| As the basic parameters of the ocean carbon cycle, the distributions and stocks of the DOC and POC are essential to the ocean carbon cycle research. Most of previous estimations of DOC and POC stocks are based on the in-situ measurements, which are restricted on the space and time resolutions compared to the remote sensing techniques. Appling the remote sensing techniques in retrieving the distribution of organic carbon stocks will be much helpful to ocean carbon cycle research. The marginal sea is equal important in the ocean carbon cycle with the open ocean, as its primary production contributes about25%of the primary production in the entire ocean. Hence, there is a great need to develop remote sensing algorithms of the estimation of organic carbon stocks for the marginal seas.The previous estimations of Organic Carbon (OC) stocks by remote sensing mainly adopted the method that combine the surface distribution of OC and its vertical profile model. As the non-light-reactive matter, both DOC and POC are difficult to retrieve directly from remote sensing spectrum information. Furthermore, the remote sensing data reflect only surface information of the sea, which brings a real challenge to the estimation of organic carbon stocks in the upper layer of ocean by remote sensing techniques. In order to overcome these negative factors, there will be a great need to introduce the joint study of marine biogeochemistry and ocean color remote sensing. Until now there are some literatures discussed about the remote sensing of POC stocks in the open ocean. However, there is merely no research on the estimation of DOC stocks from satellite.In this paper, by choosing the East China Sea(ECS) as our study area, which is a typical marginal sea in the world, in this paper our purpose is to study and establish the remote sensing algorithms for the estimation of carbon stocks in the upper layer of the ECS. Except the research on the remote sensing mechanisms and algorithms of the surface distribution of POC and DOC, we mainly focused on the remote sensing models of the vertical profiles of POC and DOC, and finally propose the detailed remote sensing estimation models for the DOC and DOC stocks. In this paper these research works were done as follows:(1) The remote sensing mechanisms of surface DOC distribution. We compared and analyzed the data from estuaries and adjacent marginal seas of global large rivers (16in the top25global large rivers according to their discharge). It has been found that aCDOM behaves conservatively in most of the estuaries with salinity, while DOC usually shows a non-conservative behavior due to the influences of biology processes. Hence we discussed the mixing behavior of DOC and CDOM and its control factors in different biogeochemical conditions. The physical mixing and biology processes such as phytoplankton production can impact the distribution of DOC. On the whole, the phytoplankton production have dominant influence on the DOC/CDOM relationship in all biology processes, which should be considered and quantified in the remote sensing algorithms.(2) The remote sensing algorithms of the surface distribution of DOC for the complicate second case water in the ECS. By analyzing the in-situ data of the ECS, we proposed the remote sensing model of surface DOC concentrations which considered the phytoplankton production influence:for the coastal regions where biological influence is significant the DOC distribution can be acquired by Conservative and Biological model, and for the coastal regions where the concentration of Chla is low, conservative model is applied to get DOC distribution. The comparison of remote sensing results with the in-situ data showed that the error of DOC is about15μM on the shelf area. The error of satellite-derived DOC mainly comes from the error of CDOM product in the region of turbid water.(3) The remote sensing of surface POC concentration. Based on the analysis of the in-situ data of four seasons in the ECS, we established the POC remote sensing models in the dry seasons and wet seasons through the particulate attenuation coefficients cp. The particles in the water mainly originated from biology processes in the spring and summer. The POC concentration showed a significant relationship with cp. however, the composition of particles changed in autumn and winter, brought a big impact on the relationship between POC and cp. Hence we proposed two different remote sensing models for the two different situations. The remote sensing result of our method shows a higher accuracy compared to the MODIS standard POC product.(4) The vertical distribution model of DOC. The in-situ data of DOC suggest that the temperature and salinity structure of water is the main reason of DOC profiles. In the data of autumn and winter cruises, DOC showed a uniformed distribution by the strong influence of vertical convection. However, in spring it showed stepped appearance with three sections:upper mixing layer, down mixing layer and middle gradient layer. Based on these analysis we proposed two different DOC vertical profile models, which can be distinguished by water mass index (3calculated by surface temperature and salinity.(5) The vertical distribution model of POC. We established four main POC vertical distribution model and methods for the retrieval of model parameters according to the biogeochemical characters and control factors of POC. The four different POC vertical profiles can be classified by using the remote sensing products such as chla, surface tempature and salinity. In the four main models, Guass model could be separate into two sub models which show distinctive shapes, and retrieve for the model parameters by fitting the in-situ data respectively. Another typical model is exponential decreasing model, which could use the significant relations between POC concentraton and decreasing rate k to derive its parameters. And the parameters of fold line model exhibited to be stable in the shelf areas, easy to be derived from the in-situ data.(6) The establishment and validation of remote sensing methods for the estimation of organic carbon stocks in the ECS. Based on the previous remote sensing models of surface concentration and vertical distribution, we proposed the remote sensing methods for the estimation of DOC and POC stocks, and derived the initial results of DOC and POC stocks in the euphotic layer and all water columns of four seasons in the ECS. Compared to the reference DOC stocks (from the estimation of the in-situ data), the remote sensing DOC stocks of three seasons have a high accuracy, the average relative difference of DOC stocks in the euphotic layer is below25%, and is below16%in the all water columns. However, the remote sensing results of POC stocks have a relatively big error when compared to the reference POC stocks. The average relative difference of POC stocks range from13.8%to42%in the euphotic layer in four seasons, and range from18%～35%in the all water columns in four seasons.On the whole, in this paper we discussed the remote sensing mechanisms of surface and vertical distribution of DOC and POC in the ECS, then established the methods for the remote sensing estimation of organic carbon, and finally derived the initial remote sensing results of seasonal DOC and POC stocks. All our works in this paper have a basis contribute to the development of observation of organic carbon in the ECS by remote sensing techniques and to the ocean carbon cycle research. |
PDF Full Text Request