Figure Captions and Tables
Figure 1: A conceptual model of the specific conductivity (SC) patterns that streamwater takes during and after a precipitation event. Highlighted are the aspects that we use to characterize the watershed response to individual precipitation events: the flush, dilution, and rate of SC recovery. Also shown is the pre-storm trajectory, or the course of streamwater SC prior to the influence of a precipitation event, which is used to quantify both the magnitude of the flush and the magnitude of the dilution.
Figure 2: A map of the Clay Brook catchment.
Figure 3: The full record of specific conductivity (SC) and precipitation data for the months we analyzed in both 2013 and 2014.
Figure 4: A visual representation of the parameter ranges for the Monte Carlo calculations. Parameter A defined the minimum amount of precipitation for an event to be considered for analysis. Parameter B defined both the time between individual precipitation measurements that separated unique events, and the range of time to define the pre-storm trajectory (PST). Parameter D defined the amount of time for a flushing response to be found after the storm began. Parameter E defined the amount of time for a dilution response to be found after the flush occurred, or after 6 hours prior to the end of the storm’s precipitation if no substantial flush was present. Parameter F defined the amount of time in which the slope of the rate of specific conductivity (SC) recovery was calculated after the dilution. Parameter G defined the number of standard deviations above or below the PST line that a peak flush or maximum dilution must have exceeded in order to be declared substantial enough for analysis.
Figure 5: The algorithm-derived points of peak flush and maximum dilution for a precipitation event beginning on May 8, 2013. and their respective distributions. Of note are the three points in which a peak flush was found across iterations, due to the variable range of Parameter D, and the increasing specific conductivity (SC) values of those points occurring in time. In this way, the algorithm seeks the truest representation of a flush over time. A) The distributions of algorithm derived peak flush values (number of times a peak flush was detected / number of times the precipitation event was detected). B) The median of the extracted peak flush and maximum dilution values from each of the algorithm’s iterations. These points overlay the SC data surrounding the precipitation event. The pre-storm trajectory (PST) and rate of SC recovery (RSC) are also displayed with estimated intercepts for appropriate visual representation. C) The distribution of algorithm derived maximum dilution values (number of times a maximum dilution was detected / number of times the precipitation event was detected).
Figure 6: The algorithm-derived points of peak flush and maximum dilution for a precipitation event beginning on October 4, 2014. and their respective distributions. Unlike Figure 5, the earlier point for peak flush is chosen more frequently by the algorithm because the first has a greater specific conductivity (SC). In this way, the algorithm retains the truest representation of a flush despite later options being available. A) The distributions of algorithm derived peak flush values (number of times a peak flush was detected / number of times the precipitation event was detected). B) The median of the extracted peak flush and maximum dilution values from each of the algorithm’s iterations. These points overlay the SC data surrounding the precipitation event. The pre-storm trajectory (PST) and rate of SC recovery (RSC) are also displayed with estimated intercepts for appropriate visual representation. C) The distribution of algorithm derived maximum dilution values (number of times a maximum dilution was detected / number of times the precipitation event was detected).
Figure 7: Kendall’s τ relationships for the magnitude of peak flush (FSC) and A) precipitation intensity (IP), B) maximum precipitation intensity (IP,max), and C) total precipitation (PT). Whiskers indicate the interquartile range of the values obtained from the 10,000 iterations. An absence of whiskers indicates 100% confidence in the value across all iterations.
Figure 8: Kendall’s τ relationships for the magnitude of maximum dilution (DSC) and A) precipitation intensity (IP), B) maximum precipitation intensity (IP,max), and C) total precipitation (PT). Whiskers indicate the interquartile range of the values obtained from the 10,000 iterations. An absence of whiskers indicates 100% confidence in the value across all iterations.
Figure 9: Kendall’s τ relationships for the rate of specific conductivity recovery (RSC) and A) precipitation intensity (PT) and B) magnitude of the dilution (DSC). Whiskers indicate the interquartile range of the values obtained from the 10,000 iterations. An absence of whiskers indicates 100% confidence in the value across all iterations.
Table 1: Ranges for each parameter’s uniform distribution. The PT is the total precipitation for a storm, the RSC is the recovery rate of SC after peak dilution, and PST is the pre-storm trajectory.