CN101424741B - Real time extracting method for satellite remote sensing sea fog characteristic quantity - Google Patents

Abstract

The invention relates to a real-time extracting method of satellite remote sensing sea fog characteristic quantity, daytime EOS/MODIS data is used and is read into a raw data file PDS, the difference of sea fog and low cloud on a spectral characteristic is analyzed in two steps that: firstly, the sea fog and the low cloud are separated, and then, the sea fog characteristic quantity is extracted, namely that a three-level identifying method is adopted to orderly filter high cloud, middle cloud, low cloud, a cloudless water body, a solar flare water body, a cloud shadow area, a sea ice and a snow cover, the sea fog detection is established, a sea fog distributing area is obtained, a file which comprises a sea fog detecting result is generated, then, the sea fog characteristic quantity of the optical thickness of a fog area, the effective radius of fog and a liquid water path are calculated according to a formula, and pictures are displayed on a microcomputer. The method not only extracts the sea fog characteristic quantity, but also can calculate the visibility and the fog top height of the fog, consequently, the quantification detection of the sea fog is realized, and a dissipationforecast of the sea fog supplies meteorological information for air safety above the sea, maritime traffic and coastal airport operation.

Description

The real time extracting method of satellite remote sensing sea fog characteristic quantity

Technical field

The invention belongs to ocean monitoring technologytechnologies, be specifically related to a kind of real time extracting method of satellite remote sensing sea fog characteristic quantity.

Background technology

Sea fog is the common weather phenomenon in China coastal waters.When sea fog was arranged, condensation vapor produced a large amount of droplets and is suspended in air space above sea, and the scattering of droplet, reflex reduce sea surface visibility, to the human marine movable material impact that produces.

There are some researches show, the foreign literature report utilizes the passage 1,3 and 4 of the AVHRR (very high resolution radiometer) of NOAA satellite, remote sensing data inverting via satellite can obtain the characteristic quantity of the mist of certain space taking-all and low clouds (below be abbreviated as mist/low clouds): mist/low clouds cloud particle effective radius r _e, opticalthickness, aqueous water path LWP (Liquidwater path).But, prior art is not separated mist and low clouds, essence is the characteristic quantity that is not finally inversed by sea fog, and that the variation of sea fog characteristic quantity and the life of sea fog disappear is relevant, can provide important evidence for the sea fog forecast of dissipating according to the variation of sea fog optical thickness, effective radius and aqueous water content.If but sea fog and low clouds not being distinguished, strict just saying can not be extracted sea fog characteristic quantity, also just can't provide information to the dissipation forecast of sea fog.

Summary of the invention

The real time extracting method that the purpose of this invention is to provide a kind of satellite remote sensing sea fog characteristic quantity is to overcome the deficiency of prior art.

The present invention makes full use of EOS/MODIS (Moderate Resolution Imaging Spectroradiometer Moderate Imaging Spectroradiomete) satellite data, this data has been compared following characteristics and advantage with the AVHRR that the NOAA satellite carries: 1) spatial resolution improves, and the substar maximum can reach 250m (the maximum 1.1km of the substar resolution of AVHRR).2) at present MODIS carries on two EOS satellites (TERRA and AQUA), when wherein the TERRA time of passing by is local time 11:30 in the morning about, AQUA is for about local time 13:30 divides, this provides the very desirable period to inverting on daytime sea fog optical thickness.3) spectral resolution improves, and the EOS/MODIS data has 36 spectrum channels (AVHRR has 6 spectrum channels), has strengthened observation and recognition capability to earth complex situations greatly.

The concrete technical scheme of the present invention is to utilize EOS/MODIS data on daytime, read in its raw data file (PDS), and divide two big steps: one,, at first sea fog and low clouds are separated, again sea fog characteristic quantity is extracted by analyzing sea fog and the difference of low clouds on spectral signature.Promptly adopt three grades declare the knowledge method successively leach in high low clouds, clear sky water body, solar flare water body, cloud shadow region, and set up the sea fog detection method and obtain the sea fog distributive province, generate the file that comprises the sea fog testing result.Then, two, go out sea fog characteristic quantity according to the formula quantitative Analysis again---visibility and mist heights of roofs in optical thickness, droplet effective radius, aqueous water path and the mist of marine fog-zone, form the sea fog characteristic quantity data file, promptly available GRADS mapping software carries out image and shows on microcomputer.

Characteristics of the present invention: the present invention separates with low clouds sea fog, has carried out features extraction at sea fog, visibility in the mist and mist heights of roofs is calculated, thereby realized the quantification monitoring of sea fog and the information that provides is provided in the dissipation of sea fog.For air space above sea flight safety, maritime traffic transportation, harbour and coastal airport operation provide effective meteorological data.

Description of drawings

Fig. 1, the schematic flow sheet that sea fog and low clouds are separated of the present invention.

Fig. 2, the present invention extract the schematic flow sheet of sea fog characteristic quantity.

Embodiment

As Fig. 1, concrete steps of the present invention are as follows:

1) data acquisition is by polar-orbiting satellite digital visual broadcast system (DVBS), obtains and read in the PDS file.

2) from the header file of PDS document data set, read in solar zenith angle, solar azimuth, satellite zenith angle, satellite aximuth, geo-location data, calibration data parameters again, utilize conventional method to carry out data pre-service and quality control (removing striped, scale of data and location etc.) the PDS file.After pre-service and quality control, generate the HDF file; Again the HDF file is carried out geometric accurate correction with method in common, generate local LD3 such as file such as data for projection such as longitude and latitude such as grade, wherein comprised the SPECTRAL DATA information of solar zenith angle, satellite zenith angle and solar satellite relative orientation angle information and each passage.

3) carry out one-level and declare knowledge, object is tentatively divided into: cloud (the middle high cloud that comprise sea fog/low clouds, contain sea ice, snow covers information), solar flare water body, clear sky water body, cloud shade.Concrete grammar is at first to read in the LD3 file, calculates solar flare table data by the sun in the data for projection and intersatellite relative angle, reads in extra large land template data, in monthly average, 30 SST (extra large surface temperature) data simultaneously.With extra large land template data continent and ocean are separated; Consider the water body that solar flare pollutes,, bring erroneous judgement easily, with solar flare table for reference data the solar flare water body is rejected usually at visible light, near infrared and short infrared even spectral characteristic middle-infrared band and mist is very approaching.The identification of cloud (the middle high cloud that comprise sea fog/low clouds, contain sea ice, snow covers information), cloud shade all adopts the method for the bright temperature of general LONG WAVE INFRARED and visible light, near infrared albedo and medium wave infrared brightness wyntet's sign threshold value to be distinguished.Utilize normalized differential vegetation index NDVI greater than the preliminary filtering clear sky of-0.05 condition water body (comprising algae information wherein).Here pixel is defined as: satellite sensor carries out the minimum unit of scanning sample to ground scenery.

4) again the cloud in the step 3 (comprise sea fog/low clouds, contain the middle high cloud that sea ice snow covers information) is carried out secondary and declare knowledge: judge whether to satisfy sea fog/low clouds feature, to contain the mixed pixel rejecting that sea ice snow covers middle high cloud and the clear sky water body and the sea ice of information, keep sea fog/low cloud sector.Method is: according to the radiation characteristic of sea fog and clear sky water body, sea ice, middle high cloud, detect meet visible albedo between 0.12～0.15, return-snow melting index NDSI between-0.1～0.1,1.6

The albedo of μ m passage is declared and is known for containing the sea fog of low clouds greater than the bright temperature of 0.15, the 10.3 μ m passages pixel greater than 270K.Do not satisfy then and cover the middle high cloud of information and the mixed pixel of clear sky water body and sea ice for containing sea ice snow.Declare knowledge by secondary, Preliminary detection goes out to contain the sea fog of low clouds, and this sea fog (containing low clouds) pixel is labeled as Fog1.The clear sky water body pixel that obtains in step 3 and the step 4 all is labeled as CS.

5) carry out three grades for the sea fog that contains low clouds in the step 4 and declare knowledge, be about to sea fog and separate with low clouds.At first will compare obvious non-warm property low clouds on the low side with sea surface temperature and separate, its method is: for the sea fog that obtains in the step 4 (containing low clouds) pixel Fog1, look for the most contiguous clear sky water body pixel CS in 100 pixels on the same parallel of Fog1.If found clear sky water body pixel CS, then compare two the bright temperature difference Ts of pixel on thermal infrared (wavelength is 11 μ m) passage _{11 μ m}, this bright temperature difference is expressed as Δ T _{11 μ m}=| T _Fog1-T _CS|.If the bright temperature difference is within 3K, the pixel that then this is labeled as Fog1 is labeled as Fog2.If more than 3K, then declaring this pixel to know, the bright temperature difference is non-warm property low clouds (C1) pixel.

If on the same parallel of Fog1 pixel, do not find clear sky water body pixel in 100 pixels, then around the Fog1 pixel, look for the most contiguous clear sky water body pixel CS in 20 * 20 matrix of picture elements.If found clear sky water body pixel, then compare two the bright temperature difference Ts of pixel on the thermal infrared passage _{11 μ m}, Δ T _{11 μ m}=| T _Fog1-T _CS|.If the bright temperature difference is within 3K, the pixel that then this is labeled as Fog1 is labeled as Fog2.If more than 3K, then declaring this pixel to know, the bright temperature difference is non-warm property low clouds (C2) pixel.

If on the same parallel of Fog1 pixel, all do not find clear sky water body pixel in 20 * 20 matrix of picture elements in 100 pixels and on every side, then with using in monthly average, 30 SST to replace the bright temperature value of clear sky water body to seek the possible sea fog pixel of proximity.Calculate Δ T _{11 μ m}=| T _Fog1-T _CS|, if the bright temperature difference in 4K, the pixel that then this is labeled as Fog1 is labeled as Fog2.

Consider three grades declare know step and finish after, the fog-zone that obtains also may comprise the broken low clouds of warm property, for broken warm property low clouds are rejected from the fog-zone, uses the region growing method again for the pixel that is labeled as Fog2, carries out the space expansion.

Region growing method: will be labeled as the pixel that meets threshold condition among the Fog2 most and be labeled as Fog3, with Fog3 is the center, the contiguous pixel relatively and the visible albedo of Fog3 pixel, in infrared (wavelength is 3.7 μ m), thermal infrared (wavelength is 10.3 μ m) difference, if difference respectively 0.02, within 1K, the 0.5K, this contiguous pixel is as new Fog3 pixel.Circulation is thus declared and is known the fog-zone of going to sea.The condition that requires when the region growing analysis has not met, and just cuts to pieces to know to be the warm broken low clouds of property.

Consider the cloud edge filtering that will be clipped in the sea fog district, the sea fog that obtains more than further inciting somebody to action again detects data texture condition analysis.Method is: get N * N matrix, N 〉=11 are if the pixel of 60-70% flag F og3 in the matrix then declares to know and is sea fog; If have only the pixel of 30-40% to be labeled as Fog3 in the matrix, then declare knowledge and be the cloud edge; So just determined low cloud sector, sea fog district, in high cloud sector and clear sky district.

6) will detect for low cloud sector, sea fog district, in the result in high cloud sector and clear sky district be output as binary sea fog testing result file by pixel, and store.

As Fig. 2,7) on the basis that sea fog detects, calculate sea fog characteristic quantity.Read in the pixel that is labeled as mist in the sea fog testing result file, read in 1 corresponding in LD3 file passage (wavelength 0.62-0.67 μ m) albedo numerical value simultaneously, read in solar zenith angle u with pixel position, fog-zone _sData and visual contrast threshold value ε, by formula (1) pursues pixel calculating, obtains moonscope fog-zone optical thickness value τ constantly.Computing formula is as follows:

$\mathrm{\τ}=\frac{a_f{\mathrm{\μ}}_0}{(1-a_f)\mathrm{\β}\left({\mathrm{\μ}}_0\right)}---\left(1\right)$

τ is an optical thickness in the formula, a _fBe the albedo of mist, μ ₀Be the cosine (μ of solar zenith angle ₀=cos (u _s)), β (μ ₀) be backscattering coefficient, can obtain by existing meticulous multiple scattering mode computation, the MODIS satellite time basic fixed of passing by, backscattering coefficient result of calculation has been formulated to form, puts into program, for routine call).Calculating good optical thickness value outputs in the file of optical thickness value τ with binary format.

8) read in optical thickness value τ file, and calculate the aqueous water path by formula (2):

$\mathrm{LWP}={10}^{{(0.5454\·\mathrm{\τ})}^{0.254}}---\left(2\right)$

LWP is aqueous water path (gm in the formula ^-2).Output in the file of aqueous water path with binary format calculating good aqueous water path values.

9) read in fog-zone optical thickness file and aqueous water path file, can obtain the effective radius of droplet particle, formula is as follows:

$r_e=\frac{3}{2}\·\frac{\mathrm{LWP}}{\mathrm{\ρ}\·\mathrm{\τ}}---\left(3\right)$

R in the formula _eBe effective radius (μ m), LWP is aqueous water path (gm ^-2), ρ is aqueous water density (gm ^-3), τ is an optical thickness.Therefore, r _eApplication in remote sensing is meant the radius on the weighting meaning of droplet size in the survey region.Result of calculation is outputed to file r with binary format _eIn.

10) inverting of mist heights of roofs (daytime) is only relevant with optical thickness, reads in the optical thickness file, and using formula (4) calculates the mist heights of roofs:

H＝45τ ^2/3 (4)

The unit of H is a rice in the formula.Because what this experimental formula obtained is mist top geometric thickness, and sea fog occurs on the sea level, can regard the thickness of mist as the mist heights of roofs.The result is outputed in the mist heights of roofs file with binary format.

11) horizontal visibility can be liked close assorted (Koschmieder) formula by Ke and tries to achieve:

$\mathrm{VIS}=\frac{1}{{\mathrm{\β}}_{\mathrm{ext}}}\·\ln \left(\frac{1}{\mathrm{\ϵ}}\right)---\left(5\right)$

ε in the formula=0.02, β _ExtBe extinction coefficient, can be expressed as the rate of change of relative optical radiation energy in the unit distance, obtain by formula (6):

${\mathrm{\β}}_{\mathrm{ext}}=\frac{\mathrm{\Δ\τ}}{\mathrm{\ΔH}}---\left(6\right)$

ε and β _ExtSubstitution formula (5) all, then the visibility formula can be reduced to:

$\mathrm{VIS}=\frac{1}{{\mathrm{\β}}_{\mathrm{ext}}}\·\ln \left(\frac{1}{\mathrm{\ϵ}}\right)=\frac{3.19}{{\mathrm{\β}}_{\mathrm{ext}}}---\left(7\right)$

Read in optical thickness file and mist heights of roofs file, (7) formula of utilization just can calculate visibility in the mist.The visibility result is outputed in the mist in the visibility file with binary format.

12) The above results is exported with the GRADS mapping software, represented the size of numerical value with different colors or isoline.

Obviously, the present invention has effectively realized the real-time monitoring of sea fog, for air space above sea flight safety, maritime traffic transportation, harbour and coastal airport operation provide the prediction meteorological data.

Claims (3)

1. the real time extracting method of a satellite remote sensing sea fog characteristic quantity is characterized in that

Step 1 is by polar-orbiting satellite digital visual broadcast system, obtains and read in the PDS file;

Step 2 is to utilize conventional method to carry out data pre-service and quality control generation HDF file to the PDS file; Again the HDF file is carried out geometric accurate correction with method in common, generate local LD3 such as file such as projected dataset such as longitude and latitude such as grade; Step 3 is that one-level is declared knowledge, and object is tentatively divided into: the cloud of sea fog and low clouds, the middle high cloud that contains sea ice and Xue Gai information, clear sky water body, solar flare water body, cloud shade; With extra large land template data continent and ocean are separated; With solar flare table for reference data the solar flare water body is rejected; To comprise the cloud of sea fog and low clouds, the middle high cloud that contains sea ice and Xue Gai information, the differentiation of cloud shade with the bright temperature of LONG WAVE INFRARED and visible light, near infrared albedo and medium wave infrared brightness wyntet's sign threshold value, utilize the pixel of normalized differential vegetation index NDVI again greater than-0.05 condition, preliminary filtering comprises the clear sky water body of algae information, has finished one-level and has declared knowledge;

Step 4 be to above-mentioned one-level declare the cloud that comprises sea fog and low clouds known among the result, the middle high cloud that contains sea ice and Xue Gai information carries out secondary and declares knowledge, judge whether to satisfy sea fog and low clouds feature: be about to contain the mixed pixel rejecting that sea ice snow covers middle high cloud and the clear sky water body and the sea ice of information, keep sea fog and low cloud sector; Method is: according to the radiation characteristic of sea fog and clear sky water body, sea ice, middle high cloud, detection meet visible albedo between 0.12～0.15, normalizing snow melting index NDSI between-0.1～0.1, the albedo of 1.6 μ m passages greater than the pixel of 270K, the normalized differential vegetation index NDVI pixel greater than-0.05 condition, is judged as the sea fog that contains low clouds greater than the bright temperature of 0.15,10.3 μ m passages; Not satisfying then is middle high cloud; Get final product Preliminary detection or identify the sea fog that contains low clouds, this sea fog pixel is labeled as Fog1; Clear sky water body pixel all is labeled as CS, has finished secondary and has declared knowledge;

Step 5 is to declare the sea fog that contains low clouds in the knowledge for secondary to carry out three grades and declare knowledge, being about to sea fog separates with low clouds: at first will comparing obviously with sea surface temperature, non-warm property low clouds on the low side separate, the fog-zone that obtains is labeled as Fog2, consider that Fog2 also may comprise the broken low clouds of warm property, for the broken low clouds of warming up property are rejected from the fog-zone, use the region growing method again for the pixel that is labeled as Fog2, carry out the space and expand, the broken low clouds of warming up property are rejected from the fog-zone, and it is Fog3 that the sea fog that obtains detects data markers;

To be clipped in the cloud edge filtering in the sea fog district then, promptly further will more than the sea fog that obtains detect data texture condition analysis, method is: get N * N matrix, N 〉=11 are if the pixel of 60-70% flag F og3 in the matrix declares then that to know be sea fog; If have only the pixel of 30-40% to be labeled as Fog3 in the matrix, then declare knowledge and be the cloud edge; So just determined low cloud sector, sea fog district, in high cloud sector and clear sky district;

Step 6 is that the result is output as binary sea fog testing result file by pixel, and stores and demonstration;

Step 7 is on the basis that sea fog detects, calculate sea fog characteristic quantity: read in the pixel that is labeled as the fog-zone in the sea fog testing result file, reading in wavelength corresponding with pixel position, fog-zone in the LD3 file simultaneously is the 1 passage albedo numerical value of 0.62-0.67 μ m, reads in solar zenith angle u _sData and visual contrast threshold value ε, by formula (1) pursues pixel calculating, obtains moonscope fog-zone optical thickness value τ constantly, and computing formula is as follows:

$\mathrm{\τ}=\frac{a_f{\mathrm{\μ}}_0}{(1-a_f)\mathrm{\β}\left({\mathrm{\μ}}_0\right)}---\left(1\right)$

τ is an optical thickness in the formula, a _fBe the albedo of mist, μ ₀Be the cosine of solar zenith angle, i.e. μ ₀=cos (u _s), β (μ ₀) be backscattering coefficient, can obtain by existing meticulous multiple scattering mode computation, the MODIS satellite time basic fixed of passing by, backscattering coefficient result of calculation has been formulated to form, put into program, for routine call, calculate good optical thickness value and output in the file of optical thickness value τ with binary format;

Step 8 is to read in optical thickness value τ file, and calculates the aqueous water path by formula (2):

$\mathrm{LWP}={10}^{{(0.5454\·\mathrm{\τ})}^{0.254}}---\left(2\right)$

LWP is the aqueous water path in the formula, and its unit is gm ^-2, output in the file of aqueous water path with binary format calculating good aqueous water path values;

Step 9 is to read in fog-zone optical thickness file and aqueous water path file, calculates the effective radius of droplet particle, and formula is as follows:

$r_e=\frac{3}{2}\·\frac{\mathrm{LWP}}{\mathrm{\ρ}\·\mathrm{\τ}}---\left(3\right)$

R in the formula _eBe effective radius, its unit is μ m, and LWP is the aqueous water path, and its unit is gm ^-2, ρ is an aqueous water density, its unit is gm ^-3, τ is an optical thickness, therefore, and r _eApplication in remote sensing is meant the radius on the weighting meaning of droplet size in the survey region, and result of calculation is outputed to file r with binary format _eIn;

Step 10 is to calculate the mist heights of roofs, and daytime, the inverting of mist heights of roofs was only relevant with optical thickness, read in the optical thickness file, and using formula (4) calculates the mist heights of roofs:

H＝45τ ^2/3 (4)

The unit of H is a rice in the formula, and the result is outputed in the mist heights of roofs file with binary format;

Step 11 is a calculated level visibility, and horizontal visibility is liked close assorted (Koschmieder) formula by Ke and tried to achieve:

$\mathrm{VIS}=\frac{1}{{\mathrm{\β}}_{\mathrm{ext}}}\·\ln \left(\frac{1}{\mathrm{\ϵ}}\right)---\left(5\right)$

ε in the formula=0.02, β _ExtBe extinction coefficient, be expressed as the rate of change of relative optical radiation energy in the unit distance, obtain by formula (6):

${\mathrm{\β}}_{\mathrm{ext}}=\frac{\mathrm{\Δ\τ}}{\mathrm{\ΔH}}---\left(6\right)$

ε and β _ExtSubstitution formula (5) all, then the visibility formula can be reduced to:

$\mathrm{VIS}=\frac{1}{{\mathrm{\β}}_{\mathrm{ext}}}\·\ln \left(\frac{1}{\mathrm{\ϵ}}\right)=\frac{3.19}{{\mathrm{\β}}_{\mathrm{ext}}}---\left(7\right)$

Read in optical thickness file and mist heights of roofs file, utilize (7) formula just can calculate visibility in the mist, the visibility result is outputed in the mist in the visibility file with binary format; The above results is exported with the GRADS mapping software.

2. the real time extracting method of satellite remote sensing sea fog characteristic quantity as claimed in claim 1, it is characterized in that the method that non-warm property low clouds separate is the sea fog pixel Fog1 that obtains containing low clouds for above-mentioned, is to look for the most contiguous clear sky water body pixel CS on the same parallel of Fog1 in 100 pixels; If found clear sky water body pixel CS, the bright temperature difference T on two pixels thermal infrared passage that is 11 μ m at wavelength relatively then _{11 μ m}, this bright temperature difference is expressed as Δ T _{11 μ m}=| T _Fog1-T _CS|, if the bright temperature difference within 3K, the pixel that then this is labeled as Fog1 is labeled as Fog2; If more than 3K, then declaring this pixel to know, the bright temperature difference is non-warm property low clouds (C1) pixel;

If on the same parallel of Fog1 pixel, do not find clear sky water body pixel in 100 pixels, then around the Fog1 pixel, look for the most contiguous clear sky water body pixel CS in 20 * 20 matrix of picture elements; If found clear sky water body pixel, then compare two the bright temperature difference Ts of pixel on above-mentioned thermal infrared passage _{11 μ m}, Δ T _{11 μ m}=| T _Fog1-T _CS|; If the bright temperature difference is within 3K, the pixel that then this is labeled as Fog1 is labeled as Fog2; If more than 3K, then declaring this pixel to know, the bright temperature difference is non-warm property low clouds (C2) pixel;

If all do not find clear sky water body pixel in 20 * 20 matrix of picture elements in 100 pixels and on every side on the same parallel of Fog1 pixel, then weather sea surface temperature value replaces the bright temperature value of clear sky water body to seek the possible sea fog pixel of proximity with using for many years, and calculates Δ T _{11 μ m}=| T _Fog1-T _CS|, if the bright temperature difference in 4K, the pixel that then this is labeled as Fog1 is labeled as Fog2.

3. the real time extracting method of satellite remote sensing sea fog characteristic quantity as claimed in claim 1, it is characterized in that the region growing method is to be labeled as Fog3 with being labeled as the pixel that meets threshold condition among the Fog2 most, with Fog3 is the center, the contiguous pixel relatively and albedo, the wavelength of Fog3 pixel be 3.7 μ m in infrared, wavelength be the thermal infrared difference of 11 μ m, if difference respectively 0.02, within 1K, the 0.5K, this contiguous pixel is as new Fog3 pixel, circulation is thus declared and is known the fog-zone of going to sea; When the condition that does not meet region growing analysis requirement, just declare and know for warming up the broken low clouds of property.

Images