The limits of homogenization: How and how much can a simple model represent hydrological dynamics at the catchment scale?

: Large-scale models often use a single grid to represent an entire catchment assuming 29 homogeneity; the impacts of such an assumption on simulating evapotranspiration and streamflow 30 remains poorly understood. Here we compare hydrological dynamics at Shale Hills (PA, USA) using a 31 complex model (spatially distributed, >500 grids) and a simple model (spatially implicit, two grids using 32 “effective” parameters). We asked two questions: What hydrological dynamics can the simple model 33 reproduce? What hydrological dynamics does the simple model miss by ignoring spatial details? Results 34 show the simple model can reproduce annual runoff ratios and evapotranspiration, daily discharge peaks 35 but not discharge minima. Neither can it reproduce different streamflow arising from the distinct land 36 surface characteristics at the two sides of the catchment. The similar annual runoff ratios between the two 37 models indicate spatial details are not as important as climate for annual scale evapotranspiration and 38 discharge partitioning. Most of the calibrated parameters in the simple model are within the ranges in the 39 complex model, except that soil porosity has to be reduced to 40% of the average porosity from the 40 complex model. Comparison between the two models indicates that the “effective” porosity in the simple 41 model represents the volume-averaged porosity in the effective drainage area of the complex model that is 42 connected to the stream; it does not represent the porosity in disconnected, uphill areas. This indicates that 43 an additional uphill functional unit is needed in the Simple model to simulate the full spectrum of the 44 high-versus-low streamflow dynamics in natural catchments. 45