Here we explore the relative contribution of the Madden-Julian Oscillation (MJO) and El Niño Southern Oscillation (ENSO) to midlatitude subseasonal predictive skill of upper atmospheric circulation over the North Pacific, using an inherently interpretable neural network applied to pre-industrial control runs of the Community Earth System Model version 2. We find that this interpretable network generally favors the state of ENSO, rather than the MJO, to make correct predictions on a range of subseasonal lead times and predictand averaging windows. Moreover, the predictability of positive circulation anomalies over the North Pacific is comparatively lower than that of their negative counterparts, especially evident when the ENSO state is important. However, when ENSO is in a neutral state, our findings indicate that the MJO provides some predictive information, particularly for positive anomalies. We identify three distinct evolutions of these MJO states, offering fresh insights into opportune forecasting windows for MJO teleconnections.

Robust and reliable projections of future streamflow are essential to create more resilient water resources, and such projections must first be bias corrected. Standard bias correction techniques are applied over calendar-based time windows and leverage statistical relations between observed and simulated data to adjust a given simulated datapoint. Motivated by a desire to connect the statistical process of bias correction to the underlying dynamics in hydrologic models, we introduce a novel windowing technique for projected streamflow wherein data are windowed based on hydrograph-relative time, rather than Julian day. We refer to this method as ‘seasonally anchored’. Four existing bias correction methods, each using both the standard day-of-year and the novel windowing technique, are applied to daily streamflow simulations driven by 10 global climate models across a diverse subset of six watersheds in California to investigate how these methods alter the model climate change signals. Among the methods, only PresRat preserves projected annual streamflow changes, and does so for both windowing techniques. The seasonally anchored window PresRat reduces the ensemble bias by a factor of two compared to quantile mapping (Qmap), cumulative distribution function transform (CDFt), and equidistant quantile matching (EDCDFm) methods. For wet season flows, PresRat with seasonally anchored windowing best preserves the original model change over the entire distribution, particularly at the highest quantiles, and the other three methods show improved performance using the novel windowing method. Concerning temporal shifts in seasonality, PresRat and CDFt preserve the original model signals with both the novel and standard windowing methods.