Hail disaster affected area satellite remote sensing tracking evaluation method based on changes

Abstract

The invention provides a change-based satellite remote sensing tracking evaluation method for a hail disaster affected area. A change-based hail disaster affected area satellite remote sensing tracking evaluation method comprises the following steps: S1, based on hectometer-level spatial resolution observation data of a FY-4 stationary meteorological satellite, continuously acquiring multispectral images of a target area at a minute-level time resolution, monitoring the initial occurrence position of a severe convective cloud cluster in real time, and determining the initial occurrence position of the severe convective cloud cluster according to the initial occurrence position of the severe convective cloud cluster; the space expansion path and the intensity evolution process are tracked in a time-phase-by-time manner through a continuous tracking technology; s2, aiming at the full life cycle of the severe convection process, synchronously extracting a time-space sequence data set of visible light wave band reflectivity, infrared brightness temperature, an NDVI vegetation index and an NDWI water body index; the hail disaster affected area satellite remote sensing tracking evaluation method based on changes has the advantages of being accurate in recognition, capable of conducting high-frequency tracking on severe convective cloud clusters and high in evaluation result reliability.

Description

Technical Field

[0001] The present invention relates to the technical field of meteorological disaster monitoring, and in particular to a satellite remote sensing tracking and evaluation method for a wind and hail disaster impact area based on changes. Background Art

[0002] Traditional wind and hail disaster monitoring mainly relies on data from meteorological radars and ground observation stations, which have problems such as incomplete spatial coverage and long update time intervals (usually ≥6 minutes).

[0003] Although existing satellite remote sensing technology can achieve large-scale monitoring, it has the following defects: the single-source satellite data identification method relies on a single threshold (such as the infrared brightness temperature threshold), which is easily affected by cloud changes and surface heterogeneity, resulting in a misjudgment rate of >30%; the calculation of disaster-affected areas lacks a dynamic correction mechanism and does not consider the verification feedback of high-resolution satellites; a minute-level time series disaster development process tracking model has not been established, and the disaster evolution rate cannot be quantified.

[0004] Therefore, it is necessary to provide a satellite remote sensing tracking and assessment method for wind and hail disaster impact areas based on changes to solve the above technical problems. Summary of the Invention

[0005] The technical problem solved by the present invention is to provide a satellite remote sensing tracking and assessment method for wind and hail disaster impact areas based on changes, which has relatively accurate identification, can track strong convective cloud clusters at high frequency, and has highly reliable assessment results.

[0006] To solve the above technical problems, the present invention provides a method for tracking and evaluating wind and hail disaster impact areas based on satellite remote sensing changes, comprising the following steps:

[0007] S1: Based on observation data with a spatial resolution of 100 meters from the Fengyun-4 geostationary meteorological satellite, multispectral images of the target area are continuously acquired with a temporal resolution of minutes. The initial location of severe convective clouds is monitored in real time, and their spatial expansion path and intensity evolution are tracked phase by phase through continuous tracking technology.

[0008] S2: For the entire life cycle of severe convection, we simultaneously extract spatiotemporal series datasets of visible light reflectance, infrared brightness temperature, NDVI vegetation index, and NDWI water index. By calculating the statistical distribution differences of each index during the pre-disaster baseline period, the disaster development period, and the dissipation period, we construct a multi-feature judgment system that includes a band reflectance mutation threshold, a thermal infrared brightness temperature gradient threshold, and a NDVI / NDWI joint variation threshold.

[0009] S3: A spatial clustering algorithm is used to aggregate and analyze pixels that simultaneously meet the following conditions: a decrease in visible light reflectance ≥ 15%, a change in infrared brightness temperature gradient ≥ 3K / 10min, an NDVI attenuation exceeding 30% of the pre-disaster mean, and an NDWI fluctuation greater than 2 standard deviations. Connected areas that meet the multi-feature joint criteria are marked as the core impact area of the wind and hail disaster, and their dynamic area change curves are calculated.

[0010] S4: Within 24 hours after the end of the severe convective process, dispatch a meter-level resolution optical satellite to conduct stereoscopic imaging of the core impact area. By comparing the texture consistency between the disaster identification area and the actual damage characteristics, the confusion matrix method is used to calculate the disaster boundary positioning error, and the estimated impact area is corrected based on the high-resolution imagery.

[0011] Preferably, in S2, the calculation of the reflectivity in the visible light band adopts:

[0012] The 0.55μm, 0.65μm, and 0.85μm channels of the Fengyun-4 satellite;

[0013] The surface reflectance map is generated through radiometric calibration and atmospheric correction.

[0014] Preferably, in S2, the calculation of the NDVI / NDWI combined variation threshold satisfies:

[0015]

[0016] Among them, NDVI t NDWI t is the index value of the disaster development period;

[0017] NDVI0 and NDWI0 are the mean index values of the pre-disaster baseline period;

[0018] σNDVI and σNDWI are the standard deviations of the indices during the pre-disaster baseline period;

[0019] When Δ 复合 >4.0 is considered an area with significant abnormal vegetation-water characteristics.

[0020] Preferably, in S3, the spatial clustering algorithm includes:

[0021] Initial screening was performed on pixels with infrared brightness temperature gradient ≥ 3K / 10min;

[0022] Use morphological closing operation to eliminate discrete noise points;

[0023] The vector boundary of the disaster patch is generated through 8-neighborhood connectivity analysis.

[0024] Preferably, in S4, the texture consistency analysis includes:

[0025] Use the ResNet-50 deep learning model to extract damage features from high-resolution images;

[0026] Calculate the feature similarity between the disaster identification area and the high-resolution image;

[0027] When the similarity is less than 0.7, it is marked as a misjudged area.

[0028] Preferably, in S3, the correction formula of the dynamic area change curve is:

[0029]

[0030] Among them, A 原始 This is the initial estimated area based on the Fengyun-4 satellite;

[0031] η is the positioning error coefficient of claim 4;

[0032] T 延时 The time delay for high-resolution satellite transits;

[0033] The constant term -2 is the timeliness compensation factor, which is determined through historical data regression analysis.

[0034] Preferably, in S1, the time parameters of the continuous tracking technology are:

[0035] Initial monitoring phase: satellite images are acquired every 2 minutes;

[0036] Disaster development period: update the cloud movement vector field every 5 minutes;

[0037] Dissipation period: The residual cloud distribution is recorded every 10 minutes.

[0038] Preferably, in S2, the pre-disaster reference period is defined as:

[0039] Stable weather period within 6 hours before the start of severe convective process;

[0040] The average value of the index of three consecutive phases was selected as the baseline value;

[0041] It is required that the standard deviation of each index during the base period is less than 20% of the preset sensitivity threshold.

[0042] Preferably, in S4, revising the estimated value of the affected area includes:

[0043] Establish a mixed pixel decomposition model:

[0044]

[0045] Among them, ρ 灾前 , ρ 灾后are the reflectivity of the pixels before and after the disaster, ρ min Minimum reflectivity in the area;

[0046] When f 损毁 When >0.6, it is judged as a completely damaged area.

[0047] Preferably, it also includes:

[0048] Overlay the disaster core impact area with the agricultural land vector map;

[0049] Calculate the percentage of disaster-affected areas of different crop types;

[0050] Generate a heat map of the economic loss level distribution.

[0051] Compared with related technologies, the satellite remote sensing tracking and assessment method for wind and hail disaster impact areas based on changes provided by the present invention has the following beneficial effects:

[0052] The present invention provides a satellite remote sensing tracking and assessment method for wind and hail disaster impact areas based on changes. Through minute-level observations based on the Fengyun-4 satellite, the disaster identification response time can be shortened, while the high-frequency tracking capability of severe convective cloud clusters can be improved. Through continuous tracking technology, the disaster impact range can be predicted in advance, thereby gaining a critical time window for emergency response. The multi-feature judgment system integrates visible light, infrared, vegetation, and water body indices to effectively reduce the misjudgment rate. The error coefficient η works synergistically with the mixed pixel decomposition model to improve the accuracy of area estimation. The threshold weight and timeliness compensation factor can be dynamically adjusted according to regional characteristics, which can improve the applicability to various terrains such as plains and mountains.

[0053] The present invention provides a satellite remote sensing tracking and assessment method for wind and hail disaster impact areas based on changes, which can be used to test the accuracy of severe convection forecast areas, provide test samples for numerical forecasts, and thus better improve the forecast quality; and provide a quasi-scientific and accurate basis (affected areas, area, degree, etc.) for agricultural insurance loss assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 The present invention provides a flow chart of the satellite remote sensing tracking and assessment method for the wind and hail disaster impact area based on changes. DETAILED DESCRIPTION

[0055] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0056] Please refer to Figure 1 ,in, Figure 1 The present invention provides a flow chart of a method for tracking and assessing wind and hail disaster impact areas based on satellite remote sensing. The method includes the following steps:

[0057] S1: Based on observation data with a spatial resolution of 100 meters from the Fengyun-4 geostationary meteorological satellite, multispectral images of the target area are continuously acquired with a temporal resolution of minutes. The initial location of severe convective clouds is monitored in real time, and their spatial expansion path and intensity evolution are tracked phase by phase through continuous tracking technology.

[0058] The time parameters of the continuous tracking technology are as follows: during the initial monitoring phase (cloud top temperature ≤ -52°C), satellite images are acquired every 2 minutes; during the disaster development phase (cloud top temperature gradient ≥ 5°C / 10 minutes), the cloud movement vector field is updated every 5 minutes; during the dissipation phase (cloud top temperature rises above -30°C), the residual cloud distribution is recorded every 10 minutes.

[0059] The cloud movement vector field is calculated based on the optical flow method. The formula is:

[0060]

[0061] Among them, I(x, y, t) is the brightness temperature field of the time series, v x , v y is the moving velocity component.

[0062] S2: For the entire life cycle of severe convection, we simultaneously extract spatiotemporal series datasets of visible light reflectance, infrared brightness temperature, NDVI vegetation index, and NDWI water index. By calculating the statistical distribution differences of each index during the pre-disaster baseline period, the disaster development period, and the dissipation period, we construct a multi-feature judgment system that includes a band reflectance mutation threshold, a thermal infrared brightness temperature gradient threshold, and a NDVI / NDWI joint variation threshold.

[0063] The calculation of visible light reflectance includes: using the 0.55μm (green light), 0.65μm (red light) and 0.85μm (near infrared) channels of the Fengyun-4 satellite; generating a surface reflectance map through radiometric calibration and atmospheric correction;

[0064] Among them, through the radiation calibration formula L λ =K λ ×DN+B λ The original DN value is converted into radiance, and then the surface reflectance ρ is output through the 6S atmospheric correction model;

[0065] The calculation of the NDVI / NDWI joint variation threshold satisfies:

[0066]

[0067] Among them, NDVI t NDWI t is the index value of the disaster development period;

[0068] NDVI0 and NDWI0 are the mean index values of the pre-disaster baseline period;

[0069] σNDVI and σNDWI are the standard deviations of the indices during the pre-disaster baseline period;

[0070] When Δ 复合 >4.0 is considered an area with significant abnormal vegetation-water characteristics.

[0071] The calculation formulas for NDVI and NDWI are:

[0072]

[0073] The pre-disaster benchmark period is defined as: a period of stable weather within 6 hours before the start of a severe convective process; the average index value of three consecutive time phases is selected as the benchmark value; and the standard deviation of each index during the benchmark period is required to be less than 20% of the preset sensitivity threshold.

[0074] S3: A spatial clustering algorithm is used to aggregate and analyze pixels that simultaneously meet the following conditions: a decrease in visible light reflectance ≥ 15%, a change in infrared brightness temperature gradient ≥ 3K / 10min, an NDVI attenuation exceeding 30% of the pre-disaster mean, and an NDWI fluctuation greater than 2 standard deviations. Connected areas that meet the multi-feature joint criteria are marked as the core impact area of the wind and hail disaster, and their dynamic area change curves are calculated.

[0075] The spatial clustering algorithm includes: preliminary screening of pixels with infrared brightness temperature gradient ≥ 3K / 10min; using morphological closing operation to eliminate discrete noise points; generating disaster patch vector boundaries through 8-neighborhood connectivity analysis;

[0076] The determination formula of the variation intensity index VI is:

[0077]

[0078] The correction formula of the dynamic area change curve is:

[0079]

[0080] Among them, A 原始 This is the initial estimated area based on the Fengyun-4 satellite;

[0081] η is the positioning error coefficient;

[0082] T 延时 The time delay for high-resolution satellite transits;

[0083] The constant term -2 is the timeliness compensation factor, which is determined through regression analysis of historical data;

[0084] Among them, the calculation formula of the positioning error coefficient η is:

[0085]

[0086] Among them, TP is the FY-4 identification area, HR is the high-resolution satellite verification area, and the area correction is triggered when η>0.3.

[0087] S4: Within 24 hours after the severe convection process ends, dispatch a meter-level resolution optical satellite to conduct stereoscopic imaging of the core impact area. By comparing the texture consistency between the disaster identification area and the actual damage characteristics, the confusion matrix method is used to calculate the disaster boundary positioning error, and the estimated impact area is corrected based on the high-resolution imagery.

[0088] Texture consistency analysis includes: using the ResNet-50 deep learning model to extract damage features from high-resolution images; calculating the feature similarity between the disaster identification area and the high-resolution image; and marking the area as misjudged when the similarity is less than 0.7;

[0089] The revised impact area estimates include:

[0090] Establish a mixed pixel decomposition model:

[0091]

[0092] Among them, ρ 灾前 , ρ 灾后 are the reflectivity of the pixels before and after the disaster, ρ min Minimum reflectivity in the area;

[0093] When f 损毁 When >0.6, it is considered a completely damaged area.

[0094] In this embodiment, the method further includes: superimposing the disaster core impact area and the agricultural land vector map; calculating the proportion of disaster-affected areas of different crop types; and generating a heat map of economic loss level distribution.

[0095] The overlay of the disaster core impact area and the agricultural land vector map includes the following input data:

[0096] The revised vector boundary of the disaster core impact area;

[0097] Agricultural land parcel vector map, including crop type, planting area, and historical yield attribute fields;

[0098] The overlay method is to use the spatial join function of the geographic information system (GIS) to perform intersection calculation on the disaster-affected area and the agricultural plots to generate an overlay result layer;

[0099] The calculation formula for the disaster-stricken area in the statistics of the proportion of disaster-stricken areas of different crop types is:

[0100]

[0101] Among them, S 受灾，c is the affected area of crop type c (unit: km 2 );A i,c is the planting area of crop c in the i-th agricultural plot (unit: km 2 );f 损毁,i is the damage ratio of the i-th plot.

[0102] The formula for calculating the proportion is:

[0103]

[0104] Among them, generating a heat map of economic loss level distribution includes:

[0105] Yield loss calculation formula is:

[0106] L c =S 受灾,c ×P c ×V c

[0107] Among them, P c is the market price of crop c; V c is the expected yield of crop c.

[0108] The economic loss levels can be divided into:

[0109] Red: L c >1 million yuan / km 2 Orange: 500,000 yuan / km 2 ≤L c ≤1 million yuan / km 2 ; Yellow: L c <500,000 yuan / km 2 ;

[0110] Visualization method: A continuous heat map is generated based on the kernel density estimation algorithm, with the bandwidth set to 2 km; the loss value is mapped using the HSL color space.

[0111] Compared with related technologies, the satellite remote sensing tracking and assessment method for wind and hail disaster impact areas based on changes provided by the present invention has the following beneficial effects:

[0112] The present invention provides a satellite remote sensing tracking and assessment method for wind and hail disaster impact areas based on changes. Through minute-level observations based on the Fengyun-4 satellite, the disaster identification response time can be shortened, while the high-frequency tracking capability of severe convective cloud clusters can be improved. Through continuous tracking technology, the disaster impact range can be predicted in advance, thereby gaining a critical time window for emergency response. The multi-feature judgment system integrates visible light, infrared, vegetation, and water body indices to effectively reduce the misjudgment rate. The error coefficient η works synergistically with the mixed pixel decomposition model to improve the accuracy of area estimation. The threshold weight and timeliness compensation factor can be dynamically adjusted according to regional characteristics, which can improve the applicability to various terrains such as plains and mountains.

[0113] The present invention provides a satellite remote sensing tracking and assessment method for wind and hail disaster impact areas based on changes, which can be used to test the accuracy of severe convection forecast areas, provide test samples for numerical forecasts, and thus better improve the forecast quality; and provide a quasi-scientific and accurate basis (affected areas, area, degree, etc.) for agricultural insurance loss assessment.

[0114] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A satellite remote sensing tracking and assessment method for wind and hail disaster impact areas based on changes, characterized in that: The following steps are involved: S1: Based on observation data with a spatial resolution of 100 meters from the Fengyun-4 geostationary meteorological satellite, multispectral images of the target area are continuously acquired with a temporal resolution of minutes. The initial location of severe convective clouds is monitored in real time, and their spatial expansion path and intensity evolution are tracked phase by phase through continuous tracking technology. S2: For the entire life cycle of severe convection, we simultaneously extract spatiotemporal series datasets of visible light reflectance, infrared brightness temperature, NDVI vegetation index, and NDWI water index. By calculating the statistical distribution differences of each index during the pre-disaster baseline period, the disaster development period, and the dissipation period, we construct a multi-feature judgment system that includes a band reflectance mutation threshold, a thermal infrared brightness temperature gradient threshold, and a NDVI / NDWI joint variation threshold. S3: A spatial clustering algorithm is used to aggregate and analyze pixels that simultaneously meet the following conditions: a decrease in visible light reflectance ≥ 15%, a change in infrared brightness temperature gradient ≥ 3K / 10min, an NDVI attenuation exceeding 30% of the pre-disaster mean, and an NDWI fluctuation greater than 2 standard deviations. Connected areas that meet the multi-feature joint criteria are marked as the core impact area of the wind and hail disaster, and their dynamic area change curves are calculated. S4: Within 24 hours after the end of the severe convective process, dispatch a meter-level resolution optical satellite to conduct stereoscopic imaging of the core impact area. By comparing the texture consistency between the disaster identification area and the actual damage characteristics, the confusion matrix method is used to calculate the disaster boundary positioning error, and the estimated impact area is corrected based on the high-resolution imagery.

2. The satellite remote sensing tracking and assessment method for wind and hail disaster impact areas based on changes according to claim 1 is characterized in that: In S2, the reflectivity in the visible light band is calculated using: The 0.55μm, 0.65μm, and 0.85μm channels of the Fengyun-4 satellite; The surface reflectance map is generated through radiometric calibration and atmospheric correction.

3. The satellite remote sensing tracking and assessment method for wind and hail disaster impact areas based on changes according to claim 1 is characterized in that: In S2, the calculation of the NDVI / NDWI joint variation threshold satisfies: Among them, NDVI t NDWI t is the index value of the disaster development period; NDVI0 and NDWI0 are the mean index values of the pre-disaster baseline period; σNDVI and σNDWI are the standard deviations of the indices during the pre-disaster baseline period; When Δ 复合 >4.0 is considered an area with significant abnormal vegetation-water characteristics.

4. The satellite remote sensing tracking and assessment method for wind and hail disaster impact areas based on changes according to claim 1 is characterized in that: In S3, the spatial clustering algorithm includes: Initial screening was performed on pixels with infrared brightness temperature gradient ≥ 3K / 10min; Use morphological closing operation to eliminate discrete noise points; The vector boundary of the disaster patch is generated through 8-neighborhood connectivity analysis.

5. The satellite remote sensing tracking and assessment method for wind and hail disaster impact areas based on changes according to claim 1 is characterized in that: In S4, the texture consistency analysis includes: Use the ResNet-50 deep learning model to extract damage features from high-resolution images; Calculate the feature similarity between the disaster identification area and the high-resolution image; When the similarity is less than 0.7, it is marked as a misjudged area.

6. The satellite remote sensing tracking and assessment method for wind and hail disaster impact areas based on changes according to claim 1 is characterized in that: In S3, the correction formula of the dynamic area change curve is: Among them, A 原始 This is the initial estimated area based on the Fengyun-4 satellite; η is the positioning error coefficient of claim 4; T 延时 The time delay for high-resolution satellite transits; The constant term -2 is the timeliness compensation factor, which is determined through historical data regression analysis.

7. The satellite remote sensing tracking and assessment method for wind and hail disaster impact areas based on changes according to claim 1 is characterized in that: In S1, the time parameters of the continuous tracking technology are: Initial monitoring phase: satellite images are acquired every 2 minutes; Disaster development period: update the cloud movement vector field every 5 minutes; Dissipation period: The residual cloud distribution is recorded every 10 minutes.

8. The satellite remote sensing tracking and assessment method for wind and hail disaster impact areas based on changes according to claim 3 is characterized in that: In S2, the pre-disaster reference period is defined as: Stable weather period within 6 hours before the start of severe convective process; The average value of the index of three consecutive phases was selected as the baseline value; It is required that the standard deviation of each index during the base period is less than 20% of the preset sensitivity threshold.

9. The satellite remote sensing tracking and assessment method for wind and hail disaster impact areas based on changes according to claim 1 is characterized in that: In S4, revising the estimated value of the affected area includes: Establish a mixed pixel decomposition model: Among them, ρ 灾前 , ρ 灾后 are the reflectivity of the pixels before and after the disaster, ρ min Minimum reflectivity in the area; When f 损毁 When >0.6, it is considered a completely damaged area.

10. The satellite remote sensing tracking and assessment method for wind and hail disaster impact areas based on changes according to claim 1, characterized in that: Also includes: Overlay the disaster core impact area with the agricultural land vector map; Calculate the percentage of disaster-affected areas of different crop types; Generate a heat map of the economic loss level distribution.

Images