A documentary proxy for coastal upwelling

Coastal upwelling is a process in which sea water from intermediate depths emerges along the west coast of continents in response to the intensity of the alongshore component of the wind. These areas are economically relevant as upwelled water is rich in nutrients, creating regions of great interest for fisheries.
Quantifying upwelling intensity at multidecadal to secular time scales is quite difficult. Upwelling is typically evaluated with indices based on sea surface temperatures, a variable that at the spatial scale required, inevitably makes use of satellite retrievals. Evidently, indices based on this method are limited to the satellite era. An alternative approach makes use of the ultimate driver of the upwelling: the wind velocity. This is typically done by using reanalysis products, a kind of data that in open ocean, strongly relies on wind observations taken from the International Comprehensive Ocean-Atmosphere Data Set (ICOADS) database. However, during the last decade, it has been suggested that at multidecadal scale, due to the changes in the observations methods, ICOADS’ velocities over the oceans could be biased toward increasing values, resulting in unrealistic trends in any wind-derived climate index.
In this work we bypassed these problems disregarding the wind velocity field and making exclusively use of in-situ wind direction observations to compute a so-called “Directional Upwelling Index” (DUI). This kind of “directional” indices can be tuned to measure the persistence of the alongshore winds at the coast and therefore, they are directly related to the presence of upwelling favourable conditions. As the wind direction observation was routinely taken by sailing ships traveling between Europe and Asia circumnavigating Africa since long ago, our DUI can be used to track the history of the upwelling in parts of the African coasts since the first half of the 19^th century, avoiding any problem related to possible biases on the wind velocity. This is a period much longer than that covered for traditional upwelling indices.
We have computed DUIs for several areas along the west African coast. In general, we have not found consistent long-term trends, although the results are dependent on the specific area analysed. At the Senegal-Mauritania latitude, the upwelling intensity has been very variable during the last two centuries and there is strong evidence of a large shift in the upwelling intensity that occurred around 1900. Further north, at the Canaries latitude, we found that the Atlantic Multidecadal Variability mode seems to control a significant part of the upwelling variability at decadal scales, resulting in an alternating change in the sign of upwelling trends computed over short periods. This could explain the contradictory results currently found in the bibliography at the time of interpreting the upwelling trends based on short series.