Seasonal Hydrologic Predictability: Sources and Limitations

Abstract:

I first review sources of predictability in seasonal hydrological forecasts. Recent work shows that at short lead times (typically up to a few months), hydrologic forecast skill is mostly controlled by hydrologic initial conditions (primarily soil moisture and where and when relevant, snow water storage), but at longer lead times, climate forecast skill dominates. Unfortunately, aside from a few special situations, climate forecast skill for lead times beyond about a month is minimal. Therefore, for practical purposes, hydrological initial conditions are the primary source of hydrological forecast skill. This is the premise of the widely used Ensemble Streamflow Prediction (ESP) method. I also investigate barriers to the use of seasonal hydrological forecasts in water resource systems operation. I review in particular work approximately a decade ago by Maurer, which casts light on the potential for improved reservoir system operations through improved forecasts as a function of the usable reservoir storage relative to the mean annual inflow, relative to the simplest forecast (climatology). In general, the potential economic benefits of improved forecasts are largest for relatively small reservoirs.