3). We used maximum daily water level measured in 18 wells (Lyon et al., 2006) recorded via WT-HR 500 capacitance probes (TruTrack, Inc., New Zealand). We ran the watershed model using precipitation data measured on-site and temperature data from Delhi, NY. On days when runoff was predicted, we divided the wells into “wet” locations where our model predicted runoff generation and “dry” locations where our model predicted no runoff generation to compare water table depths between groups. Volumetric

soil moisture measurements were taken Selleck Obeticholic Acid at two field sites in Fall Creek and Cascadilla Creek watersheds (near Ithaca, NY) over the course of Fall 2012 and Spring 2013 (Fig. 4). Measurements were taken in triplicate using a TDR probe over a range of wetness classes (Buchanan et al., 2013). We assigned a wetness class to each sampling location using

a 3-m LIDAR derived STI value (same method as in Test 2). For each measurement date, we modeled the extent of saturated areas in the contributing watershed that were predicted to generate runoff on that particular date. Using this IDH inhibition breakdown, we assigned each soil moisture measurement point a predicted value of “wet” and “dry” based on whether the model predicted the point to be generating runoff or not, respectively. This was compared to the soil moisture status of these wet and dry locations. The number of wet and dry locations changed on each measurement date, depending on the extent of saturation predicted for that day. We estimated the porosity of the soil as 53% assuming minimal organic matter using the bulk density reported in the USDA SSURGO data set (USDA-NRCS, 2013). We found there FER was a significant (p < 0.001) linear relationship between Sd and SWDd, which is represented by Eq. (6) and overall coefficients reported in Table

2. equation(6) Sd=Smin+C1(SWDd)Sd=Smin+C1(SWDd)We recalculated this relationship by excluding data from each watershed individually, and found that the relationship remained significant at the p < 0.001 level for each watershed excluded, with the intercept, Smin, varying between 78 and 86 mm, and the slope, C1, varying between 3.3 and 3.5 ( Table 2 and Fig. 5). This suggests that we can use Eq. (6) to determine Sd from SWDd directly, without needing to calibrate unique coefficients for individual watersheds, i.e., we can use the average values for Smin and C1. The best-fit Tp values were well correlated (R2 = 0.80, p < 0.01) to Tc ( Fig. 6), and we determined a linear relationship that allows us to estimate Tp based on Tc: equation(7) Tp,c=C2Tc+C3Tp,c=C2Tc+C3where Tp,c is the calculated time to peak (h), C2 is a fitted slope of 0.33 (unitless), and C3 is the fitted intercept of 3.4 (h). We recalculated C2 and C3 using the leave-one-out method ( Fig. 6); R2 varied between 0.77 and 0.88 for the various combinations of nine watersheds, C2 varied between 0.28 and 0.