CN103324827B - A kind of improvement business core drives the method for bidirectional reflectance distribution function (BRDF) model hot spot - Google Patents

Abstract

The present invention relates to a kind of method that improvement business core drives bidirectional reflectance distribution function (BRDF) model hot spot.Hotspot correction is carried out by driving the volume scattering core of BRDF model RTLSR to existing MODIS business core, the new volume scattering core that one is considered focus change is proposed, the RTCLSR model that this core and former geometrical optics core carry out linear combination generation has better capability of fitting to focus, but do not affect the fitting precision of other observed ray, we with POLDER-3/BRDF database and airborne CAR data the hot spot parameters to RTCLSR model demarcate and verify, RTCLSR new model can be used for study hotspot effect with ground surface type, the change of NDVI and solar zenith angle, simultaneously, because new model maintains the linear forms of master mould very well, do not roll up the complicacy of model calculation, therefore, be MODIS? BRDF/ albedo business product up-gradation provides new algorithm basis and solution.The present invention, in Spatial Information Technology field, especially has important using value in quantitative remote sensing.

Description

A kind of improvement business core drives the method for bidirectional reflectance distribution function (BRDF) model hot spot

One, technical field

The present invention relates to a kind of method that improvement business core drives bidirectional reflectance distribution function (BRDF) model hot spot, belong to Spatial Information Technology field.

Two, background technology

Earth surface anisotropic emission is the basic macroscopic appearance of occurring in nature object to reflection of electromagnetic wave, the research of earth's surface bidirectional reflectance distribution function, is optics Quantitative Remote Sensing Studies field important component part.Since the seventies, satellite remote sensing mainly takes Vertical Observation mode to obtain earth's surface information.Along with the further investigation of mechanism of remote sensing, the development of remote sensing technology, and the various new demands faced, people more and more clearly realize that, obtaining earth's surface message context with remote sensing, multiple-angle thinking provides extra effective information source, serves irreplaceable effect.Over nearly 20 years, along with succeeding in sending up of multi-angle satellite borne sensor, earth observation have accumulated a large amount of multi-angle observation data, to effective utilization of these data of multiple angles, need dependence bidirectional reflectance distribution function (BRDF) model to a great extent, such as, the bidirectional reflect of BRDF model (RTLSR) the production world business that Moderate Imaging Spectroradiomete (MODIS) drives based on linear kernel and albedo product, and issue towards Global Subscriber, at present, quite extensive to the Application and Development of such semiempirical model, as, BRDF corrects, Surface classification, the inverting etc. of vegetation biophysical parameters and structural parameters.

Although this model has obtained applying comparatively widely, the main shortcoming of of this model underestimates hot spot-effect.Solution in early days for this problem is the expression formula being multiplied by this model with a focus core, although the correction of this method to hot spot-effect has certain effect, main Problems existing is: (1) isotropy parameter (f _iso) mainly the one of atural object optical properties express, its determines the change with angle of the height of target population volume reflection and non-targeted reflective-mode, therefore can not to this parameter application focus kernel function; (2) the geometrical optics core of the sparse reciprocity of Li Shi better features hot spot-effect, should get rid of, need not carry out secondary correction from focus core corrects; (3) model after correcting is still on the low side to the estimation of focus; (4) this bearing calibration changes the linear structure of master mould, adds computation complexity, and due to the non-linear behavior of this hotspot correction, we get rid of the comparison to this model under study for action.The replicative function that this respect recent progress also relates to based on Jupp and Strahler is theoretical, volume scattering core is corrected by having developed a hot spot factor, thus structure a new model, we are referred to as RTJLSR model, but this calibration model Problems existing is: (1) when the sun and observation at zenith direction time, isotropic scatterning parameter (f _iso) lose the physical significance as reflectivity in master mould; (2) this correction does not consider that focus may by the impact of different ground surface type, NDVI and solar zenith angle change.

The present invention is based on the focus core studied in early days and deliver, revise by checking existing volume scattering with this focus, overcome when light and sight line overlap, substance scattering nucleus does not consider the shortcoming of the two-way clearance rate correlativity of Vegetation canopy, propose and a kind ofly improve the method that business core drives bidirectional reflectance distribution function model hot spot, because focus used is endorsed with by two new argument C ₁/ C ₂adjust height and the width of focus, therefore need to demarcate in advance these two new arguments, in research, we adopt the multi-angle observation data of POLDER-3/BRDF database, by asking the method for hot spot region Local Minimum error of fitting, demarcate this two parameters, simultaneously, further study these two parameters with IGBP ground surface type, the situation of change of vegetation greenness NDVI and solar zenith angle, finally, the matching advantage of RTCLSR model in focus direction is verified further by the observation data of POLDER-3/BRDF data and airborne CAR, result shows, the present invention has stronger for multi-angle focus direction observation, capability of fitting more flexibly, its focus fitting precision improves significantly, by to two new argument C ₁/ C ₂demarcate in advance, the RTCLSR model of new correction is expected to can be used as new algorithm for the BRDF of MODIS and albedo product up-gradation, to the BRDF on inverting earth's surface, the calculating of the reflectivity especially in focus direction can be significantly improved, and embodies importance of the present invention and practical value.

Three, summary of the invention

1, object: the present invention seeks to solve MODIS existing business executing arithmetic (RTLSR algorithm) underestimates focus observation shortcoming in focus direction, the focus core delivered by utilizing early stage research, for Luo Si top layer core (RossThick, the K of existing MODIS business algorithm _rT) revise, overcome when light and sight line overlap, substance scattering nucleus does not consider the two-way clearance rate correlativity of Vegetation canopy, thus underestimates the shortcoming of focus direction observation, proposes a kind of method improving business core driving bidirectional reflectance distribution function model hot spot.The present invention, in quantitative remote sensing, especially provides correction algorithm to aspects such as utilizing BRDF model inversion Land Surface Parameters (as vegetation aggregate index), has important using value.

2, technical scheme: the present invention relates to a kind of method that improvement business core drives bidirectional reflectance distribution function (BRDF) model hot spot, concrete steps following (as Fig. 1):

Step one: the volume scattering core revising kernel-driven model

Bidirectional reflect function (BRDF) model that linear kernel drives can be expressed as:

R(θ _v，θ _s，Δφ，λ)＝f _iso(λ)+f _vol(λ)K _vol(θ _v，θ _s，Δφ)+f _geo(λ)K _geo(θ _v，θ _s，Δφ)(1)

At this, R (θ _v, θ _s, Δ φ, λ) and be bidirectional reflectance distribution function, f _iso(λ), f _vol(λ) and f _geo(λ) be three parameters of model, be called isotropy parameter, volume scattering parameter and geometrical optics parameter; K _vol(θ _v, θ _s, Δ φ) and K _geo(θ _v, θ _s, Δ φ) and be volume scattering and the geometrical optics kernel function of kernel-driven model, given vegetation structure parameter, two kernel functions are trigonometric functions of observation and position of sun; At present, the bidirectional reflect of Moderate Imaging Spectroradiomete (MODIS) business and albedo product have employed Luo Si top layer core (RossThick, K _rT) and Li Shi sparse reciprocal kernel (LiSparseR, K _lSR) the RTLSR model of linear compound, the major defect of this model is the reflectivity underestimating focus direction, and this is that Luo Si top layer kernel representation is as follows due to relativity problem when Luo Si top layer core does not consider on focus direction that light and sight line overlap:

$K_{\mathrm{RT}}=\frac{(\mathrm{\π}/2-\mathrm{\ξ})\cos \mathrm{\ξ}+\sin \mathrm{\ξ}}{{\cos \mathrm{\θ}}_v+{\cos \mathrm{\θ}}_s}-\frac{\mathrm{\π}}{4}---\left(2\right)$

Above-mentioned Luo Si top layer core formula (2) of this application is revised, and by adding a focus core, improving its capability of fitting in focus direction, constructing so-called Luo Si top layer old core (RossThickChen, K _rTC), be expressed as follows:

$K_{\mathrm{RTC}}=\frac{(\mathrm{\π}/2-\mathrm{\ξ})\cos \mathrm{\ξ}+\sin \mathrm{\ξ}}{{\cos \mathrm{\θ}}_v+{\cos \mathrm{\θ}}_s}(1+C_1{e}^{(-\mathrm{\ξ}/\mathrm{\π})C_2})-\frac{\mathrm{\π}}{2}---\left(3\right)$

Formula (3) is compared with formula (2): (i) adds focus core this focus core increases C newly ₁/ C ₂two parameters are respectively used to regulate focus height and width, and these two parameters need to demarcate in advance; (ii) constant factor-π/4 are adjusted to-pi/2, the old core in Luo Si top layer is made to move down π/4 along ordinate, this adjustment can not affect volume scattering and check multi-angle observation DATA POPULATION capability of fitting, because the capability of fitting of core is only relevant with the shape of core, and has nothing to do with the position of core; Meanwhile, this adjustment can make model isotropy parameter (f _iso) retaining its physical significance, that is, isotropy parameter can be used as the sun and the reflectance value of observation when zenith direction.

Step 2: demarcate focus nuclear parameter C ₁/ C ₂

For two newly-increased parameter C of focus core ₁/ C ₂need to demarcate in advance, due to most of polar-orbiting satellite as, MODIS, MISR etc. seldom obtain the observation data in focus direction, so we carry out model parameter demarcation with existing POLDER-3/BRDF database, this database provides comparatively complete focus observation, simultaneously, these observations carry out taxonomic revision according to existing earth's surface ecotype (IGBP), be convenient to the situation of change of study hotspot with ground surface type, revised model we be referred to as RTCLSR model, determined C can be changed according to focus with IGBP class or NDVI ₁/ C ₂parameter, for the existing BRDF of MODIS and albedo business production, timing signal we adopt hot spot region local fit error minimum principle, minimum error of fitting formula is as follows:

$\mathrm{RMSE}=\sqrt{{\frac{{\mathrm{\Σ}}_{j=1}^n(R^{\mathrm{obs}}({\mathrm{\θ}}_v,{\mathrm{\θ}}_i,\mathrm{\Δ\φ},\mathrm{\λ})-R^{\mathrm{mode}l}({\mathrm{\θ}}_v,{\mathrm{\θ}}_i,\mathrm{\Δ\φ},\mathrm{\λ}))}{n-3}}^2}---\left(4\right)$

Wherein, R ^obs(θ _v, θ _i, Δ φ, λ) and R ^model(θ _v, θ _iΔ φ, λ) be respectively the BRDF of observation and model, n is observation number, because of on the scattering direction of revised volume scattering core beyond hot spot region and substance volume scattering core have and coincide very well, so we get the calculating that the sun and the observation data of observation relative bearing ξ≤5 ° scope carry out minimum error of fitting, we are referred to as Local Minimum error of fitting method, by asking Local Minimum error of fitting, we obtain minimum C by the method for iteration ₁/ C ₂value, considers C at this ₁/ C ₂possible variation range and the complicacy of calculating, get C ₁=0.3-1.5, C ₂=0.5-1.5, obtains the C that whole POLDER-3/BRDF database is corresponding respectively ₁/ C ₂value, the C of every IGBP class (except ice and snow) in database ₁/ C ₂value, and the C of different N DVI layering ₁/ C ₂value, this analyzes the change of focus with above parameter further by contributing to;

Step 3: checking RTCLSR new model

New model checking is divided into two steps: (i) analyzes C ₁/ C ₂the situation of change of index (NDVI) and solar zenith angle (SZA) is returned-changed to parameter with ground surface type (IGBP), vegetation; (ii) comparing difference of new model (RTCLSR), master mould (RTLSR) and forefathers' revision models (RTJLSR) and observation data; Based on the C obtained above ₁/ C ₂value, we analyze focus height and width with ground surface type, the rule of NDVI and solar zenith angle change, the necessity of indirect verification parametrization focus; Meanwhile, we use the POLDER observation data of onboard flight CAR data and the different ground surface type collected, and compare new model RTCLSR, master mould RTLSR and the difference of forefathers' revision models RTJLSR in focus, carry out indirectly and directly verifying to new model.

3, advantage and effect: this invention exploits a kind of method that improvement business core drives bidirectional reflectance distribution function (BRDF) model hot spot, the focus core delivered by utilizing early stage research, Luo Si top layer core for existing MODIS business executing arithmetic revises, overcome when light and sight line overlap, substance scattering nucleus does not consider the two-way clearance rate correlativity of Vegetation canopy, thus underestimates the shortcoming of focus direction observation.The inventive method is especially adapted to sensor has the data of multiple angles of intensive sampling matching in focus direction, but parameter is after demarcating, also the matching that focus direction lacks the data of multiple angles of observation is used in, the latter will produce an accurate focus based on priori in focus direction, this invention clear logic, method is simple, strong adaptability, model after correction is expected to can be used as the upgrading of new algorithm for MODIS whole world BRDF and albedo product, to the BRDF on inverting earth's surface, especially the reflectivity improved in focus direction can be significantly improved, thus have important using value and realistic meaning to the inversion accuracy etc. improving vegetation structure (as vegetation aggregate index).

Four, accompanying drawing explanation

The schematic flow sheet of Fig. 1 the inventive method

Fig. 2 substance falls apart core K _rT(black line), the volume scattering core K that forefathers correct _rTMwork as phase angle xi ₀when=1.5 ° (green solid line), work as phase angle xi ₀when=3 ° (green dotted line), the volume scattering core K that the present invention corrects _rTCwork as C ₁=1/C ₂when=1 (red dotted line), upwards translation π/4 and C ₁=1/C ₂when=1 (red solid line), C ₁=1/C ₂when=0.6 (blue dotted line), C ₁=0.6/C ₂when=1 (pink dotted line)

Fig. 3 (a) model RTLSR, RTJLSR and RTCLSR is 6 wave band Local Minimum errors of fitting, and new RTCLSR model uses all data of POLDER database (RTCLSR_syn) and every IGBP class (RTCLSR_1GBP) to demarcate respectively

Fig. 3 (b) model RTLSR, RTJLSR and RTCLSR red wave band IGBP class Local Minimum error of fitting compares

Fig. 3 (c) model RTLSR, RTJLSR and RTCLSR near-infrared band IGBP class Local Minimum error of fitting compares

Fig. 4 principal plane airborne CAR measurement data (red point), RTLSR model (blue dotted line), RTJLSR model (green dotted line) and RTCLSR model (black solid line), now get C ₁=0.6/C ₂=0.6

Fig. 5 principal plane spaceborne POLDER data (red point), RTLSR model (blue dotted line), RTJLSR model (green dotted line) and RTCLSR model (black solid line), now get C ₁=0.4 ~ 1.0/C ₂=1.0, each subgraph marked source data, IGBP class and solar zenith angle etc.

Five, embodiment

The present invention relates to a kind of method that improvement business core drives bidirectional reflectance distribution function (BRDF) model hot spot, the method concrete steps following (as Fig. 1):

Step one: the volume scattering core revising kernel-driven model

Bidirectional reflect function (BRDF) model that linear kernel drives can be expressed as:

R(θ _v，θ _s，Δφ，λ)＝f _iso(λ)+f _vol(λ)K _vol(θ _v，θ _s，Δφ)+f _geo(λ)K _geo(θ _v，θ _s，Δφ)(1)

At this, R (θ _v, θ _s, Δ φ, λ) and be bidirectional reflectance distribution function, f _iso(λ), f _vol(λ) and f _geo(λ) be three parameters of model, be called isotropy parameter, volume scattering parameter and geometrical optics parameter; K _vol(θ _v, θ _s, Δ φ) and K _geo(θ _v, θ _s, Δ φ) and be volume scattering and the geometrical optics kernel function of kernel-driven model, given vegetation structure parameter, two kernel functions are trigonometric functions of observation and position of sun; At present, the bidirectional reflect of Moderate Imaging Spectroradiomete (MODIS) business and albedo product have employed Luo Si top layer core (RossThick, K _rT) and Li Shi sparse reciprocal kernel (LiSparseR, K _lSR) the RTLSR model of linear compound, the major defect of this model is the reflectivity underestimating focus direction, and this is that Luo Si top layer kernel representation is as follows due to relativity problem when Luo Si top layer core does not consider on focus direction that light and sight line overlap:

$K_{\mathrm{RT}}=\frac{(\mathrm{\π}/2-\mathrm{\ξ})\cos \mathrm{\ξ}+\sin \mathrm{\ξ}}{{\cos \mathrm{\θ}}_v+{\cos \mathrm{\θ}}_s}-\frac{\mathrm{\π}}{4}---\left(2\right)$

Above-mentioned Luo Si top layer core formula (2) of this application is revised, and by adding a focus core, improving its capability of fitting in focus direction, constructing so-called Luo Si top layer old core (RossThickChen, K _rTC), be expressed as follows:

$K_{\mathrm{RTC}}=\frac{(\mathrm{\π}/2-\mathrm{\ξ})\cos \mathrm{\ξ}+\sin \mathrm{\ξ}}{{\cos \mathrm{\θ}}_v+{\cos \mathrm{\θ}}_s}(1+C_1{e}^{(-\mathrm{\ξ}/\mathrm{\π})C_2})-\frac{\mathrm{\π}}{2}---\left(3\right)$

Formula (3) is compared with formula (2): (i) adds focus core this focus core increases C newly ₁/ C ₂two parameters are respectively used to regulate focus height and width, and these two parameters need to demarcate in advance; (ii) constant factor-π/4 are adjusted to-pi/2, the old core in Luo Si top layer is made to move down π/4 along ordinate, this adjustment can not affect volume scattering and check multi-angle observation DATA POPULATION capability of fitting, because the capability of fitting of core is only relevant with the shape of core, and has nothing to do with the position of core; Meanwhile, this adjustment can make model isotropy parameter (f _iso) retaining its physical significance, that is, isotropy parameter can be used as the sun and the reflectance value of observation when zenith direction.

Step 2: demarcate focus nuclear parameter C ₁/ C ₂

For two newly-increased parameter C of focus core ₁/ C ₂need to demarcate in advance, due to most of polar-orbiting satellite as, MODIS, MISR etc. seldom obtain the observation data in focus direction, so we carry out model parameter demarcation with existing POLDER-3/BRDF database, this database provides comparatively complete focus observation, simultaneously, these observations carry out taxonomic revision according to existing earth's surface ecotype (IGBP), be convenient to the situation of change of study hotspot with ground surface type, revised model we be referred to as RTCLSR model, determined C can be changed according to focus with IGBP class or NDVI ₁/ C ₂parameter, for the existing BRDF of MODIS and albedo business production, timing signal we adopt hot spot region local fit error minimum principle, minimum error of fitting formula is as follows:

$\mathrm{RMSE}=\sqrt{{\frac{{\mathrm{\Σ}}_{j=1}^n(R^{\mathrm{obs}}({\mathrm{\θ}}_v,{\mathrm{\θ}}_i,\mathrm{\Δ\φ},\mathrm{\λ})-R^{\mathrm{mode}l}({\mathrm{\θ}}_v,{\mathrm{\θ}}_i,\mathrm{\Δ\φ},\mathrm{\λ}))}{n-3}}^2}---\left(4\right)$

Wherein, R ^obs(θ _v, θ _i, Δ φ, λ) and R ^model(θ _v, θ _iΔ φ, λ) be respectively the BRDF of observation and model, n is observation number, because of on the scattering direction of revised volume scattering core beyond hot spot region and substance volume scattering core have and coincide very well, so we get the calculating that the sun and the observation data of observation relative bearing ξ≤5 ° scope carry out minimum error of fitting, we are referred to as Local Minimum error of fitting method, by asking Local Minimum error of fitting, we obtain minimum C by the method for iteration ₁/ C ₂value, considers C at this ₁/ C ₂possible variation range and the complicacy of calculating, get C ₁=0.3-1.5, C ₂=0.5-1.5, obtains the C that whole POLDER-3/BRDF database is corresponding respectively ₁/ C ₂value, the C of every IGBP class (except ice and snow) in database ₁/ C ₂value, and the C of different N DVI layering ₁/ C ₂value, this analyzes the change of focus with above parameter further by contributing to;

Step 3: checking RTCLSR new model

New model checking is divided into two steps: (i) analyzes C ₁/ C ₂parameter is with the situation of change of ground surface type (IGBP), vegetation normalization index (NDVI) and solar zenith angle (SZA); (ii) comparing difference of new model (RTCLSR), master mould (RTLSR) and forefathers' revision models (RTJLSR) and observation data; Based on the C obtained above ₁/ C ₂value, we analyze focus height and width with ground surface type, the rule of NDVI and solar zenith angle change, the necessity of indirect verification parametrization focus; Meanwhile, we use the POLDER observation data of onboard flight CAR data and the different ground surface type collected, and compare new model RTCLSR, master mould RTLSR and the difference of forefathers' revision models RTJLSR in focus, carry out indirectly and directly verifying to new model.

Embodiment 1:

Intel (R) Core is configured with at one ^tM2.5GHz4 processor, 4G internal memory, the computing machine of ATIRadeonHD5670 figure video card is implemented, for airborne CAR data of multiple angles and spaceborne POLDER data of multiple angles, adopt this method, matching is carried out to the RTCLSR model after above data acquisition corrects, adopts RTLSR model and RTJLSR models fitting to carry out direct and indirectly checking (Fig. 1) of the present invention simultaneously.

Contrast substance scattering nucleus K _rTwith the volume scattering core K of forefathers' development _rTMand the volume scattering core K that the present invention corrects _rTC(Fig. 2) can find out, the volume scattering core K that the present invention corrects _rTCby adjusting two hot spot parameters C ₁/ C ₂, more responsive and flexible to the change of focus; Relatively in the Local Minimum error of fitting (Fig. 3 (a)) of hot spot region, three models can find out that RTLSR model is maximum in the local fit error of hot spot region to spaceborne POLDER data 6 wave bands, when RTCLSR model adopts all data scaling C of POLDER-3/BRDF ₁/ C ₂during parameter, the local fit error of this model is suitable with RTJLSR model, but slightly unconspicuous improvement, but when RTCLSR model adopts every IGBP class to demarcate C ₁/ C ₂during parameter, local fit error has greatly improved relative to RTJLSR model; For every IGBP class, Fig. 3 (a) and Fig. 3 (b) compares red and near-infrared band, the otherness of three model local fit errors, can find out, the RTCLSR model that the present invention corrects has obvious improvement to most of IGBP type in the matching of hot spot region;

Contrast the performance (Fig. 4) of airborne CAR observation data and three models on principal plane, we also can clearly find out, the RTCLSR model that the present invention corrects, better the matching observation of focus; The IGBP type different to spaceborne POLDER data compares (Fig. 5) further and shows on principal plane, the RTCLSR model that the present invention corrects equally can matching POLDER focus observation data well, and former RTLSR model generally underestimates focus observation, RTJLSR model has then generally over-evaluated focus observation.

Claims (1)

1. improve the method that business core drives bidirectional reflectance distribution function (BRDF) model hot spot, step comprises:

Step one: the volume scattering core revising kernel-driven model

Bidirectional reflect function (BRDF) model that linear kernel drives can be expressed as:

R(θ _v，θ _s，Δφ，λ)＝f _iso(λ)+f _vol(λ)K _vol(θ _v，θ _s，Δφ)+f _geo(λ)K _geo(θ _v，θ _s，Δφ)(1)

At this, R (θ _v, θ _s, Δ φ, λ) and be bidirectional reflectance distribution function, f _iso(λ), f _vol(λ) and f _geo(λ) be three parameters of model, be called isotropy parameter, volume scattering parameter and geometrical optics parameter; K _vol(θ _v, θ _s, Δ φ) and K _geo(θ _v, θ _s, Δ φ) and be volume scattering and the geometrical optics kernel function of kernel-driven model, given vegetation structure parameter, two kernel functions are trigonometric functions of observation and position of sun; At present, the bidirectional reflect of Moderate Imaging Spectroradiomete (MODIS) business and albedo product have employed Luo Si top layer core (RossThick, K _rT) and Li Shi sparse reciprocal kernel (LiSparseR, K _lSR) the RTLSR model of linear compound, the major defect of this model is the reflectivity underestimating focus direction, and this is that Luo Si top layer kernel representation is as follows due to relativity problem when Luo Si top layer core does not consider on focus direction that light and sight line overlap:

$K_{RT}=\frac{(\mathrm{\π}/2-\mathrm{\ξ})\cos \mathrm{\ξ}+\sin \mathrm{\ξ}}{{\mathrm{cos\θ}}_v+{\mathrm{cos\θ}}_s}-\frac{\mathrm{\π}}{4}---\left(2\right)$

Above-mentioned Luo Si top layer core formula (2) is revised, by adding a focus core, improving its capability of fitting in focus direction, constructing so-called Luo Si top layer old core (RossThickChen, K _rTC), be expressed as follows:

$K_{RTC}=\frac{(\mathrm{\π}/2-\mathrm{\ξ})cos\mathrm{\ξ}+\sin \mathrm{\ξ}}{{\mathrm{cos\θ}}_v+{\mathrm{cos\θ}}_s}(1+C_1{e}^{\left(-\mathrm{\ξ}/\mathrm{\π}\right)C_2})-\frac{\mathrm{\π}}{2}---\left(3\right)$

Formula (3) is compared with formula (2): (i) adds focus core this focus core increases C newly ₁/ C ₂two parameters are respectively used to regulate focus height and width, and these two parameters need to demarcate in advance; (ii) constant factor-π/4 are adjusted to-pi/2, the old core in Luo Si top layer is made to move down π/4 along ordinate, this adjustment can not affect volume scattering and check multi-angle observation DATA POPULATION capability of fitting, because the capability of fitting of core is only relevant with the shape of core, and has nothing to do with the position of core; Meanwhile, this adjustment can make model isotropy parameter f _isoretain its physical significance, that is, isotropy parameter can as the sun and observation the reflectance value when zenith direction;

Step 2: demarcate focus nuclear parameter C ₁/ C ₂

For two newly-increased parameter C of focus core ₁/ C ₂need to demarcate in advance, because most of polar-orbiting satellite seldom obtains the observation data in focus direction, so we carry out model parameter demarcation with existing POLDER-3/BRDF database, this database provides comparatively complete focus observation, meanwhile, these observations carry out taxonomic revision according to existing earth's surface ecotype IGBP, are convenient to the situation of change of study hotspot with ground surface type, revised model we be referred to as RTCLSR model, determined C can be changed according to focus with IGBP class or NDVI ₁/ C ₂parameter, for the existing BRDF of MODIS and albedo business production, timing signal we adopt hot spot region local fit error minimum principle, minimum error of fitting formula is as follows:

$RMSE=\sqrt{\frac{{\Σ}_{j=1}^n{\left(R^{obs}\left({\mathrm{\θ}}_v,{\mathrm{\θ}}_i,\mathrm{\Δ}\mathrm{\φ},\mathrm{\λ}\right)-R^{\mathrm{mod}el}\left({\mathrm{\θ}}_v,{\mathrm{\θ}}_i,\mathrm{\Δ}\mathrm{\φ},\mathrm{\λ}\right)\right)}^2}{n-3}}---\left(4\right)$

Wherein, R ^obs(θ _v, θ _i, Δ φ, λ) and R ^model(θ _v, θ _iΔ φ, λ) be respectively the BRDF of observation and model, n is observation number, because of on the scattering direction of revised volume scattering core beyond hot spot region and substance volume scattering core have and coincide very well, so we get the calculating that the sun and the observation data of observation relative bearing ξ≤5 ° scope carry out minimum error of fitting, we are referred to as Local Minimum error of fitting method, by asking Local Minimum error of fitting, we obtain minimum C by the method for iteration ₁/ C ₂value, considers C at this ₁/ C ₂possible variation range and the complicacy of calculating, get C ₁=0.3-1.5, C ₂=0.5-1.5, obtains the C that whole POLDER-3/BRDF database is corresponding respectively ₁/ C ₂value, the C of all IGBP classes in database beyond ice and snow ₁/ C ₂value, and the C of different N DVI layering ₁/ C ₂value, this analyzes the change of focus with above parameter further by contributing to;

Step 3: checking RTCLSR new model

New model checking is divided into two steps: (i) analyzes C ₁/ C ₂parameter is with the situation of change of earth's surface type i GBP, vegetation normalization index NDVI and solar zenith angle SZA; (ii) comparing difference of new model RTCLSR, master mould RTLSR and forefathers' revision models RTJLSR and observation data; Based on the C obtained above ₁/ C ₂value, we analyze focus height and width with ground surface type, the rule of NDVI and solar zenith angle change, the necessity of indirect verification parametrization focus; Meanwhile, we use the POLDER observation data of onboard flight CAR data and the different ground surface type collected, and compare new model RTCLSR, master mould RTLSR and the difference of forefathers' revision models RTJLSR in focus, carry out indirectly and directly verifying to new model.