We present a hydrological modeling framework that treats the landscape as comprising a set of dynamic, cascading, hierarchical, non-linear filters. In this talk, we will present an example of its application to elucidate patterns of inter-annual variability in observed catchment responses. We introduce the Horton Index, H=E/W, which is the ratio of annual evapotranspiration, E, to annual plant available water, W. Our work has been motivated by the fact that H, estimated for over 400 catchments around continental United States, remained remarkably constant between years, with relatively small variance, regardless of climatic variability, suggesting that vegetation may have developed in a way to use a maximum fraction of plant available water. Detailed comparative analyses demonstrated that apart from mean climate, between-year variability of H was governed, to first order, by within-year patterns of precipitation events. Even then, models based on within-year climatic variability under-estimated the inter-annual variability of H, suggesting a more dynamic role being played by vegetation functioning. In order to understand this, we analyzed high resolution flux data from 12 Fluxnet sites. Classically, hydrologists have estimated evapotranspiration (ET) by first determining a maximum limit to ET as a function of radiation, temperature, leaf area and relative humidity (e.g., Penman-Monteith equation), and then censoring it as a function of soil moisture availability to account for water limitation and its effects on the conductance of the plant hydraulic apparatus.  We evaluated this classical paradigm through a combination of systematic data analysis and modeling, in which we followed a top-down modeling approach with the use of models of increasing complexity. 

Results/Conclusions

The classical predictive approach implemented via a Penman-Monteith equation was found to reproduce daily, seasonal and annual ET fluxes well for arid, shallow rooted systems. At deep-rooted sites, however, this approach over-estimated the effects of water limitation.  Analysis of latent heat fluxes and soil moisture indicated that plants were accessing groundwater sources that were effectively decoupled from precipitation at these sites.  Accounting for the additional water sources improved predictions of ET, except for a seasonal lag that was observed at several temperate sites.  Comparison of ET, radiation and leaf area index suggested that phenological factors controlled ET variability at these sites.  Parameterizing these effects with a simple growing season index further improved predictions of ET at these sites.