Abstract
The main objective of this thesis is to assess global trends over a 25-year period in ocean colour variables and phytoplankton bloom timings in a changing climate using merged ocean colour data. Satellite-derived ocean colour data provide continuous, daily measurements of global waters and are a relevant tool for monitoring their health in a changing climate. Phytoplankton dynamics and blooms are important to monitor because they are an essential part of the marine food web the carbon cycle. However, the significant time series inconsistencies that arise from combining ocean colour data from different missions within merged datasets must first be addressed, which is the secondary goal of this thesis. The merged ocean colour data set used, was compiled by the ‘Ocean Colour Climate Change Initiative’ project (OC-CCI-v6). Merging observations from different satellite sensors is necessary for long-term and continuous climate research because the lifetime of these ocean colour sensors is limited. Although satellite mission calibration bias corrections have been performed on merged data set products, significant inconsistencies between missions remain. These inconsistencies appear as sudden steps in the time series of these products when a satellite mission is added to- or removed from the merged data set. This inter-mission inconsistency is not caused by poor correction of sensor sensitivities but by differences in the ability of a sensor to observe certain waters. Merged data sets have a significantly increased spatio-temporal coverage compared to the individual sensor data sets. However, the greater coverage is unevenly distributed in time and space, depending on the used ocean colour data set(s). This has implications for the interpretation of long-term trends of ocean colour and phytoplankton bloom phenology time series. As a result, previous studies that used merged ocean colour data may be biased. It was found that coastal waters, high latitudes, and areas subject to changing cloud cover are most affected by coverage variability between missions. The “Temporal Gap Detection Method” is introduced, which temporally homogenises the observations per-pixel of the time series and consequently minimises the magnitude of the inter-mission inconsistencies. The method presented is suitable to be transferred to other merged satellite-derived data sets that exhibit inconsistencies due to changes in coverage over time. The corrected data set allows the examination of long-term trends and statistics of variables derived from ocean colour data with greater accuracy. Global and local trends of the chlorophyll-a concentration, optical properties, and bloom phenology were inferred from the corrected ocean colour data set. Sea surface temperature, salinity, and several climate oscillations were included in the analysis to gain insight into the underlying processes of the derived trends. The bloom phenology analysis includes a per-pixel trend estimation of bloom initiation, peak timing, duration, and bloom height. Results show that the bloom dynamics are changing substantially in the Southern Ocean and the North Atlantic Ocean. The derived trends indicate a significant increase in chlorophyll-a concentration in the polar waters, a decrease in chlorophyll-a concentration in some equatorial waters, and point to ocean darkening, predominantly in the polar waters, due to an increase in non-phytoplankton absorption. This study contributes to broader knowledge of global trends of ocean colour variables and their relation to a changing environment, and it is the first to use a consistently covered merged ocean colour data set.
