3_Figure_7.zip

This zipped folder contains eight python dictionary objects saved in Pickle format. The files ending in “36s” have results using the optimal parameter set, while the files ending in “37s” have results assuming no inverse storage effect (i.e., lambda=0). The files beginning with “frac_temp” have 4 layers of keys. The first key [“TW”,“LH”] indicates either Lower Hafren or Tanllwyth. The second key [0,1,2,3,4,5] refers to five different rainfall scenarios. Number zero is full rainfall variability. Number one is number zero minus seasonality. Number twp is number one minus within storm variability. Number three is number two minus between storm variability. Number four is no variability. Number five is the historic period from 1999-2008. The third key [“yearly”, “summer”, “dec_mean”, “winter”, “winter_mean”, “jun_mean”, “summer_mean”, “mean”] indicates a time period of averaging. For example, “yearly” contains the mean age fraction of water per year, “dec_mean” contains the mean age fraction of water each year in December, and “mean” contains a single value for the entire period. The data record is 100 years long for rainfall scenarios 0-4, and 10 years long for rainfall scenario 5. The last key [1,2,3,…365] indicates the age fraction of water reported in days. For example, “90” return the fraction of water less than 90 days old in discharge. Thus, to summarize, the object frac_temp[‘TW’][1][‘summer’][90] returns an array with 100 values representing the yearly mean fraction of water younger than 90 days in the Tanllwyth watershed given a synthetically generated, 100 year precipitation input with full variability minus seasonality. The files beginning with “disch”, “prec”, and “stor” have the daily discharge [mm/day], precipitation [mm/day], and storage anomaly [mm] time-series for each watershed and for each of the five categories of rainfall. The time-series are 100 years long.