Acoustic trawl surveys are routinely conducted around the world. They consist on following pre-defined transects, imaging the water column acoustically using downward active acoustic systems (so called scientific echosounders) and conducting regular trawling. This type of survey is used to assess the state of marine species (most often fish) at a specific point in time. Whilst echosounder data and trawling information are the base of an acoustic trawl survey, various types of ancillary data can be collected to complement the survey. An important part of the processing of data collected during a fish species acoustic survey is the scrutinisation process. It consists on the categorization and allocation of echo traces into species or species groups. Through expert judgement of the analyst, this relies on acoustic (frequency components, school morphology, time of day, depth etc) and biological (fishing) information. This information is usually specific to each survey and each species. For difficult scrutinisation cases, ancillary data (e.g. additional acoustic data, video data) can significantly aid this process. In this report, the use of two type of ancillary acoustic data collected during the herring Acoustic Survey (HERAS) onboard Tridens II are investigated: the multi-beam echosounder data (MBES, ME70) and the multi-frequency data from the split-beam echosounders (EK60/EK80 Continuous Wave). First, MBES data from HERAS 2016 were analysed with the aim of: (1) building a database of 3D fish school descriptors for potential future use during HERAS survey scrutinisation, (2) comparing MBES data with single beam echosounder data (EK60) in the context of abundance estimation. The results of the comparison between the ME70 and the EK60 showed that results are in line though an increased coverage of the water column with the ME70 induces greater variability. Historical HERAS data were analysed in order to investigate acoustic fingerprints for herring during this survey. The HERAS surveys from 2015 to 2019 were reanalysed using LSSS (Large Scale Survey System) automatic routines for fish school detection. The fish school descriptors and acoustic fingerprints were then used for classification using a Neural Network (NN). The NN classifier yielded an accuracy for herring of ~70%.