| The case-2 waters include coastal waters, lakes, and rivers. Since the optical complexity and regional differences of the water components of different waters, we still do not have a universal remote sensing algorithm which can estimate chlorophyll a (Chla) concentration in different case-2 waters based on satellite images. It limits the application breadth and depth of remote sensing technique in the eutrophication monitoring, estimation of primary productivity, evaluation of carbon sink of case-2 waters. The lack of a universal Chla concentration estimation algorithm and a stable atmospheric correction algorithm are the two main reasons why we cannot estimation the Chla concentration of case-2 waters globally or regionally. For these two key areas,we tried to build a real universal Chla estimation algorithm and a precise atmospheric correction method, respectively. With the efforts of these two aspects, we try to make it possible to simultaneously estimate the Chla concentration of case-2 waters in the global scale. The main research content and conclusions are below.(1) We collet a total of 1121 samples from 18 case-2 waters. After analyzing the difference of the remote sensing reflectance and water inherent optical properties of different case-2 waters,a universal assumption of removing the absorption signal of yellow matter (denoted as aym(λ), the sum of the absorption of dissolved organic matte (aCDOM(λ1) and non-pigmented particulates (ad(λ1)) at about the absorption peak of Chla expressed as aym(λ1) = ηaym (λ2)+(1-η)aym(λ3) was proposed.When λ1,λ2,λ3 and η equal to 665 nm, 560 nm, 709 nm, and 0.2, respectively, the assumption estimates aym(665) well with an mean relative error (MRE) of 16.498%.In comparison, the assumption of classic three band model has a MRE of 65.168%when estimating aym(665) only using the 709 nm. Based on the assumption and a determination method of the location of near-infrared band which was used to remove the signal of the backscatter of the water, we present an universal model of case-2 waters for the estimation of Chla concentrations (denoted as UMOC). For the whole 1121 samples, the MRE between the in situ Chla concentration and the estimated Chla of UMOC is only 30.418%. Whether for the "Gulf type" of case-2 waters,or the "lake type" of case-2 waters, UMOC both get accurate and consistent Chla estimation results.Among the 18 different case-2 waters, only in the Terrebonne Bay, the MRE are great than 50%. UMOC are also applied to the simulated MERIS and Sentinel-2a data, it also gets similar well results like using the in situ remote sensing reflectance. All of these results indicate that UMOC is a model which can estimate the Chla concentration of case-2 waters in the global or regional scale.(2) In Lake Taihu, an improved land target-based iterative method (denoted as dense dark vegetation (DDV)-WC) was proposed for the atmospheric correction of MODIS and synchronous MERIS images. The improvements include the use of six aerosol models in OPAC, which fully consider the impact of relative humidity (RH) on aerosol properties, and a rigorous land dark target selection rule (DDV-WC-Selection).The inversion results of six aerosols models were compared by using the in situ AOT data that were measured by sun photometers. The aerosol model that is used in the aerosol retrieval in different months of the year is the aerosol model that has the best retrieval performance of the six aerosol models. The validation results also prove that DDV-WC-Selection method can exclude the non-DDV and mixed pixels more steadily than the DDV selection method in the MODIS standard algorithm. Whether for the MODIS image and synchronous MERIS image, DDV-WC method both get good atmospheric correction results.(3) We put the UMOC algorithm and DDV-WC algorithm together to get the Chla concentration distribution maps using the long time series MERIS RR images between 2002 and 2012 in Lake Taihu and Lake Hongze. The good correlation between the in situ Chla concentration and MERIS-retrieved Chla concentration indicates that the combination of UMOC and DDV-WC can get reliable Chla concentration estimation results when applied to the Chla estimation of case-2 waters for MERIS images. The distribution of Chla concentration in Lake Taihu and Lake Hongze both exhibit a significant spatial heterogeneity. For Lake Taihu, the Meiliang Bay, Zhushan Bay, and the north-west area of the open area are the three regions which have a high Chla concentration. For Lake Hongze, the Chla concentrations of Chengzi Bay and Lihe Bay are higher than the east part of the Lake. The Chla concentration distribution of Lake Taihu also shows significantly seasonal changes. However, the Lake Hongze does not.Between 2003 and 2011, Lake Taihu can divide to three periods. From 2003 to 2004,the mean Chla concentration of the lake is relatively low; from 2005 to 2008, the mean Chla concentration are high, and from 2009 to 2011, the mean concentration reduces.The internal variability of Lake Hognze is inconspicuous. But the Chla concentration of the Chengzi Bay in Lake Hongze rises which need attention. Whether Lake Taihu or Lake Hongze, the Chla concentrations of the water both have a significant negative correlation with the wind speed. The winds speed decreasing climate background may increase the eutrophication degree of these two lakes. |
PDF Full Text Request