Inversion For The Hy-1a Ccd Atmospheric Correction And Water Body Components

Supported by national advanced technology research and development projects and HY-1A satellite application projects, to meet the demand of ocean color remote sensing using HY-1A CCD. the paper probes into atmospheric correction and oceanic constituents retrieval. The thesis is divided into the following parts:(1) Based on radiative transfer simulations and lookup table techniques, an atmospheric correction algorithm is established for HY-1A CCD in case 1 waters. Water-leaving reflectance of band 1 and 2, as well as other parameters, can be retrieved from TOA reflectance of band 1 to 4. Rayleigh reflectance can be approximated using Fourier series relevant to angles and wavelength. Gas absorptive transmittance can be approximated by a three order polynomial of air mass describing its angle dependency, and content related water vapor transmittance and ozone transmittance can also be approximated by a simple formula using air mass and gas content. Applying the Rayleigh lookup tables and gas transmittance formulas to HY-1A CCD data processing, the overall performance is good. Atmospheric scattering transmittance lookup tables are also built. Two kinds of aerosol lookup tables—τa ←→γ = ρmix / ρr and ρas ←→ ρA are built (taking account of absorbing aerosols), and four aerosol correction algorithms are put forward for case 1 waters. When applying them to simulated data, it turns out that the algorithm based on τa ←→ γ lookup tables is the best, and acquires good results.(2) Based on radiative transfer simulations and Artificial Neural Network (ANN) techniques, an atmospheric correction algorithm is established for HY-1A CCD in case 1 waters. Water-leaving reflectance of band 1 and 2, as well as atmospheric scattering transmittance, aerosol optical depths and other parameters, can be retrieved from TOA reflectance of band 3 and 4, as well as there angles. The ANN models are successfully applied to simulated data, and not sensitive to random added noise of ±5%. Comparison with algorithms based on lookup tables, the ANN algorithm has the highest R2 and lowest RMSE. When applying them to the data with aerosols not included in the lookup tables and ANN model construction, the performance of ANN algorithm is evidently better. Thus, it can be believed that the ANN algorithm has better interpolation ability among different aerosols. Because of its high speed of data processing, the ANN algorithm is another good choice for atmospheric correction in case 1 waters.(3) Based on radiative transfer simulations and ANN techniques, an atmospheric correction algorithm is established for HY-1A CCD in case 2 waters. Water-leaving reflectance of band 1 to 2, aerosol optical depths and other parameters can be retrieved from Rayleigh and gas absorption corrected TOA reflectance of band 1 to 4, as well as there angles. The overall performance of ANN models using simulated data is good, but comparing with the ANN algorithm for case 1 waters, the ANN algorithm for case 2 waters is more sensitive to random noise. Applying ANN models to HY-1A CCD data, the results appears reasonable except for highly turbid areas.(4) Using a three-component concentration retrieval algorithm based on genetic optimizing technique for HY-IA CCD, three component concentrations can be simultaneously retrieved from R(0') of band 1 to 4. Applying it to simulated data, the performance is quite good. SPM has the highest accuracy, then CDOM, and then chlorophyll a (relatively bad in low concentrations).It can be believed that HY-IA CCD can be used to oceanic constituents retrieval in case 2 waters, especially for SPM. The highest mean relative error is below 16% when introducing ±10% error to the input reflectance.In the end, the paper points out the future work concerning algorithm improvement.