Effects of MERIS L1b radiometric calibration on regional land cover mapping and land products

The information derived from remotely sensed data must be carefully used because there are many sources of error that potentially affect its quality. In this respect, an accurate full calibration process of any sensor is essential because it can minimise various uncertainties and errors that reduce the final quality of the products derived from remotely sensed data. This paper presents several calibration efforts performed on MERIS data and subsequently focuses on the smile effect as well as on the vicarious calibration corrections. The implications of these corrections are evaluated using a MERIS full resolution level 1b image that was acquired over The Netherlands. A thematic approach, based on regional land cover mapping using linear spectral unmixing, and a continuous approach, based on continuous variables (FAPAR, MTCI and NDVI), are used to quantify these implications. Even though MERIS has a very high radiometric quality, results point out that these radiometric effects are consistently present in the final MERIS products. We also conclude that MERIS, after including all potential corrections investigated here, does not exhibit significant ( radiometric) deficiencies. However, from a strict point of view, all the proposed radiometric corrections should be applied to the data so that the retrieval of quantitative information can be done with the highest possible quality. The use of fully radiometrically corrected data will also facilitate multitemporal comparisons. Therefore, we finally conclude that a systematic application of all relevant calibration parameters will increase the long term comparability of MERIS measurements in such a way that more emphasis can be put on the retrieval of MERIS products.