CN114112065A - Method for judging and recognizing fire danger by satellite remote sensing - Google Patents

Abstract

The invention discloses a method for judging and identifying a satellite remote sensing fire, which comprises the following steps: step 1, preprocessing data, acquiring original satellite data, and performing positioning, calibration, quality inspection and local map processing; step 2, fire point identification is carried out on the preprocessed satellite picture; step 3, judging and identifying the intensity level of the fire point; step 4, fire passing area identification and fire point space-time distribution statistics; step 5, generating a satellite remote sensing fire monitoring result; the method solves the technical problems that in the process of monitoring the satellite remote sensing fire, a meteorological satellite detector is interfered by solar radiation, and the brightness temperature of some special ground objects (interference sources for short) on the underlying surface in a remote sensing image exceeds a target heat source, so that false fire points are generated, and the final satellite remote sensing fire judgment and identification result has larger error.

Description

Method for judging and recognizing fire danger by satellite remote sensing

Technical Field

The invention belongs to the technical field of fire monitoring; in particular to a method for judging and identifying the satellite remote sensing fire danger.

Background

In winter and spring every year, the rainfall is less, air-dried materials are dry, forest and grassland fires are easy to occur, and serious threats are brought to the safety of forestry, agriculture and animal husbandry and people's lives and properties. Moreover, forest-covered mountainous areas are often rare, traffic blockage is caused, and the method has great limitation and difficulty only depending on manual monitoring. With the development of science and technology, satellite remote sensing is more and more widely applied to fire monitoring of forests, grasslands and the like. The meteorological satellite remote sensing has the characteristics of wide visual field, dense observation frequency and sensitivity to ground high-temperature heat sources, and gradually becomes an important means for monitoring forest and grassland fires.

The remote sensing fire danger of forest and grassland mainly utilizes the characteristic that a satellite detector has high sensitivity to abnormal temperature rise in fire in a middle infrared band. In the process of monitoring the satellite remote sensing fire, a meteorological satellite detector is interfered by solar radiation, the brightness temperature of some special ground objects (called interference sources for short) on the underlying surface in a remote sensing image exceeds a target heat source, so that a false fire point is generated, and the final satellite remote sensing fire judgment result has a larger error.

Disclosure of Invention

The technical problems to be solved by the invention are as follows: the method for judging the satellite remote sensing fire danger has the technical problems that in the process of monitoring the satellite remote sensing fire, a meteorological satellite detector is interfered by solar radiation, the brightness temperature of some special ground objects (interference sources for short) on an underlying surface in a remote sensing image can exceed a target heat source, so that false fire points are generated, the final satellite remote sensing fire danger judgment result has larger error, and the like.

The technical scheme of the invention is as follows:

a method for judging fire danger by satellite remote sensing comprises the following steps:

step 1, preprocessing data, acquiring original satellite data, and performing positioning, calibration, quality inspection and local map processing;

step 2, fire point identification is carried out on the preprocessed satellite picture;

step 3, judging and identifying the intensity level of the fire point;

step 4, fire passing area identification and fire point space-time distribution statistics;

and 5, generating a satellite remote sensing fire monitoring result.

The specific method for preprocessing the data comprises the following steps:

step 1.1, calibration and positioning: positioning and calibrating the original satellite data; after the image is positioned, the image is checked by a landmark, and if an error exists, the image is corrected by positioning; the requirement of the positioning precision after correction is not more than 1 pixel;

step 1.2, local area map processing: generating a local area map of a monitoring area from the projection of the preprocessed data, wherein the size of the image is set according to the range of the monitoring area;

and step 1.3, synthesizing the image by multiple channels.

Step 1.3 the multichannel composite image specifically includes:

step 1.3.1, multichannel synthesis of daytime images:

a) channel synthesis mode: respectively endowing middle infrared, near infrared and visible light channels with red, green and blue colors for synthesis;

b) channel enhancement: each channel is subjected to image enhancement to highlight heat source points and earth surface characteristics;

c) image color effect: the fire point is bright red; the vegetation area is green; clouds or smog are white or grey; the water body is blue, black or dark purple;

step 1.3.2, multichannel synthesis of night images:

a) channel synthesis mode: respectively endowing middle infrared, far infrared and far infrared channels with red, green and blue colors for synthesis;

b) channel enhancement: each channel is subjected to image enhancement to highlight heat source points and earth surface characteristics;

c) image color effect: the fire point is bright red; the fireless zone is dark grey.

The fire point judging method in the step 2 comprises an automatic judging method and a human-computer interaction judging method;

the specific automatic identification method comprises the following steps:

step 2.1, generating a pixel marking map, which specifically comprises the following steps:

(1) cloud region pixel mark

If the pixel satisfies R_VIS＞R_{VIS_TCTH}And T_FIR＜T_{FIR_TCTH}Marked as cloud area pixels;

in the formula: r_VISRepresents the visible channel reflectance, expressed as a percentage;

R_{VIS_TCTH}representing a cloud area judgment threshold value of the reflectivity of a visible light channel, and representing the cloud area judgment threshold value in percentage;

T_FIRexpressing the brightness temperature of the far infrared channel, and the unit is Kelvin;

T_{FIR_TCTH}the unit of the cloud area identification threshold value of the brightness and the temperature of the far infrared channel is Kelvin;

(2) water body picture element mark

If the pixel satisfies R_NIR＜R_{NIR_TWTH}And (R)_NIR﹣R_VIS) If less than 0, marking as a water body pixel; in the formula:R_NIRrepresents the near infrared channel reflectance, expressed as a percentage; r_{NIR_TWTH}Representing a water body identification threshold value of the reflectivity of the near infrared channel, and representing the water body identification threshold value by percentage;

R_VISrepresents the visible channel reflectance, expressed as a percentage;

(3) desert region pixel mark

If the land utilization type of the pixel is a desert area, marking the pixel as the desert area pixel;

(4) blazed spot area pixel mark

If the pixel satisfies S_glintMarking the image element as a flare area at an angle of less than or equal to 15 degrees; in the formula: s_glintRepresenting solar flare angle in degrees.

(5) Low temperature picture element marking

If the pixel T_MIR＜T_{MIR_TLOWTH}And is marked as a low-temperature area pixel; in the formula: t is_MIRRepresents the brightness temperature of the middle infrared channel, and has the unit of Kelvin (K);

T_{MIR_TLOWTH}the judgment threshold value of the brightness temperature low-temperature region of the middle infrared channel is represented, and the unit is Kelvin;

(6) clear sky vegetation pixel mark

If the pixel is not a cloud area, a water body, a desert area, a blazing area or a low-temperature pixel, marking as a clear sky vegetation pixel;

step 2.2, calculating background temperature:

(1) background region pixel selection

Judging suspected high-temperature pixel

If the clear sky in the neighborhood around the detection pixel indicates that the pixel meets the following conditions, the pixel is taken as a suspected high-temperature pixel:

T_MIR＞(T_{MIR_AVG}+△T_MIR) And T_M-F＞(T_{M-F_AVG}+8K), or T_MIR＞T_{MIR_WM}

In the formula: t is_{MIR_AVG}The brightness and temperature average value of clear air in 7 x 7 pixels around an infrared channel in a detection pixel is expressed in Kelvin;

T_{M-F_AVG}for detecting the clear empty finger quilt pixel in the 7 x 7 pixels around the pixelThe average value of the brightness temperature difference of the middle infrared channel and the far infrared channel is expressed in Kelvin;

in the formula: delta T_MIRIndicating a brightness temperature increment threshold value of a middle infrared channel for judging the suspected high-temperature pixel, wherein the unit is Kelvin;

T_M-Fexpressing the brightness temperature difference between the middle infrared channel and the far infrared channel, and the unit is Kelvin;

T_{MIR_WM}indicating that the brightness and temperature threshold of the middle infrared channel of the suspected high-temperature pixel is judged, wherein the unit is Kelvin;

(2) background zone mean temperature and standard deviation calculation

1) Calculating T_MIRBG，T_FIRBG，T_M-FBG

In the formula: t is_MIRBGRepresenting the average value of the brightness and the temperature of the background area of the intermediate infrared channel, and the unit is Kelvin; t is_FIRBGThe average value of the brightness and the temperature of the background area of the far infrared channel is represented, and the unit is Kelvin; t is_M-FBGThe average value of the brightness and temperature difference between the infrared channel and the far infrared channel in the background area is represented, and the unit is Kelvin; t is_FIR,iThe unit of the temperature of the ith pixel of the far infrared channel is Kelvin; t is_MIR,iThe temperature of the ith pixel brightness temperature of the middle infrared channel is represented in Kelvin;

2) calculating delta T_MIRBGAnd δ T_FIRBG

In the formula: delta T_MIRBGExpressing the standard deviation of the brightness and the temperature of the background area of the intermediate infrared channel, and the unit is Kelvin (K);

δT_FIRBGexpressing the standard deviation of the brightness and the temperature of the background area of the far infrared channel, and the unit is Kelvin (K);

3) calculating delta T_M-FBG

In the formula, delta T_M-FBGStandard deviation representing the mean value of the difference in brightness and temperature between the infrared channel and the far infrared channel in the background region, in kelvin (K);

4) correction of standard deviation

When delta T_MIRBGLess than δ T_bgminIs set to δ T_MIRBGIs δ T_bgmin；δT_FIRBGLess than δ T_bgminWhen, set delta T_FIRBGIs δ T_bgmin；δT_M-FBGLess than δ T_bgminWhen, set delta T_M-FBGIs δ T_bgmin；δT_bgminThe reference value is 2K, and when the sun zenith angle is more than 87 degrees, delta T_bgminReference value is 1.5K;

when delta T_MIRBGGreater than δ T_bgmaxWhen, set delta T_MIRBGIs δ T_bgmax；δT_FIRBGGreater than δ T_bgmaxWhen, set delta T_FIRBGIs δ T_bgmax；δT_M-FBGGreater than δ T_bgmaxWhen, set delta T_M-FBGIs δ T_bgmax，δT_bgmaxThe reference value is 3K, and when the sun zenith angle is more than 87 degrees, delta T_bgmaxReference value is 2.5K;

in the formula: delta T_bgmaxThe standard deviation upper limit of the infrared channel brightness and temperature of the background area representing the fire point identification is expressed in Kelvin (K);

δT_bgminthe lower limit of standard deviation of the infrared channel brightness and the temperature in the background area representing the fire point identification is expressed in Kelvin (K);

step 2.3, fire pixel confirmation

If one pixel meets the following conditions, the pixel is determined as a fire point pixel preliminarily;

d) polar orbit satellite: t is_MIR≥(T_MIRBGTen 4 delta T_MIRBG) And T_M-F≥(T_M-FBG+4δT_M-FBG)；

e) A stationary satellite: t is_MIR≥(T_MIRBGTen 3 delta T_MIRBG) And T_M-F≥(T_M-FBG+3δT_M-FBG)；

If the preliminarily determined fire point pixel meets one of the following cloud pollution conditions, the fire point pixel is excluded, otherwise, the fire point pixel is determined;

f)R_VIS＞(R_VISBG+ 10%) and T_MIR＜T_MIRTC；

In the formula: t is_MIRTCThe judgment threshold value of the brightness and the temperature of the infrared channel in fire point pixel cloud pollution is represented, the unit is Kelvin (K), and the initial value is 330K;

g)T_FIR＜(T_FIRBG-△T_{FIR_TCR})；

in the formula: delta T_{FIR_TCR}The method comprises the steps of representing a far infrared channel brightness and temperature identification threshold value polluted by fire point cloud, wherein the unit is Kelvin (K), and the initial value is 5K;

h)R_VIS＞R_VISBGand T_FIR＜T_FIRBGAnd T_MIR＜(T_MIRBG+6δT_MIRBG) And T_M-F＜(T_M-FBG+6δT_M-FBG)；

Step 2.4, grading the credibility of the point pixel

The credibility is divided into four grades according to the following steps:

a) when (T)_MIR–T_MIRBG) Not less than 15K and (T)_M-F–T_M-FBG) When the K is more than or equal to 15K, the image element is a first-class fire point image element, which is also called a confirmed fire point image element;

b) when (T)_MIR-T_MIRBG) < 15K or (T)_M-F–T_M-FBG) When the temperature is less than 15K, the pixel is a secondary fire point pixel, also called a suspected fire point pixel;

c) satisfying a) or b), and when cloud area pixels exist at the periphery of the fire point pixel within 2 pixels (including two pixels), the fire point pixel is a three-level fire point pixel, also called a cloud area edge fire point pixel;

d) when the 8 pixels around the fire point pixel are allIs not a fire phantom, and (T)_MIR–T_MIRBG) When the temperature is higher than 20K, the pixel is a four-stage fire point pixel, also called a noise fire point pixel;

step 2.5, fire pixel partitioning

The adjacent fire point pixels are divided into the same fire area and numbered from north to south and from west to east.

The specific man-machine interaction identification method comprises the following steps:

step 2.6, the daytime image fire point identification specifically comprises the following steps:

step 2.6.1, daytime fire monitoring multichannel image processing, including:

1) image enhancement

Performing exponential enhancement processing on the mid-infrared channel image to highlight heat source point information; performing linear enhancement processing on the near infrared and visible light channel images to highlight the surface features;

2) channel synthesis

RGB synthesis is carried out on the intermediate infrared channel image, the near infrared channel image and the visible light channel image to generate a daytime fire monitoring multichannel synthetic image;

step 2.6.2, visual fire point identification of images in daytime: in the daytime fire monitoring multichannel synthetic image, the color effect of the meteorological satellite image is as follows: the fire point is bright red; the fire passing area is dark red or black; the vegetation area without fire is green; cloud or smoke is white or grey; the water body is blue, black or dark purple;

step 2.7, night image man-machine interaction fire point identification, which specifically comprises the following steps:

step 2.7.1, night fire monitoring multichannel image processing, including:

image enhancement: performing exponential enhancement processing on the mid-infrared channel image to highlight heat source point information; respectively carrying out linear enhancement and exponential enhancement treatment on the far infrared channel and the far infrared split window channel to highlight the surface characteristics;

channel synthesis: RGB synthesis is carried out on the intermediate infrared channel, the far infrared channel and the far infrared channel split window channel to generate a night fire monitoring multichannel synthetic graph;

step 2.7.2, identifying the fire point by visual observation of the night image; in the multi-channel synthetic image for monitoring the fire condition at night, the color effect of the meteorological satellite image is as follows: the fire point is bright red; the fireless zone is dark grey.

And step 3, the fire point intensity grade identification comprises the following steps:

step 3.1, estimating the area ratio of the sub-pixel fire points and the fire point temperature, and specifically comprising the following steps:

step 3.1.1, channel data selection

If T_MIR≥T_MIRthIf the brightness and temperature of the middle infrared channel are saturated, selecting far infrared channel data for estimation; otherwise, selecting intermediate infrared and far infrared channel data and estimating by using a Newton iteration method; if the Newton iteration method is not converged, selecting mid-infrared channel data for estimation;

step 3.1.2, Dual channel data estimation

Will N_MIR、N_MIRbg、N_FIR、N_FIRbg、P₀，T₀Substituting into a Newton iteration method formula to estimate P and T;

step 3.1.3 Single channel data estimation

The middle infrared channel estimation is shown in a formula 5-9, the far infrared channel solution is shown in a formula 5-10, and T is set to be 750K in the formula;

P＝(N_MIR-N_MIRbg)/(N_MIRt-N_MIRbg)…………(5-9)

wherein the content of the first and second substances,

P＝(N_FIR-N_FIRbg)/(N_FIRt-N_FIRbg)…………(5-10)

wherein the content of the first and second substances,

step 3.2, calculating the fire point intensity

The fire point intensity is obtained by the formula (5-11):

FRP＝S_f×σT⁴……………(5-11)

in the formula, FRPThe Radiation Power (Fire Radiation Power) of the open Fire in the pixel, namely the intensity of the Fire point, and the unit is watt (W); sigma is Stefan Boltzmann constant, sigma is 5.6704 × 10^-8(W·m^-2·K^-4)；S_fIs the area of the sub-pixel fire point and has the unit of square meter (m)²)；

S_f＝P×S……………(5-12)

Wherein S is the area of the fire pixel;

and 3.3, grading the fire point intensity.

Step 4, the fire passing area identification and fire point space-time distribution statistics comprise:

step 4.1, judging and identifying the fire passing area at a single time, which comprises the following specific steps:

step 4.1.1, single-channel identification threshold

If the pixel is a non-water body pixel and the following conditions are met, the pixel is identified as a fire passing area pixel;

R_Nir＜R_{Nir_th}………(5-13)

in the formula, R_{Nir_th}A reflectivity threshold value for the near-infrared channel fire passing region identification;

in the formula: r_{Nir_th}The reference value is 10%,

step 4.1.2 NDVI identification threshold method

If the pixel is a non-water body pixel and the following conditions are met, the pixel is identified as a fire passing area pixel;

NDVI＝(R_Nir-R_Red)/(R_Nir+R_Red)……(5-15)

in the formula, R_Red(ii) reflectance in percent (%) for the visible red channel;

NDVI＜NDVI_th………(5-16)

in the formula, NDVI_thNDVI threshold value for fire passing area pixel identification; in the formula: NDVI_thThe reference value is 0;

4.2, judging and identifying the fire passing area of the two time phase images before and after the fire occurs; if the pixel is a non-water body pixel and the following conditions are met, the pixel is identified as a fire passing area pixel;

NDVI_Before-NDVI_After＞NDVI_{BA_th}………(5-17)

in the formula, NDVI_BeforePre-fire NDVI values; NDVI_AfterNDVI after fire; NDVI_{BA_th}An NDVI threshold for fire regions is identified based on the two-phase images.

Step 4, the fire passing area identification and fire point space-time distribution statistics further comprise:

step 4.3, verifying and correcting the man-machine interaction fire passing area identification information, which specifically comprises the following steps:

step 4.3.1, manually checking the fire passing area identification effect:

superposing fire passing area identification information on the fire monitoring multichannel synthetic graph, and manually checking the fire passing area identification effect;

step 4.3.2, correcting fire passing area identification through human-computer interaction:

if the fire passing area judgment information has misjudgment or missed judgment, correcting the judgment error by correcting the judgment threshold value until the manual judgment effect is met;

step 4.3.3, fire passing area estimation, which specifically comprises the following steps:

calculating the area of the pixel of the fire passing area of single satellite data:

calculating the pixel area of the fire passing area by remote sensing of the meteorological satellite and the high spatial resolution satellite: calculating the vegetation coverage of the high spatial resolution satellite:

C_G＝(NDVI_Mix-NDVI_S)/(NDVI_V-NDVI_S)……(5-19)

in the formula, NDVI_MixA single pixel NDVI value for a high spatial resolution satellite; NDVI_VA single pixel vegetation end-member value for a high spatial resolution satellite; NDVI_SA single pixel bare earth end member value for a high spatial resolution satellite;

remote sensing of meteorological satellites and high spatial resolution satellites is integrated to estimate the area of a fire passing area:

in the formula, C_iC in single pixel space range of meteorological satellite_GCalculating an average value, namely:

in the formula, C_GjThe vegetation coverage of the jth high spatial resolution satellite pixel in the coverage range of the meteorological satellite pixel;

and m is the number of high-spatial-resolution satellite pixels contained in a single pixel of the meteorological satellite.

The invention has the beneficial effects that:

according to the invention, false fire points caused by interference of a photovoltaic power generation field and a thermal power plant in the existing meteorological satellite remote sensing fire monitoring product are eliminated through means of fire point discrimination, fire point intensity discrimination, fire passing area discrimination, fire point space-time distribution statistics and the like, and the reliability of the monitoring product is improved.

On the basis of the existing satellite remote sensing fire point identification method, the satellite remote sensing fire point identification method utilizes the collected geographic information data of interference sources (photovoltaic power plants and thermal power plants) to secondarily identify the satellite remote sensing fire point based on a GIS (geographic information system) technical method (including establishing an interference source information data set, creating a buffer area and the like), and eliminates false fire points generated by the interference sources in monitoring. The invention is applied to the daily business of satellite remote sensing fire monitoring, and a multi-stage service material is manufactured, so that the reliability of a satellite remote sensing fire monitoring product is effectively improved.

The method solves the technical problems that in the process of monitoring the satellite remote sensing fire, a meteorological satellite detector is interfered by solar radiation, and the brightness temperature of some special ground objects (interference sources for short) on the underlying surface in a remote sensing image exceeds a target heat source, so that false fire points are generated, and the final satellite remote sensing fire judgment result has larger error.

Detailed Description

The method specifically comprises the following steps:

step 1: preprocessing data, the satellite original data is processed by preprocessing processes such as positioning, calibration, quality inspection and local map processing,

the specific method for preprocessing the data in the step 1 comprises the following steps:

step 1.1, scaling and positioning

The original data of the satellite is preprocessed by positioning, calibration and the like, and the visible light channel is corrected by the solar altitude angle.

Step 1.2, local area map generation

And generating a local area map of the monitoring area from the projection of the preprocessed data, wherein the size of the image can be set according to the range of the monitoring area.

Step 1.3, multichannel Synthesis of images

Step 1.3.1 daytime image

a) Channel synthesis mode: respectively endowing middle infrared, near infrared and visible light channels with red, green and blue colors for synthesis;

b) channel enhancement: each channel needs image enhancement to highlight heat source points and earth surface characteristics;

c) image color effect:

1, ignition point: bright red color

2) Vegetation area: green colour

3) Cloud or fog, white or grayish-grey

4) Water body of blue, black or dark purple

Step 1.3.2 night image

a) Channel synthesis mode: respectively endowing middle infrared, far infrared and far infrared channels with red, green and blue colors for synthesis;

b) channel enhancement: each channel needs image enhancement to highlight heat source points and earth surface characteristics;

c) image color effect:

1) fire point: bright red color

2) A fireless area: dark grey

Step 1.4 positioning accuracy

Image localization should be checked by landmarks. If the error exists, the positioning correction is needed. And the requirement of the positioning accuracy after correction is not more than 1 pixel.

Step 2: the fire monitoring mainly comprises fire point identification;

and 2, judging the fire point, namely, judging automatically and judging man-machine interaction.

Automatic identification:

the information content of the judgment and identification result comprises a satellite/sensor identification, the resolution ratio (unit is meter) of a middle infrared channel, observation time, the serial number of a fire point image element, the serial number of a fire district, longitude and latitude, names of provinces, cities and counties, land coverage types, reliability, result generation time, processing personnel and the like.

Step 2.1, generating a pixel label chart

(1) Cloud region pixel mark

If the pixel satisfies R_VIS＞R_{VIS_TCTH}And T_FIR＜T_{FIR_TCTH}And the mark is a cloud area pixel.

In the formula: r_VISThe visible light channel reflectance is expressed in percent (%).

R_{VIS_TCTH}The cloud area identification threshold value of the visible light channel reflectivity is expressed in percentage (%), and the reference value is 20%.

T_FIRIndicating the far infrared channel brightness temperature in kelvin (K).

T_{FIR_TCTH}The unit of the cloud area identification threshold value of the far infrared channel brightness temperature is Kelvin (K), and the reference value is 270K.

(2) Water body picture element mark

If the pixel satisfies R_NIR＜R_{NIR_TWTH}And (R)_NIR﹣R_VIS) If less than 0, marking as a water body pixel.

In the formula: r_NIRThe near infrared channel reflectance is expressed in percent (%).

R_{NIR_TWTH}The water body identification threshold value of the reflectivity of the near infrared channel is expressed in percentage (%), and the reference value is 10%.

R_VISThe visible light channel reflectance is expressed in percent (%).

(3) Desert region pixel mark

And if the land utilization type of the pixel is the desert area, marking the pixel as the desert area pixel.

(4) Blazed spot area pixel mark

If the pixel satisfies S_glintAnd marking the image element as a flare area at an angle of less than or equal to 15 degrees.

In the formula: s_glintDenotes the solar flare angle in degrees (degree).

(5) Low temperature picture element marking

If the pixel T_MIR＜T_{MIR_TLOWTH}And marking as a low-temperature area pixel.

In the formula: t is_MIRRepresents the mid-infrared channel luminance temperature in kelvin (K).

T_{MIR_TLOWTH}The unit of the identification threshold value of the brightness temperature low-temperature region of the middle infrared channel is Kelvin (K), and the reference value is 265K.

(6) Clear sky vegetation pixel mark

If the pixels are not cloud areas, water bodies, desert areas, blazing areas and low-temperature pixels, the pixels are marked as clear-sky vegetation pixels.

Step 2.2, calculation of background temperature

(1) Background region pixel selection

1) Judging suspected high-temperature pixel

If the clear null indicator pixel in the peripheral neighborhood of the detection pixel meets the following conditions, the clear null indicator pixel is taken as a suspected high-temperature pixel:

T_MIR＞(T_{MIR_AVG}+△T_MIR) And T_M-F＞(T_{M-F_AVG}+8K), or T_MIR＞T_{MIR_WM}

In the formula: t is_{MIR_AVG}Detecting brightness and temperature average values of pixels in 7 x 7 around an infrared channel in the pixel, wherein the unit is Kelvin (K);

T_{M-F_AVG}the clear space in 7 pixels around the detection pixel refers to the average value of the brightness and temperature differences of the infrared channel and the far infrared channel in the detected pixel, and the unit is Kelvin (K).

In the formula: delta T_MIRIndicating and judging brightness and temperature of middle infrared channel of suspected high-temperature pixelThe incremental threshold value is set to a value that is,

the unit is Kelvin (K), the reference value is 10K.

T_M-FThe brightness temperature difference between the mid-infrared channel and the far-infrared channel is expressed in kelvin (K).

T_{MIR_WM}The unit of the middle infrared channel brightness temperature threshold value for judging the suspected high-temperature pixel is Kelvin (K), and the reference value is 330K.

(2) Background zone mean temperature and standard deviation calculation

1) Calculating T_MIRBG，T_FIRBG，T_M-FBG

In the formula: t is_MIRBGAnd the average value of the brightness and the temperature of the background area of the intermediate infrared channel is expressed in Kelvin (K).

T_FIRBGAnd the unit of the brightness temperature average value of the background area of the far infrared channel is Kelvin (K).

T_M-FBGAnd the average value of the brightness temperature difference between the infrared channel and the far infrared channel in the background area is expressed in Kelvin (K).

T_FIR,iThe unit of the temperature of the ith pixel of the far infrared channel is Kelvin (K).

T_MIR,iAnd the unit of the temperature of the ith pixel brightness of the intermediate infrared channel is Kelvin (K).

2) Calculating delta T_MIRBG，δT_FIRBG

In the formula: delta T_MIRBGAnd the standard deviation of the brightness and the temperature of the background area of the intermediate infrared channel is expressed in units of Kelvin (K).

δT_FIRBGAnd the standard deviation of the brightness and the temperature of the background area of the far infrared channel is expressed in the unit of Kelvin (K).

3) Calculating deltaT_M-FBG

In the formula: delta T_M-FBGAnd the standard deviation of the mean value of the brightness temperature difference between the infrared channel and the far infrared channel in the background area is expressed in Kelvin (K).

4) Correction of standard deviation

When delta T_MIRBGLess than δ T_bgminIs set to δ T_MIRBGIs δ T_bgmin；δT_FIRBGLess than δ T_bgminWhen, set delta T_FIRBGIs δ T_bgmin；δT_M-FBGLess than δ T_bgminWhen, set delta T_M-FBGIs δ T_bgmin。δT_bgminThe reference value is 2K, and when the sun zenith angle is more than 87 degrees, delta T_bgminThe reference value is 1.5K.

When delta T_MIRBGGreater than δ T_bgmaxWhen, set delta T_MIRBGIs δ T_bgmax；δT_FIRBGGreater than δ T_bgmaxWhen, set delta T_FIRBGIs δ T_bgmax；δT_M-FBGGreater than δ T_bgmaxWhen, set delta T_M-FBGIs δ T_bgmax，δT_bgmaxThe reference value is 3K, and when the sun zenith angle is more than 87 degrees, delta T_bgmaxThe reference value was 2.5K.

In the formula: delta T_bgmaxAnd the standard deviation upper limit of the infrared channel brightness and temperature of the background area representing the fire point identification is expressed in Kelvin (K).

δT_bgminAnd the lower limit of the standard deviation of the infrared channel brightness and the temperature in the background area representing the fire point identification is expressed in Kelvin (K).

Step 2.3, fire pixel confirmation

If a pixel meets the following conditions, the pixel can be determined as a fire point pixel preliminarily:

i) polar orbit satellite: t is_MIR≥(T_MIRBGTen 4 delta T_MIRBG) And T_M-F≥(T_M-FBG+4δT_M-FBG)；

j) A stationary satellite: t is_MIR≥(T_MIRBGTen 3 delta T_MIRBG) And T_M-F≥(T_M-FBG+3δT_M-FBG)。

If the preliminarily determined fire pixel meets one of the following cloud pollution conditions, the fire pixel is excluded, otherwise, the fire pixel is determined as the fire pixel:

k)R_VIS＞(R_VISBG+ 10%) and T_MIR＜T_MIRTC；

In the formula: t is_MIRTCThe unit of the threshold value for judging the brightness and the temperature of the infrared channel in the fire point pixel cloud pollution is Kelvin (K), and the initial value is 330K.

l)T_FIR＜(T_FIRBG-△T_{FIR_TCR})；

In the formula: delta T_{FIR_TCR}The unit of the identification threshold value of the far infrared channel brightness and temperature of the fire point cloud pollution is Kelvin (K), and the initial value is 5K.

m)R_VIS＞R_VISBGAnd T_FIR＜T_FIRBGAnd T_MIR＜(T_MIRBG+6δT_MIRBG) And T_M-F＜(T_M-FBG+6δT_M-FBG)。

Step 2.4, grading the credibility of the point pixel

The credibility is divided into four grades according to the following steps:

a) when (T)_MIR–T_MIRBG) Not less than 15K and (T)_M-F–T_M-FBG) When the K is more than or equal to 15K, the image element is a first-class fire point image element, which is also called a confirmed fire point image element;

b) when (T)_MIR-T_MIRBG) < 15K or (T)_M-F–T_M-FBG) When the temperature is less than 15K, the pixel is a secondary fire point pixel, also called a suspected fire point pixel;

c) satisfying a) or b), and when cloud area pixels exist at the periphery of the fire point pixel within 2 pixels (including two pixels), the fire point pixel is a three-level fire point pixel, also called a cloud area edge fire point pixel;

d) when the 8 pixels around the fire pixel are not fire pixels, and (T)_MIR–T_MIRBG) And when the temperature is higher than 20K, the pixel is a four-stage fire point pixel, also called a noise fire point pixel.

Step 2.5, fire pixel partitioning

The adjacent fire point pixels are divided into the same fire area and numbered from north to south and from west to east.

Human-computer interaction recognition

Step 2.6, daytime image fire point identification method

Step 2.6.1, daytime fire monitoring multichannel image processing

1) Image enhancement

Performing exponential enhancement processing on the mid-infrared channel image to highlight heat source point information; and performing linear enhancement processing on the near-infrared and visible light channel images to highlight the earth surface characteristics.

2) Channel synthesis

And (4) performing RGB synthesis on the intermediate infrared, near infrared and visible light channel images to generate a daytime fire monitoring multichannel synthetic image.

Step 2.6.2, daytime image visual fire point identification

In general, in the daytime fire monitoring multichannel synthetic image, the color effect of the meteorological satellite image is as follows:

1) fire point: bright red color

2) A fire passing area: dark red or black

3) Vegetation areas without fire: green colour

4) Cloud or smoke: white or bluish grey

5) Water body: blue, black or dark purple

Step 2.7, night image man-machine interaction fire point identification method

Step 2.7.1, multi-channel image processing for monitoring fire condition at night

1) Image enhancement

Performing exponential enhancement processing on the mid-infrared channel image to highlight heat source point information; and respectively carrying out linear enhancement and exponential enhancement treatment on the far infrared channel and the far infrared split window channel to highlight the surface characteristics.

2) Channel synthesis

And performing RGB synthesis on the intermediate infrared channel, the far infrared channel and the far infrared channel split window channel to generate a night fire monitoring multichannel synthetic graph.

(2) Night image visual fire point identification method

In general, in a multi-channel synthetic image for monitoring the fire condition at night, the color effect of a meteorological satellite image is as follows:

1) fire point: bright red.

2) A fireless area: dark grey.

Step 3, judging and identifying the intensity level of the fire point; the method specifically comprises the following steps:

step 3.1, subpixel fire area ratio and fire temperature estimation method

Step 3.1.1, channel data selection

If T_MIR≥T_MIRthAnd if the brightness temperature of the middle infrared channel is saturated, selecting far infrared channel data for estimation. Otherwise, selecting middle infrared and far infrared channel (namely dual-channel) data, and estimating by using a Newton iteration method; if the Newton iteration method is not converged, selecting mid-infrared channel data for estimation; t is_MIRthSee formula (5-4):

C₁is constant, with a value of: 1.1910659X 10^-5mW/

(m²·sr·cm^-4)。

C₂Is a constant value of 1.438833K/cm^-1。

V_MIRRepresenting the mid-infrared channel center wavenumber.

N_MIRCAAnd (3) representing the radiance corresponding to the intercept of the scaling coefficient of the intermediate infrared channel, wherein the unit is as follows: mW/(m)²·sr·cm^-1)。

Step 3.1.2, two-channel data estimation method

Will N_MIR、N_MIRbg、N_FIR、N_FIRbg、P₀，T₀Substituting into Newton's iterative method formula to estimate P and T

N_MIR、N_MIRbg、N_FIR、N_FIRbgAccording to the publicationFormulas (5-5) - (5-8); p₀，T₀The dichotomy calculation is used.

Step 3.1.3 Single channel data estimation method

The mid-infrared channel estimation is shown in formulas 5-9, the far-infrared channel solution is shown in formulas 5-10, and T is set to 750K in the formulas.

P＝(N_MIR-N_MIRbg)/(N_MIRt-N_MIRbg)…………(5-9)

Wherein the content of the first and second substances,

P＝(N_FIR-N_FIRbg)/(N_FIRt-N_FIRbg)…………(5-10)

wherein the content of the first and second substances,

step 3.2, calculating the fire point intensity

The fire point intensity is obtained by the formula (5-11):

FRP＝S_f×σT⁴……………(5-11)

in the formula, FRP is the Radiation Power (Fire Radiation Power) of the open Fire in the pixel, namely the intensity of the Fire point, and the unit is watt (W); sigma-stipitisBoltzmann constant, σ 5.6704 × 10^-8(W·m^-2·K^-4)；S_fSub-pixel fire area in square meters (m)²)；N_FIRRepresents far infrared channel radiance, unit: mW (milliwatt) and/or liver/kidney

(m²·sr·cm^-1)。

N_FIRBGRepresents the background radiance of the far infrared channel, unit: mW/(m)²·sr·cm^-1)。

N_FIRtAnd (3) expressing the sub-pixel fire point radiance of the far infrared channel, unit: mW/(m)²·sr·cm^-1)。

N_MIRRepresents the mid-infrared channel radiance, unit: mW/(m)²·sr·cm^-1)。

N_MIRBGRepresents the mid-infrared channel background radiance, unit: mW/(m)²·sr·cm^-1)。

N_MIRtAnd (3) representing the radiance of the fire point of the sub-pixel of the middle infrared channel, wherein the unit is as follows: mW/(m)²·sr·cm^-1)。

P represents the area ratio of the sub-pixel fire points.

P₀And (4) representing the iteration initial value of the variable P in the Newton iteration method formula.

T represents the sub-pixel fire temperature, unit: K.

T₀and (4) representing the iteration initial value of the variable T in the Newton iteration method formula.

V_FIRRepresenting the far infrared channel center wave number.

S_f＝P×S……………(5-12)

In the formula, S is the area of the fire pixel.

Step 3.3, fire point intensity grading

The fire intensity ratings are defined in table 1.

Table 1: fire intensity rating definition

Step 4, processing procedures such as fire passing area identification, fire point space-time distribution statistics and the like are carried out to generate a satellite remote sensing fire monitoring result

Step 4, area estimation of fire passing area

Step 4.1, single-time fire passing area identification method

Step 4.1.1, single-channel identification threshold method

If the pixel is a non-water body pixel and the following conditions are met, the pixel is identified as a fire passing area pixel:

R_Nir＜R_{Nir_th}………(5-13)

in the formula, R_{Nir_th}-a reflectivity threshold for the near infrared channel fire crossing zone identification.

In the formula: r_{Nir_th}The reference value is 10%.

Step 4.1.2 NDVI identification threshold method

If the pixel is a non-water body pixel and the following conditions are met, the pixel is identified as a fire passing area pixel:

NDVI＝(R_Nir-R_Red)/(R_Nir+R_Red)……(5-15)

in the formula, R_RedThe visible red channel reflectance, expressed as a percentage (%).

NDVI＜NDVI_th………(5-16)

In the formula, NDVI_th-NDVI threshold value of picture element identification of fire passing area.

In the formula: NDVI_thThe reference value is 0,.

Step 4.2, fire area judgment method based on two time phase images before and after fire

If the pixel is a non-water body pixel and the following conditions are met, the pixel is identified as a fire passing area pixel:

NDVI_Before-NDVI_After＞NDVI_{BA_th}………(5-17)

in the formula, NDVI_Before-pre-fire NDVI values; NDVI_After-post fire NDVI values; NDVI_{BA_th}-identifying the NDVI threshold of the fire passing zone based on the two temporal images, from the vicinity of the fire passing zone and the fire passing zoneThe difference between the NDVI of two phases for areas having the same land cover type is determined.

Step 4.3, verifying and correcting the identification information of the fire passing area through human-computer interaction

Step 4.3.1, artificially inspecting the fire passing area identification effect

And superposing the fire passing area identification information on the fire monitoring multichannel synthetic graph, and manually checking the fire passing area identification effect. The visual identification method of the fire passing area is disclosed in the daytime image visual fire point identification method of the 7.1.2.1 polar orbit meteorological satellite of QX/T344.2.

Step 4.3.2, the fire passing area judgment is corrected through human-computer interaction

If the fire passing area judgment information has the condition of misjudgment or missed judgment, the judgment error is corrected by correcting the judgment threshold value until the manual judgment effect is met.

Step 4.3.2 fire passing area estimation method

2.3.4.1 method for calculating area of image element of single satellite data fire passing area

2.3.4.2 meteorological satellite and high spatial resolution satellite remote sensing integrated fire passing area pixel area meter

Calculation method

a) High spatial resolution satellite vegetation coverage calculation

C_G＝(NDVI_Mix-NDVI_S)/(NDVI_V-NDVI_S)…………(5-19)

In the formula, NDVI_Mix-high spatial resolution satellite single pixel NDVI values; NDVI_V-high spatial resolution satellite single pixel vegetation end-member values; NDVI_SHigh spatial resolution satellite single pixel bare earth end-member values.

b) Meteorological satellite and high spatial resolution satellite remote sensing comprehensive fire passing area estimation

In the formula, C_iC in single pixel space range of meteorological satellite_GCalculating an average value, namely:

in the formula, C_GjThe vegetation coverage of the jth high spatial resolution satellite pixel within the coverage range of the meteorological satellite pixel;

m is the number of high spatial resolution satellite pixels contained in a single pixel of the meteorological satellite.

And 5: and performing secondary judgment based on the fire monitoring result to obtain a satellite remote sensing fire monitoring result.

Based on the fire monitoring product carries out the secondary and judges and know, mainly includes:

step 5.1, establishing an interference source information database

An interference source information database is established based on ArcGIS software by utilizing geographic information data of photovoltaic power generation fields, thermal power plants and the like.

Step 5.2, secondary identification of remote sensing fire point

And (3) creating a buffer area of 5 kilometers of an interference source (a photovoltaic power generation field and a thermal power plant) by using a GIS technology, and judging and identifying the fire monitoring result again. If the fire point is in the buffer area, the fire point is identified as a false fire point, and if the fire point is outside the buffer area, the remote sensing fire hazard material is manufactured according to the flow.

Claims (8)

1. A method for judging fire danger by satellite remote sensing comprises the following steps:

step 1, preprocessing data, acquiring original satellite data, and performing positioning, calibration, quality inspection and local map processing;

step 2, fire point identification is carried out on the preprocessed satellite picture;

step 3, judging and identifying the intensity level of the fire point;

step 4, fire passing area identification and fire point space-time distribution statistics;

and 5, generating a satellite remote sensing fire monitoring result.

2. The method for judging the satellite remote sensing fire risk according to claim 1, characterized in that: the specific method for preprocessing the data comprises the following steps:

step 1.1, calibration and positioning: positioning and calibrating the original satellite data; after the image is positioned, the image is checked by a landmark, and if an error exists, the image is corrected by positioning; the requirement of the positioning precision after correction is not more than 1 pixel;

step 1.2, local area map processing: generating a local area map of a monitoring area from the projection of the preprocessed data, wherein the size of the image is set according to the range of the monitoring area;

and step 1.3, synthesizing the image by multiple channels.

3. The method for judging the satellite remote sensing fire risk according to claim 2, characterized in that: step 1.3 the multichannel composite image specifically includes:

step 1.3.1, multichannel synthesis of daytime images:

a) channel synthesis mode: respectively endowing middle infrared, near infrared and visible light channels with red, green and blue colors for synthesis;

b) channel enhancement: each channel is subjected to image enhancement to highlight heat source points and earth surface characteristics;

c) image color effect: the fire point is bright red; the vegetation area is green; clouds or smog are white or grey; the water body is blue, black or dark purple;

step 1.3.2, multichannel synthesis of night images:

a) channel synthesis mode: respectively endowing middle infrared, far infrared and far infrared channels with red, green and blue colors for synthesis;

b) channel enhancement: each channel is subjected to image enhancement to highlight heat source points and earth surface characteristics;

c) image color effect: the fire point is bright red; the fireless zone is dark grey.

4. The method for judging the satellite remote sensing fire risk according to claim 1, characterized in that: the fire point judging method in the step 2 comprises an automatic judging method and a human-computer interaction judging method;

the specific automatic identification method comprises the following steps:

step 2.1, generating a pixel marking map, which specifically comprises the following steps:

(1) cloud region pixel mark

If the pixel satisfies R_VIS＞R_{VIS_TCTH}And T_FIR＜T_{FIR_TCTH}Marked as cloud area pixels;

in the formula: r_VISRepresents the visible channel reflectance, expressed as a percentage;

R_{VIS_TCTH}representing a cloud area judgment threshold value of the reflectivity of a visible light channel, and representing the cloud area judgment threshold value in percentage;

T_FIRexpressing the brightness temperature of the far infrared channel, and the unit is Kelvin;

T_{FIR_TCTH}the unit of the cloud area identification threshold value of the brightness and the temperature of the far infrared channel is Kelvin;

(2) water body picture element mark

If the pixel satisfies R_NIR＜R_{NIR_TWTH}And (R)_NIR﹣R_VIS) If less than 0, marking as a water body pixel; in the formula: r_NIRRepresents the near infrared channel reflectance, expressed as a percentage; r_{NIR_TWTH}Representing a water body identification threshold value of the reflectivity of the near infrared channel, and representing the water body identification threshold value by percentage;

R_VISrepresents the visible channel reflectance, expressed as a percentage;

(3) desert region pixel mark

If the land utilization type of the pixel is a desert area, marking the pixel as the desert area pixel;

(4) blazed spot area pixel mark

If the pixel satisfies S_glintMarking the image element as a flare area at an angle of less than or equal to 15 degrees; in the formula: s_glintRepresenting solar flare angle in degrees.

(5) Low temperature picture element marking

If the pixel T_MIR＜T_{MIR_TLOWTH}And is marked as a low-temperature area pixel; in the formula: t is_MIRRepresents the brightness temperature of the middle infrared channel, and has the unit of Kelvin (K);

T_{MIR_TLOWTH}the judgment threshold value of the brightness temperature low-temperature region of the middle infrared channel is represented, and the unit is Kelvin;

(6) clear sky vegetation pixel mark

If the pixel is not a cloud area, a water body, a desert area, a blazing area or a low-temperature pixel, marking as a clear sky vegetation pixel;

step 2.2, calculating background temperature:

(1) background region pixel selection

Judging suspected high-temperature pixel

If the clear sky in the neighborhood around the detection pixel indicates that the pixel meets the following conditions, the pixel is taken as a suspected high-temperature pixel:

T_MIR＞(T_{MIR_AVG}+△T_MIR) And T_M-F＞(T_{M-F_AVG}+8K), or T_MIR＞T_{MIR_WM}

In the formula: t is_{MIR_AVG}The brightness and temperature average value of clear air in 7 x 7 pixels around an infrared channel in a detection pixel is expressed in Kelvin;

T_{M-F_AVG}the detection pixel is characterized in that clear sky in 7 × 7 pixels around the detection pixel refers to the average value of the brightness and temperature difference between an infrared channel and a far infrared channel in the detected pixel, and the unit is Kelvin;

in the formula: delta T_MIRIndicating a brightness temperature increment threshold value of a middle infrared channel for judging the suspected high-temperature pixel, wherein the unit is Kelvin;

T_M-Fexpressing the brightness temperature difference between the middle infrared channel and the far infrared channel, and the unit is Kelvin;

T_{MIR_WM}indicating that the brightness and temperature threshold of the middle infrared channel of the suspected high-temperature pixel is judged, wherein the unit is Kelvin;

(2) background zone mean temperature and standard deviation calculation

1) Calculating T_MIRBG，T_FIRBG，T_M-FBG

In the formula:T_MIRBGrepresenting the average value of the brightness and the temperature of the background area of the intermediate infrared channel, and the unit is Kelvin; t is_FIRBGThe average value of the brightness and the temperature of the background area of the far infrared channel is represented, and the unit is Kelvin; t is_M-FBGThe average value of the brightness and temperature difference between the infrared channel and the far infrared channel in the background area is represented, and the unit is Kelvin; t is_FIR,iThe unit of the temperature of the ith pixel of the far infrared channel is Kelvin; t is_MIR,iThe temperature of the ith pixel brightness temperature of the middle infrared channel is represented in Kelvin;

2) calculating delta T_MIRBGAnd δ T_FIRBG

In the formula: delta T_MIRBGExpressing the standard deviation of the brightness and the temperature of the background area of the intermediate infrared channel, and the unit is Kelvin (K);

δT_FIRBGexpressing the standard deviation of the brightness and the temperature of the background area of the far infrared channel, and the unit is Kelvin (K);

3) calculating delta T_M-FBG

In the formula, delta T_M-FBGStandard deviation representing the mean value of the difference in brightness and temperature between the infrared channel and the far infrared channel in the background region, in kelvin (K);

4) correction of standard deviation

When delta T_MIRBGLess than δ T_bgminIs set to δ T_MIRBGIs δ T_bgmin；δT_FIRBGLess than δ T_bgminWhen, set delta T_FIRBGIs δ T_bgmin；δT_M-FBGLess than δ T_bgminWhen, set delta T_M-FBGIs δ T_bgmin；δT_bgminThe reference value is 2K, and when the sun zenith angle is more than 87 degrees, delta T_bgminReference value is 1.5K;

when delta T_MIRBGGreater than deltaT_bgmaxWhen, set delta T_MIRBGIs δ T_bgmax；δT_FIRBGGreater than δ T_bgmaxWhen, set delta T_FIRBGIs δ T_bgmax；δT_M-FBGGreater than δ T_bgmaxWhen, set delta T_M-FBGIs δ T_bgmax，δT_bgmaxThe reference value is 3K, and when the sun zenith angle is more than 87 degrees, delta T_bgmaxReference value is 2.5K;

in the formula: delta T_bgmaxThe standard deviation upper limit of the infrared channel brightness and temperature of the background area representing the fire point identification is expressed in Kelvin (K);

δT_bgminthe lower limit of standard deviation of the infrared channel brightness and the temperature in the background area representing the fire point identification is expressed in Kelvin (K);

step 2.3, fire pixel confirmation

If one pixel meets the following conditions, the pixel is determined as a fire point pixel preliminarily;

a) polar orbit satellite: t is_MIR≥(T_MIRBGTen 4 delta T_MIRBG) And T_M-F≥(T_M-FBG+4δT_M-FBG)；

b) A stationary satellite: t is_MIR≥(T_MIRBGTen 3 delta T_MIRBG) And T_M-F≥(T_M-FBG+3δT_M-FBG)；

If the preliminarily determined fire point pixel meets one of the following cloud pollution conditions, the fire point pixel is excluded, otherwise, the fire point pixel is determined;

a)R_VIS＞(R_VISBG+ 10%) and T_MIR＜T_MIRTC；

In the formula: t is_MIRTCThe judgment threshold value of the brightness and the temperature of the infrared channel in fire point pixel cloud pollution is represented, the unit is Kelvin (K), and the initial value is 330K;

b)T_FIR＜(T_FIRBG-△T_{FIR_TCR})；

in the formula: delta T_{FIR_TCR}The method comprises the steps of representing a far infrared channel brightness and temperature identification threshold value polluted by fire point cloud, wherein the unit is Kelvin (K), and the initial value is 5K;

c)R_VIS＞R_VISBGand T_FIR＜T_FIRBGAnd T_MIR＜(T_MIRBG+6δT_MIRBG) And T_M-F＜(T_M-FBG+6δT_M-FBG)；

Step 2.4, grading the credibility of the point pixel

The credibility is divided into four grades according to the following steps:

a) when (T)_MIR–T_MIRBG) Not less than 15K and (T)_M-F–T_M-FBG) When the K is more than or equal to 15K, the image element is a first-class fire point image element, which is also called a confirmed fire point image element;

b) when (T)_MIR-T_MIRBG) < 15K or (T)_M-F–T_M-FBG) When the temperature is less than 15K, the pixel is a secondary fire point pixel, also called a suspected fire point pixel;

c) satisfying a) or b), and when cloud area pixels exist at the periphery of the fire point pixel within 2 pixels (including two pixels), the fire point pixel is a three-level fire point pixel, also called a cloud area edge fire point pixel;

d) when the 8 pixels around the fire pixel are not fire pixels, and (T)_MIR–T_MIRBG) When the temperature is higher than 20K, the pixel is a four-stage fire point pixel, also called a noise fire point pixel;

step 2.5, fire pixel partitioning

The adjacent fire point pixels are divided into the same fire area and numbered from north to south and from west to east.

5. The method for satellite remote sensing fire judgment according to claim 4, wherein the method comprises the following steps: the specific man-machine interaction identification method comprises the following steps:

step 2.6, the daytime image fire point identification specifically comprises the following steps:

step 2.6.1, daytime fire monitoring multichannel image processing, including:

1) image enhancement

Performing exponential enhancement processing on the mid-infrared channel image to highlight heat source point information; performing linear enhancement processing on the near infrared and visible light channel images to highlight the surface features;

2) channel synthesis

RGB synthesis is carried out on the intermediate infrared channel image, the near infrared channel image and the visible light channel image to generate a daytime fire monitoring multichannel synthetic image;

step 2.6.2, visual fire point identification of images in daytime: in the daytime fire monitoring multichannel synthetic image, the color effect of the meteorological satellite image is as follows: the fire point is bright red; the fire passing area is dark red or black; the vegetation area without fire is green; cloud or smoke is white or grey; the water body is blue, black or dark purple;

step 2.7, night image man-machine interaction fire point identification, which specifically comprises the following steps:

step 2.7.1, night fire monitoring multichannel image processing, including:

image enhancement: performing exponential enhancement processing on the mid-infrared channel image to highlight heat source point information; respectively carrying out linear enhancement and exponential enhancement treatment on the far infrared channel and the far infrared split window channel to highlight the surface characteristics;

channel synthesis: RGB synthesis is carried out on the intermediate infrared channel, the far infrared channel and the far infrared channel split window channel to generate a night fire monitoring multichannel synthetic graph;

step 2.7.2, identifying the fire point by visual observation of the night image; in the multi-channel synthetic image for monitoring the fire condition at night, the color effect of the meteorological satellite image is as follows: the fire point is bright red; the fireless zone is dark grey.

6. The method for judging the satellite remote sensing fire risk according to claim 1, characterized in that: and step 3, the fire point intensity grade identification comprises the following steps:

step 3.1, estimating the area ratio of the sub-pixel fire points and the fire point temperature, and specifically comprising the following steps:

step 3.1.1, channel data selection

If T_MIR≥T_MIRthIf the brightness and temperature of the middle infrared channel are saturated, selecting far infrared channel data for estimation; otherwise, selecting intermediate infrared and far infrared channel data and estimating by using a Newton iteration method; if the Newton iteration method is not converged, selecting mid-infrared channel data for estimation;

step 3.1.2, Dual channel data estimation

Will N_MIR、N_MIRbg、N_FIR、N_FIRbg、P₀，T₀Substituting into a Newton iteration method formula to estimate P and T;

step 3.1.3 Single channel data estimation

The middle infrared channel estimation is shown in a formula 5-9, the far infrared channel solution is shown in a formula 5-10, and T is set to be 750K in the formula;

P＝(N_MIR-N_MIRbg)/(N_MIRt-N_MIRbg)…………(5-9)

wherein the content of the first and second substances,

P＝(N_FIR-N_FIRbg)/(N_FIRt-N_FIRbg)…………(5-10)

wherein the content of the first and second substances,

step 3.2, calculating the fire point intensity

The fire point intensity is obtained by the formula (5-11):

FRP＝S_f×σT⁴……………(5-11)

in the formula, the FRP is the Fire Radiation Power (Fire Radiation Power) in the pixel, i.e. the Fire intensity, and the unit is watt (W); sigma is Stefan Boltzmann constant, sigma is 5.6704 × 10^-8(W·m^-2·K^-4)；S_fIs the area of the sub-pixel fire point and has the unit of square meter (m)²)；

S_f＝P×S……………(5-12)

Wherein S is the area of the fire pixel;

and 3.3, grading the fire point intensity.

7. The method for judging the satellite remote sensing fire risk according to claim 1, characterized in that: step 4, the fire passing area identification and fire point space-time distribution statistics comprise:

step 4.1, judging and identifying the fire passing area at a single time, which comprises the following specific steps:

step 4.1.1, single-channel identification threshold

If the pixel is a non-water body pixel and the following conditions are met, the pixel is identified as a fire passing area pixel;

R_Nir＜R_{Nir_th}………(5-13)

in the formula, R_{Nir_th}A reflectivity threshold value for the near-infrared channel fire passing region identification;

in the formula: r_{Nir_th}The reference value is 10%,

step 4.1.2 NDVI identification threshold method

If the pixel is a non-water body pixel and the following conditions are met, the pixel is identified as a fire passing area pixel;

NDVI＝(R_Nir-R_Red)/(R_Nir+R_Red)……(5-15)

in the formula, R_Red(ii) reflectance in percent (%) for the visible red channel;

NDVI＜NDVI_th………(5-16)

in the formula, NDVI_thNDVI threshold value for fire passing area pixel identification; in the formula: NDVI_thThe reference value is 0;

4.2, judging and identifying the fire passing area of the two time phase images before and after the fire occurs; if the pixel is a non-water body pixel and the following conditions are met, the pixel is identified as a fire passing area pixel;

NDVI_Before-NDVI_After＞NDVI_{BA_th}………(5-17)

in the formula, NDVI_BeforePre-fire NDVI values; NDVI_AfterNDVI after fire; NDVI_{BA_th}An NDVI threshold for fire regions is identified based on the two-phase images.

8. The method for satellite remote sensing fire judgment according to claim 7, wherein the method comprises the following steps: step 4, the fire passing area identification and fire point space-time distribution statistics further comprise:

step 4.3, verifying and correcting the man-machine interaction fire passing area identification information, which specifically comprises the following steps:

step 4.3.1, manually checking the fire passing area identification effect:

superposing fire passing area identification information on the fire monitoring multichannel synthetic graph, and manually checking the fire passing area identification effect;

step 4.3.2, correcting fire passing area identification through human-computer interaction:

if the fire passing area judgment information has misjudgment or missed judgment, correcting the judgment error by correcting the judgment threshold value until the manual judgment effect is met;

step 4.3.3, fire passing area estimation, which specifically comprises the following steps:

calculating the area of the pixel of the fire passing area of single satellite data:

calculating the pixel area of the fire passing area by remote sensing of the meteorological satellite and the high spatial resolution satellite:

calculating the vegetation coverage of the high spatial resolution satellite:

C_G＝(NDVI_Mix-NDVI_S)/(NDVI_V-NDVI_S)……(5-19)

in the formula, NDVI_MixA single pixel NDVI value for a high spatial resolution satellite; NDVI_VA single pixel vegetation end-member value for a high spatial resolution satellite; NDVI_SA single pixel bare earth end member value for a high spatial resolution satellite;

remote sensing of meteorological satellites and high spatial resolution satellites is integrated to estimate the area of a fire passing area:

in the formula, C_iC in single pixel space range of meteorological satellite_GCalculating an average value, namely:

in the formula, C_GjFor the first in the meteorological satellite pixel coverageVegetation coverage of j high spatial resolution satellite pixels;

and m is the number of high-spatial-resolution satellite pixels contained in a single pixel of the meteorological satellite.