Synthesis of Major Findings at PUB 2011 and Recommendations for Future Directions

This chapter discusses three tasks that are considered necessary to advance
improvements in prediction in ungauged basins in applied situations. First,
incorporation of new hydrological process understanding into predictive
models needs to continue. This should enable models to predict more than
the traditional end point; streamflow, and expand into others such as storage
in the snowpack and subsurface. It will also reduce uncertainty when dealing
with non-stationarity. Second, methods need to be introduced that help
constrain, rather than calibrate, model parameters. Traditional calibration
could be used to solve for these parameters; but there are reasons to avoid
this approach, non-stationarity being only one. Constraining parameters is
useful as it helps locate uncertainty, and how it could be addressed.
Information for constraining parameters can come from a wealth of sources,
including existing hydrological indices, independent measurements from
remote sensing, or research catchments. Third, work must continue to
encourage the adoption and implementation of robust tools by practicing
hydrologists. Unfamiliarity and adhering to accepted standards of practice
are two reasons why practitioners are sometimes hesitant to adopt new
approaches. Time limitations and a lack of easy access to new tools are
significant logistical impediments to adoption as well. Some countries have
addressed these problems with training courses as a means of technology
and knowledge transfer. Such collaborative efforts are crucial to improve
water management and prediction systems.