Summary: | F I G U R E 3 The residual error between observed and predicted water temperature (top panel), and the in-situ water temperature (black line) and long short-term memory neural network water temperature prediction (red crosses) for the training (1961–1994) and validation (1995–2019) dataset in the Mill Race (bottom panel). Years excluded due to accumulation of internal sate (green), prolonged periods of missing data (blue shaded), and measurement error (red shaded) are shown in the top panel, and the delineation of the training and validation period is shown by the vertical dashed line in both panels. Published as part of Rinaldo, Adrian, Eyto, Elvira de, Reed, Thomas, Gjelland, Karl Øystein & McGinnity, Philip, 2024, Global warming is projected to lead to increased freshwater growth potential and changes in pace of life in Atlantic salmon Salmo salar, pp. 647-661 in Journal of Fish Biology 104 (3) on page 652, DOI:10.1111/jfb.15603, http://zenodo.org/record/10981564
|