incident angle and time consistency terms.

[0065] S6. Solve for the set of registered point pairs under matching consistency constraints and distance threshold constraints. The matching consistency constraints include at least one of the following: one-to-one correspondence constraint, bidirectional consistency constraint, global optimum or near-optimal constraint, and iterative convergence consistency constraint.

[0066] S7. The obtained registration point pairs are classified and marked as foresight and backsight according to the azimuth symbol or equivalent sight criterion. The registration point pairs obtained by different scanning circle intervals, different scanning circles and different distance gates are accumulated and deduplicated to obtain the global registration point set.

[0067] Set the scanning circle number . No. The set of pulse indices of the circle is Distance from gate number slant distance .

[0068] Define footprint points:

[0069]

[0070] Footprint Collection:

[0071]

[0072] azimuth sequence Calculate the difference ,when The point is determined as the wraparound point, and an index set for each scan circle is constructed accordingly. .

[0073] For the first Circle, distance gate Intra-circle pulse Calculate footprint coordinates:

[0074] (1)

[0075] For fixed Starting from the adjacent scan circle, define the scan circle to be matched as... and order Increasing. (Regarding) and First, spatial gating is performed to generate a set of candidate point pairs:

[0076] (2)

[0077] in It can be set according to the Euler grid resolution or equivalent footprint size.

[0078] The increment stops when the termination condition is met. Termination conditions may include: reaching a preset upper limit. If, after several consecutive increments, no new valid registration point pairs are generated, or the track displacement estimate no longer meets the overlap width criterion.

[0079] For candidate point pairs Cost of constructing a match:

[0080] (3)

[0081] The weights and scales are used to balance the dimensions and confidence levels of different physical quantities, and can be adaptively set according to system experience or data statistics.

[0082] Global assignment optimization under one-to-one correspondence constraints:

[0083] (4)

[0084] , , (5)

[0085] And apply threshold filtering, for example:

[0086] (6)

[0087] This yields the set of effective registration point pairs under the distance gate for this circle pair. .

[0088] In another implementation, iterative consistency can be used, where the forward and reverse mappings satisfy a bidirectional consistency constraint:

[0089] (7)

[0090] And the rate of change of the set is used as the convergence criterion:

[0091] (8)

[0092] For fixed Union of registration sets with different intervals:

[0093]

[0094] A global set is formed for all scan circles and distance gates:

[0095]

[0096] When the same observation index participates in multiple registration point pairs, it is retained at the lowest cost:

[0097]

[0098] Alternatively, the number of registrations a single observation index can be limited to a preset number to improve overall stability.

[0099] In this embodiment, under the system parameters of a satellite platform altitude of 500 km, a satellite flight speed of 7.6 km / s, and a radar beam incident angle range of 25° to 45°, multiple sets of comparative experiments and statistical analyses were conducted on the search for spatial matching points in forward-looking and backward-looking observations under a helical scanning system. To ensure the objectivity and repeatability of the comparison results, the experiments were performed on the same computer platform configured with an i7-13700F CPU (2.10 GHz) and 32 GB of memory.

[0100] Under the above conditions, the method described in this invention and the conventional greedy search algorithm were used to search for matching points on the same dataset. The computation time, matching success rate, and positioning error of each experimental group were statistically analyzed, and the results are shown in Table 1. Experimental results show that, compared with the conventional greedy search algorithm, the method described in this invention can significantly reduce the computational load and computation time while maintaining or improving matching accuracy. This improves the processing efficiency and reliability of forward-looking and backward-looking spatial matching, demonstrating good engineering applicability and promotional value.

[0101] Table 1 Performance Comparison of Different Methods

[0102]

[0103] The above description is merely a preferred embodiment of the present invention. Any equivalent substitutions made by those skilled in the art to the candidate gating method, matching cost construction method, consistency constraint form, and optimization solution method without departing from the concept of the present invention should fall within the protection scope of the present invention.

Claims

1. A rapid registration method for forward-looking and backward-looking observations using beam conical spiral scanning of a spaceborne cloud radar, characterized in that, Includes the following steps: S1. Acquire observation data from the spaceborne conical scanning radar within multiple consecutive scanning circles. The observation data includes radial Doppler velocity. Reflectivity Angle of incidence Azimuth Slope distance Satellite platform altitude and along-rail speed Where the radial Doppler velocity is the radial velocity after platform motion compensation has been completed; S2. Divide the continuous observation into multiple scanning circles according to the azimuth circumference point to obtain the first... A set of pulse indices for each scan cycle ; S3. Based on satellite orbit parameters and radar scanning geometry, calculate the location coordinates of the footprint in the geographic coordinate system for each scan circle, each range gate, and each pulse, and construct a footprint point set. ; S4. Perform incremental traversal matching with the scan circle interval as the independent variable. Starting from the adjacent scan circle, increase the interval for subsequent scan circle pairs and generate candidate point pairs until the termination condition is met, so as to cover all cross-registration points that can overlap. S5. Construct a matching cost function for candidate point pairs. The matching cost function includes a footprint spatial distance term and at least one of reflectivity, radial velocity, incident angle and time consistency terms. S6. Solve for the set of registered point pairs under matching consistency constraints and distance threshold constraints. The matching consistency constraints include at least one of the following: one-to-one correspondence constraint, bidirectional consistency constraint, global optimum or near-optimal constraint, and iterative convergence consistency constraint. S7. The obtained registration point pairs are classified and marked as foresight and backsight according to the azimuth symbol or equivalent sight criterion. The registration point pairs obtained by different scanning circle intervals, different scanning circles and different distance gates are accumulated and deduplicated to obtain the global registration point set.

2. The rapid registration method for forward and backward observation of spaceborne cloud radar beam conical spiral scanning according to claim 1, characterized in that, In step S2, the continuous observation is divided into multiple scanning circles according to the azimuth angle loop point, specifically satisfying the following: for the discrete azimuth angle sequence calculate ,when Time determination Here is the point of return, where This is a preset threshold.

3. The rapid registration method for forward and backward observation of spaceborne cloud radar beam conical spiral scanning according to claim 1, characterized in that, Step S3 specifically refers to: Each scanning circle, pulse within the circle Distance Gate The plane coordinates of the footprint are: ； in , The radar range sampling interval or range gate width is determined by the system bandwidth and sampling configuration. This represents the relative time within the scan circle compared to the start time of the scan circle. For the first Within the first scanning circle The incident angle of each pulse, For the first Within the first scanning circle The azimuth angle of each pulse.

4. The rapid registration method for forward and backward observation of spaceborne cloud radar beam conical spiral scanning according to claim 1, characterized in that, In step S4, the termination condition includes at least one of the following: a. The scanning circle interval reaches the preset upper limit. ; b. The number of newly registered point pairs that pass the threshold constraint after a continuous preset number of interval increment matching is zero. c. The overlap width criterion estimated by the displacement along the track is no longer satisfied.

5. The rapid registration method for forward and backward observation of spaceborne cloud radar beam conical spiral scanning according to claim 1, characterized in that, In step S4, after the candidate point pairs are generated, spatial gating is applied to satisfy: ； in, and The scan circle number to be matched. Number the distance gates; Indicates the first The pulse index used for matching within each scan cycle. Indicates the first The pulse index used for matching within each scan cycle; in , These are the coordinate components of the footprint point in the planar coordinate system. Indicates transpose. Describing the Euclidean norm, The threshold radius.

6. The rapid registration method for forward and backward observation of spaceborne cloud radar beam conical spiral scanning according to claim 5, characterized in that, In step S5, the matching cost function is: ； in, These are the weighting coefficients. This is the normalization scale.

7. The rapid registration method for forward and backward observation of spaceborne cloud radar beam conical spiral scanning according to claim 1, characterized in that, In step S6, the one-to-one correspondence constraint determines the set of registration point pairs through global optimization, satisfying: ， ， ， ， in, For scanning circle With scanning circle At the distance of the door The following set of candidate point pairs is generated by spatial gating. For candidate point pairs The matching cost, This indicates that the candidate point pair was selected as a valid registration point pair.

8. The rapid registration method for forward and backward observation of spaceborne cloud radar beam conical spiral scanning according to claim 5, characterized in that, In step S6, the bidirectional consistency constraint includes: forward matching mapping. Reverse matching mapping satisfy: in, This represents the bidirectional consistency distance threshold, used to limit the upper limit of the resubmission error after forward matching and then reverse matching; And the iterative convergence consistency constraint satisfies: when ； Stop iteration when, where For the first The set of registration point pairs in the next iteration The cardinality of a set. This represents the symmetric difference or discrepancy set of the sets of registration point pairs from two iterations. This represents the iteration convergence threshold.

9. The rapid registration method for forward and backward observation of spaceborne cloud radar beam conical spiral scanning according to claim 5, characterized in that, In step S6, the threshold filtering for the distance threshold constraint includes at least one of the following: a. Distance threshold , This represents the distance threshold, used to limit the upper limit of the footprint space distance between candidate point pairs; b. Cost threshold , Indicates candidate point pairs The matching cost, This represents the cost threshold, used to remove point pairs with excessive matching costs. c. Observational consistency threshold or , Indicates the reflectivity consistency threshold. The radial velocity consistency threshold is used to limit the upper limit of the difference between the registration point pairs in reflectivity and radial velocity observations.

10. The rapid registration method for forward and backward observation of spaceborne cloud radar beam conical spiral scanning according to claim 1, characterized in that, In step S7, accumulation and deduplication satisfy the requirements of fixed... Union of the registration sets obtained at different intervals: in, For the scan circle number, Number the distance gate. Scan circle interval, Indicates scanning circle With scanning circle At the distance of the door The resulting set of valid registration point pairs consists of observation index pairs. ,in , ; Represents the union of sets; The set of globally registered point pairs formed by all scan circles and the distance gate is: ； When the same observation index participates in multiple registration point pairs, the cost of preservation is retained. The smallest registration point pair, or a limit on the number of registration point pairs that the same observation index can participate in at most a preset number of times.

Images