Historically, seasonal drought forecasting has been achieved through T and Prcp, and understanding the relationships these variables have to large-scale coupled oceanic-atmospheric processes such as ENSO. To advance E0 forecasting (at any timescale; weather or seasonal) beyond post-processing of dynamical model output, we should strive to achieve better physical understanding of the relationships between E0 and sources of predictability (Tian et al.  and McEvoy et al. [2016b] touch on this for E0). This includes ENSO for seasonal forecasts but other indices including the Madden-Julian oscillation for subseasonal forecasts. Using modern, high-resolution datasets to build upon past work establishing T and Prcp relationships to large-scale climate patterns (e.g., Cayan et al. 1999; Redmond and Koch 1991) for E0 and the individual drivers is a logical step toward improving E0 forecasting. Such real-time E0 seasonal forecast products are currently in development.
For climate-scale drought vulnerability and ecological assessments requiring evaporative estimates, we must develop a robust climate-scale E0 or similar metric of aridity, and it must incorporate all physically relevant drivers— including the vegetative effects of increased atmospheric CO2—with uncertainties that are well expressed and transmissible through the analyses to users in a useful manner. Further, decisions at strategic planning timescales require support from more regional analyses.
Such advances will help us address more-specific questions. To what degree have ENSO, the Pacific Decadal Oscillation, and the Interdecadal Pacific Oscillation been affected by climate change? What are their effects on long-term ET and E0? Will nutrient cycling constrain increases in GPP with atmospheric CO2? Are the results of Roderick et al. (2015) robust across GCMs? How will their global conclusions regionalize or seasonalize?