Final Activity Report Summary - PEIPSIREM (Satellite based remote sensing of aquatic environment in Estonia by the example of Lake Peipsi)
The project was developed to meet the increasing need for cost-effective ways to elaborate the present monitoring of water resources in a new member state of the European Union? Estonia. This requires more frequent and comprehensive spatial and temporal data coverage than provided by common sampling. Satellite-based remote sensing offers the best solution to this problem. Coastal and lake waters are especially sensitive to anthropogenic activities. Because of complex optical properties of these waters, the use of satellite data is not yet operational for their monitoring.
Processing of satellite images includes the application of atmospheric correction and suitable algorithms to derive water quality parameters, such as chlorophyll concentration, suspended matter and dissolved organic matter concentration in the water. The project is a follow-up to a study carried out during the original EU Marie Curie individual fellowship in Sweden. Now, the research involves all three largest lakes in the EU? L. Peipsi, Vänern and Vättern, located east and west of the Baltic Proper. Regarding the optical properties of the water the investigation extended from clear and oligotrophic lake water to turbid and humic coastal, and lake waters. In addition to the MODIS and SeaWiFS data MERIS/Envisat data was available for the project.
Main scientific objectives and results:
1. Validation and comparison of standard remote sensing products. We have shown that the SeaWiFS and MODIS standard chlorophyll products do not work in the Baltic Sea, where the chlorophyll estimation error is around 200% in best cases. In lakes the standard products are often just masked out. Despite the complex optical conditions over the Baltic Sea region, the SeaWiFS, MODIS and MERIS results are similar at the earlier stage of a cyanobacterial bloom. Large and uncertain discrepancies occur between the satellite data during strong blooms, which are common in the Baltic Sea and L. Peipsi. Validation of the MODIS Sea Surface Temperature product over large European lakes showed very good (R2>0.97) accordance with measured data.
2. Optimising atmospheric correction. We have shown that the failure of atmospheric correction over the Baltic Sea and inland lakes is caused by two factors: a) poorly defined aerosol optical properties over the Baltic Sea region and b) high water-leaving radiance in the near-infrared bands above turbid waters. A radiative transfer code for the coupled atmosphere-ocean system was tested on SeaWiFS data and this tends to produce better results over clear lakes than the standard method. Comparing aerosol optical thickness data measured on either side of the Baltic Sea close to the investigated lakes, both in Sweden and in Estonia, it has been shown that in most cases there is a very good agreement between satellite-derived and measured aerosol optical thickness over the whole region. Therefore, the same models can be applied over the whole area of investigation.
2. Developing algorithms for water quality estimates. It has been demonstrated that turning satellite remote sensing algorithms to the water type greatly improves product retrieval. Most large European lakes belong to the Clear water type. L. Peipsi, however, differs from these as its waters are mostly classified as Moderate type, indicating that they contain a mixture of different substances. However, Moderate and Turbid water types cannot be distinguished by applying the MODIS standard processing scheme because it lacks of suitable bands. For optically complex conditions (strong horizontal variability, cyanobacterial blooms, etc), a hyperspectral sensor such as Hyperion is most useful, but the MODIS high resolution bands may also be suitable. However, only the MERIS sensor has a suitable set of spectral bands to estimate water quality in optically complex waters.
Processing of satellite images includes the application of atmospheric correction and suitable algorithms to derive water quality parameters, such as chlorophyll concentration, suspended matter and dissolved organic matter concentration in the water. The project is a follow-up to a study carried out during the original EU Marie Curie individual fellowship in Sweden. Now, the research involves all three largest lakes in the EU? L. Peipsi, Vänern and Vättern, located east and west of the Baltic Proper. Regarding the optical properties of the water the investigation extended from clear and oligotrophic lake water to turbid and humic coastal, and lake waters. In addition to the MODIS and SeaWiFS data MERIS/Envisat data was available for the project.
Main scientific objectives and results:
1. Validation and comparison of standard remote sensing products. We have shown that the SeaWiFS and MODIS standard chlorophyll products do not work in the Baltic Sea, where the chlorophyll estimation error is around 200% in best cases. In lakes the standard products are often just masked out. Despite the complex optical conditions over the Baltic Sea region, the SeaWiFS, MODIS and MERIS results are similar at the earlier stage of a cyanobacterial bloom. Large and uncertain discrepancies occur between the satellite data during strong blooms, which are common in the Baltic Sea and L. Peipsi. Validation of the MODIS Sea Surface Temperature product over large European lakes showed very good (R2>0.97) accordance with measured data.
2. Optimising atmospheric correction. We have shown that the failure of atmospheric correction over the Baltic Sea and inland lakes is caused by two factors: a) poorly defined aerosol optical properties over the Baltic Sea region and b) high water-leaving radiance in the near-infrared bands above turbid waters. A radiative transfer code for the coupled atmosphere-ocean system was tested on SeaWiFS data and this tends to produce better results over clear lakes than the standard method. Comparing aerosol optical thickness data measured on either side of the Baltic Sea close to the investigated lakes, both in Sweden and in Estonia, it has been shown that in most cases there is a very good agreement between satellite-derived and measured aerosol optical thickness over the whole region. Therefore, the same models can be applied over the whole area of investigation.
2. Developing algorithms for water quality estimates. It has been demonstrated that turning satellite remote sensing algorithms to the water type greatly improves product retrieval. Most large European lakes belong to the Clear water type. L. Peipsi, however, differs from these as its waters are mostly classified as Moderate type, indicating that they contain a mixture of different substances. However, Moderate and Turbid water types cannot be distinguished by applying the MODIS standard processing scheme because it lacks of suitable bands. For optically complex conditions (strong horizontal variability, cyanobacterial blooms, etc), a hyperspectral sensor such as Hyperion is most useful, but the MODIS high resolution bands may also be suitable. However, only the MERIS sensor has a suitable set of spectral bands to estimate water quality in optically complex waters.