Remote sensing of tropical coastal waters

Study of the Berau estuary, East Kalimantan, Indonesia

Abstract of Wiwin Ambarwulan's PhD thesis

The coastal zone is a shallow area where land, ocean and atmosphere strongly interact. It is covering approximately 7% of the surface of the global ocean. The coastal zones are increasingly attracting interest because they are the most populated and utilized areas on earth. Despite of its relatively small surface area, the coastal zone plays a considerable role in the biogeochemical cycles because virtually all land-derived materials (water, sediments, dissolved and particulate nutrients, pollutants) enter this region through surface runoff or groundwater flow. Thus, preservation of the ecological quality status of these zones including the quality of the coastal waters is a key issue of coastal zone management. Monitoring this zone requires approaches that can monitor their complex and heterogeneous processes and rapid changes. Remote sensing is one approach that can be used to monitor effectively the dynamics of the coastal zone.

Many coastal zones in Indonesia and in the world are characterized by a high concentration of suspended matter which is transported by rivers into the sea. There are many approaches and algorithms available for monitoring the Case-1 waters (open ocean). Those algorithms fail however, when applied to Case-2 waters (coastal and inland waters). Recently, many studies have been done in development of algorithms for Case-2 waters by using empirical, semiempirical, as well as analytical approaches. However, most of the algorithms were developed and validated in high-latitude coastal and inland waters, and only few have been tested in equatorial tropical regions. In order to fill this gap, this research focused on the study of ocean colour remote sensing for monitoring water quality parameters in the tropical equatorial region with frequent cloud cover.

This thesis used explorative approaches and algorithms in the monitoring of water quality in the Case-2 waters. The research was driven by field observations, MERIS satellite data analysis and application of algorithms for quantifying the water quality parameters. In order to achieve the objective, the methodology is divided into 4 steps: (1) establishing an inverse bio-optical model for estimating IOP and SIOP from in situ measurement of AOP and biogeochemical parameters and then applying the approach to the MERIS data for retrieving water quality parameters; (2) a semi-analytical approach, especially a neural network analysis and validated with in situ measurements; (3) a coupled radiative transfer model for atmospheric correction and a semianalytical approach, based on the Kubelka-Munk model for estimating TSM concentration, and (4) a spectral unmixing model combined with the inverse KM model which was applied for retrieval of TSM concentration. The methods were developed for particular conditions of the Berau estuary, but they can be applied to other Case-2 waters in Indonesia or other environments, since the research location was chosen based on its representativeness for the major phenomena in Indonesian waters. A series of field observations were done in the period of 2006 into 2008. Totally, more than 100 field locations were measured, although not all of the in situ measurements were in concordance with satellite overpasses.

The results showed that estimating water qualities parameters (e.g. TSM and Chl-a concentrations) can be done by following a simple empirical approach and by more complex semi-empirical approaches. It was concluded that a semianalytical bio-optical model via estimation of IOP and associate SIOP give a better estimation TSM and Chl-a concentration.

The existing Case-2 algorithm regional processors (e.g. C2R and FUB) and the global algorithm (MERIS L2), as well as a purely empirical approach can be used to understand the spatial distribution of water quality in a more quantitative way. However, these are only partly successful because the standard atmospheric corrections in these algorithms do not work properly under conditions of partial cloud cover and highly complex turbid waters. In this study it was found that application of a spectral unmixing algorithm helped in reducing the effects of haze variations while leaving the spectral effects of sediment variation intact. Combined with a standard atmospheric correction based on the MODTRAN radiative transfer model, the inverse semiempirical K-M model was next applied for estimating TSM concentration. This methodology proved to be quite robust, although for TSM concentrations below 50 mg l-1 the standard Case-2 algorithms are equally successful.

PhD Defence Ceremony
PhD Thesis

previous page
more news

  1. Home »
  2. Organisation »