In order to reduce the calibration uncertainty of the reflectance-based method brought by the assumption of the aerosol model, the irradiance-based method, known as improved reflectance-based method, was proposed. The irradiance-based method is described in this paper. The radiometric calibration field campaign was performed at Dunhuang test site on 27 August, 2014. A hyperspectral irradiance meter (HSIM) developed by Anhui Institute of Optics and Fine Mechanics (AIOFM) was used to measure the diffuse-to-global spectral irradiance ratio. The irradiance-based method and the reflectance-based method were performed to calibrate the first four bands of Moderate Resolution Imaging Spectroradiometer (MODIS). The results of two methods were compared with result of MODIS on-board calibrator. The comparison shows that the result of irradiance-based method has a good consistency with on-board calibration and reflectance-based method results. The difference of calibration coefficients between irradiance-based and on-board method was less than 1.4%. Due to the limitations of the irradiance-based method, a clear sky and stable atmospheric condition is required for the entire half of the calibration day to provide the data necessary for the extrapolation of diffuse-to-global ratio in viewing direction. A study on the effects of aerosol mode assumption on the final apparent reflectance was performed on both the irradiance-based method and the reflectance-based method by selecting different aerosol modes to predict the apparent reflectance. The results show that aerosol mode assumption has a great effect on the reflectance-based method, however slight effect on the irradiance-based method.
With the progress of quantitative remote sensing, the acquisition of surface BRDF becomes more and more important. In order to improve the accuracy of the surface BRDF measurements, a VNIR-SWIR Bidirectional Reflectance Automatic Measurement System, which was developed by Hefei Institutes of Physical Science (HIPS), is introduced that allows in situ measurements of hyperspectral bidirectional reflectance data. Hyperspectral bidirectional reflectance distribution function data sets taken with the BRDF automatic measurement system nominally cover the spectral range between 390 and 2390 nm in 971 bands. In July 2007, September 2008, June 2011, we acquired a series of the BRDF data covered Dunhuang radiometric calibration test site in terms of the BRDF measurement system. We have not obtained such comprehensive and accurate data as they are, since 1990s when the site was built up. These data are applied to calibration for FY-2 and other satellites sensors. Field BRDF data of a Dunhuang site surface reveal a strong spectral variability. An anisotropy factor (ANIF), defined as the ratio between the directional reflectance and nadir reflectance over the hemisphere, is introduced as a surrogate measurement for the extent of spectral BRDF effects. The ANIF data show a very high correlation with the solar zenith angle due to multiple scattering effects over a desert site. Since surface geometry, multiple scattering, and BRDF effects are related, these findings may help to derive BRDF model parameters from the in-situ BRDF measurement remotely sensed hyperspectral data sets.
Test site vicarious calibration provides an absolute radiometric calibration for sensors. Surface reflectance is a critical parameter to be measured during a vicarious calibration field campaign. In order to realize long-term high precision observations of surface spectral reflectance in solar reflective bands, Automated Self-Calibration Spectra-Radiometer (ASCSR) was developed. ASCSR measures the global irradiance and the ground reflected radiance respectively with high spectral resolution from 400nm-2400 nm, the ratio of the two measurements is the surface reflectance. The degradation influences of instrument sensors and optical elements are removed by ratio-measurements and self-calibration. In the past two years ASCSR deployed in Dunhuang test site for continuous spectral reflectance measurements over 4 weeks. The measurements result of ASCSR is compared with traditional measurements which used SVC spectra-radiometer.
In order to realize unmanned vicarious calibration, Automated Test-site Radiometer (ATR) was developed for surface reflectance measurements. ATR samples the spectrum from 400nm-1600 nm with 8 interference filters coupled with silicon and InGaAs detectors. The field of view each channel is 10 ° with parallel optical axis. One SWIR channel lies in the center and the other seven VNIR channels are on the circle of 4.8cm diameters which guarantee each channel to view nearly the same section of ground. The optical head as a whole is temperature controlled utilizing a TE cooler for greater stability and lower noise. ATR is powered by a solar panel and transmit its data through a BDS (China’s BeiDou Navigation Satellite System) terminator for long-term measurements without personnel in site. ATR deployed in Dunhuang test site with ground field about 30-cm-diameter area for multi-spectral reflectance measurements. Other instruments at the site include a Cimel sunphotometer and a diffuser-to-globe irradiance meter for atmosphere observations. The methodology for band-averaged reflectance retrieval and hyperspectral reflectance fitting process are described. Then the hyperspectral reflectance and atmospheric parameters are put into 6s code to predict TOA radiance which compare with MODIS radiance.
This paper proposes a new spatial scale conversion method, which validates moderate resolution imaging spectroradiometer (MODIS) leaf area index (LAI) product when geometry information from the MODIS 1B product and classification result is combined. The in situ LAI data, Landsat Thematic Mapper (TM), and MODIS 1B product were utilized in this research. An object-oriented method was used to classify TM imaging, where each class was computed using an empirical model to achieve LAI respectively. The 30-m TM LAI image was aggregated into the MODIS 1B product based on the geometry information of MODIS 1B. The simulated MODIS 1B image was then converted into a MODIS LAI product and compared with the simulated LAI map pixel by pixel. The results showed a lower root mean square error and higher normalization of the absolute error with the new method. In addition, the field LAI was not significantly correlated with MODIS LAI, but it did show a strong correlation with TM LAI. The new method achieved a higher correlate coefficient with the MODIS product than the conventional methods. Using this validation method based on classification and image simulation can improve the accuracy of product certification.