The operational principle of offshore wind farms (OWF) is to extract kinetic energy from the atmosphere and convert it into electricity. Consequently, a region of reduced wind speed in the shadow zone of an OWF, the so-called wind-wake, is generated. As there is a horizontal wind speed deficit between the wind-wake and the undisturbed neighboring regions, the locally reduced surface stress results in an adjusted Ekman transport. Subsequently, the creation of a dipole pattern in sea surface elevation induces corresponding anomalies in the vertical water velocities. The dynamics of these OWF wind-wake induced upwelling/downwelling dipoles have been analyzed in earlier model studies, and strong impacts on stratified pelagic ecosystems have been predicted. Here we provide for the first time empirical evidence of the existence of such upwelling/downwelling dipoles. The data were obtained by towing a remotely operated vehicle (TRIAXUS ROTV) through leeward regions of operational OWFs in the summer stratified North Sea. The undulating TRIAXUS transects provided high-resolution CTD data which enabled the characterization of three different phases of the ephemeral life cycle of a wind-wake-induced upwelling/downwelling dipole: development, operation, and erosion. We identified two characteristic hydrographic signatures of OWF-induced dipoles: distinct changes in mixed layer depth and potential energy anomaly over a distance < 5 km and a diagonal excursion of the thermocline of ~10–14 m over a dipole dimension of ~10–12 km. Whether these anthropogenically induced abrupt changes are significantly different from the corridor of natural variability awaits further investigations.

The general task of image classification seems to be solved due to the development of modern convolutional neural networks (CNNs). However, the high intraclass variability and interclass similarity of plankton images still prevents the practical identification of morphologically similar organisms. This prevails especially for rare organisms. Every CNN requires a vast amount of manually validated training images which renders it inefficient to train study‐specific classifiers. In most follow‐up studies, the plankton community is different from before and this data set shift (DSS) reduces the correct classification rates. A common solution is to discard all uncertain images and hope that the remains still resemble the true field situation. The intention of this North Sea Video Plankton Recorder (VPR) study is to assess if a combination of a Capsule Neural Network (CapsNet) with probability filters can improve the classification success in applications with DSS. Second, to provide a guideline how to customize automated CNN and CapsNet deep learning image analysis methods according to specific research objectives. In community analyses, our approach achieved a discard of uncertain predictions of only 5%. CapsNet and CNN reach similar precision scores, but the CapsNet has lower recall scores despite similar discard ratios. This is due to a higher discard ratio in rare classes. The recall advantage of the CNN decreases with increasing DSS. We present an alternative method to handle rare classes with a CNN achieving a mean recall of 96% by manually validating an average of 6.5% of the original images.