Introduction and Limits of Use
The surfaces were generated in a three step process: (1) estimation of hydroregime variables for 34 point locations (weather stations) across the Central Valley using the PHYDO hydroregime model, (2) fitting regression models to these stations, (3) applying the regression models to capture large-scale trends, (4) estimating residuals for each station, (5) interpolating residuals across the Valley to capture fine-scale variation and restore exact modeled values at each station, and (6) add the regression models and the interpolated residuals to create final raster surfaces (see Pyke (2002) for a detailed discussion of methods). In some cases these surfaces were cleaned with conditional operators to pick off outlying values that sometimes occur along the edges of the Valley. The most typical operation removed flood durations greater than 365 or less than 0 and replaced them with 0 values. These filters were not applied at all to the"pools-only" surfaces.