Angles Normalization Model Of Vegetation Canopy Reflectance Based On Geostationary Satellite Remote Sensing Data

With the development of band design and spatial resolution of VNIR(Visible-Near Infrared,400-900nm)multi-channel scanning radiometer on geostationary satellite,the geostationary satellite have not only been used to include the traditional application such as meteorology,communications and broadcasting but also been applied to the monitoring of ocean water color and land.In particular,a high frequency imaging characteristic allows the geostationary satellite(GSS)to capture the diurnal variation of vegetation canopy spectra,which has a very important practical significance for the monitoring of vegetation through the remote sensing(RS)approach.Moreover,this can also greatly improve the efficiency of multi-spectral RS data acquisition in cloudy and rainy areas.There is a problem that the high frequency GSS RS data usually have large differences in observation angle and solar angle,and the differences in bidirectional reflectance characteristic of among vegetation canopy spectra are significant,which make it necessary to process angles normalization on GSS RS data.Whereas,the 16day BRDF(bidirectional reflectance distribution function)products and NBAR(Nadir BRDF adjusted Reflectance)products of MODIS(MODerate resolution Imaging Spectroradiometer)and VIIRS(Visible Infrared Imaging Radiometer Suite)are commonly used,the angle normalization algorithm(in used,kernel-driven model)requires the use of 16 days period of cloud free angular surface reflectances that adequately sample(7 at least)the viewing geometry for the model parameters solution.The kernel-driven model exists time scale applicability issues when used for the angle normalization of GSS RS data,so the research on the angle normalization of GSS RS data is still in urgent.In order to resolve this problem,this paper established an angles normalization model for remote sensing vegetation canopy reflectance based on GOCI imaging characteristic that each pixel has fixed observation angle.GOCI images(Geostationary,Ocean Color Imager)were obtained from GSS COMS(Communication?Ocean and Meteorological Satellite).The established model referenced topographic correction method for vegetation canopies based on path length correction,solar zenith angle normalization model and Minnaert model.And it also considered the characteristic of diurnal variations in vegetation canopy spectra.The established model can provide important reference value for the normalization of Gao Fen-4 satellite data.Specifically,the paper firstly confirmed the selection of bands and time window for GOCI data analysis based on the diurnal variation characteristic of winter wheat canopy spectra,and analyzed the effects of angles on spectral characteristic parameters and vegetation indices.Secondly,the paper completed the topographic correction,solar angle normalization and observation angle normalization according to the definition of ground object reflectance normalization;all the processes integrally established the model of vegetation canopy reflectance angles normalization for geostationary satellite remote sensing data.Finally,experiments were carried out on field-measured vegetation canopy spectra with multi-observed angles and GOCI images obtained on22th Apr,2015 to validate the performance of the proposed angles normalization model.Results showed that the angles normalization model achieved the desired normalization effect on GOCI images.The main conclusions of this paper are as follows:(1)Diurnal variations in vegetation canopy spectra of winter wheat in jointing stage were revealed,solar and detect angles effects on vegetation canopy spectra were uncovered.The integral reflectance in the VNIR(S_VNIR)computed from field-measured data showed obvious fluctuations and the double-peak characteristic.Reflectance in the afternoon(13:30-15:00)was higher than that in the morning(9:30-11:00).And the band reflectance variance of the field-measured data and simulated data had curves that were shaped similarly to the vegetation spectra.The RHL of reflectance in range 400-730nm were significantly larger than these in range 730-900nm.The polarization diagram of S_VNIR presented irregular bowl-shaped distribution affected by detect angle,and spectral characteristic parameters and vegetation indices usually appeared high value or low value at hot-spot position.(2)Vegetation canopy reflectance angles normalization model(ASNM)was established based on geostationary satellite data.In order to take advantage of the high frequency and large width imaging characteristic of a geostationary satellite RS data,the ASNM includes three core steps:Firstly considering the geotropism of vegetation,ASNM introduced the path length correction method to process topographic correction for vegetation canopies.Secondly,on the base of ground condition that slope of each pixel had been corrected to horizontal plane,the paper resorted to cosine of solar zenith angle to carry out the solar angle normalization,and solar incidence angle of each pixel had been normalized to the local zenith direction.Finally the paper referenced the Minnaert model to eliminate the bidirectional reflection differences caused by the changes in solar incidence angles.During this process,the time window(before 11:00)was used to retain the diurnal variations characteristics of vegetation canopy spectral in high frequency RS data.(3)Validity of the established angles normalization model was verified by GOCI images processing.The ASNM has a better normalization effect on GOCI image ground reflectance than PLC model and the single solar zenith angle normalization model.The coefficient of determination(²)for the linear relationship between band reflectance of each GOCI images in morning time window have been significantly improved,with the slope of linear fitting equation all close to 1.On the contrary,the slope of linear fitting equation between band reflectance of each GOCI images in double-peak time window was significantly declined.Furthermore,the comparisons of the results after topographic correction or angle(s)normalization showed that:ASNM has a better normalization effect on both solar incidence angle and slope rather than solar zenith angle,observation zenith angle and the angle from sensor observation direction to the ground surface normal.The innovations of this paper are:(1)Minnaert model was referenced to eliminate the bidirectional reflection differences caused by changes of solar incidence angle,and revealed solution of Minnaert model coefficient k in pixel by pixel mode with data filter.(2)Vegetation canopy reflectance angles normalization model was established with the help of the geostationary satellite sensors characteristic that high frequency imaging and each pixel have fixed detect angle.The ASNM is able to improve the time resolution of angles normalized products for RS vegetation canopy reflectance to hourly level.