Estimating prediction horizon of reservoir computer on L63 system when observed variables are incomplete

Abstract Reservoir computer (RC) is an attractive neural computing framework that can well predict the dynamics of chaotic systems. Previous knowledge RC performance established on case all variables in a system are completely observed. However, practical circumstances observed from dynamical usually incomplete, among which there lack understanding performance. Here we utilize mean error growth curve to estimate prediction horizon Lorenz63 (L63), and particularly investigate scenario univariate time series. Our results demonstrate outperforms local analogs L63, state-space embedding technique improve incomplete observations. We then test conclusion more complicated systems, extend method intraseasonal predictability atmospheric circulation indices. These could provide indications for future developments applications RC.