Fig. 5: Predicted sensitivity index estimates using functional relationships based on basin characteristics.
From: The sensitivity of simulated streamflow to individual hydrologic processes across North America
The total Sobol' sensitivity indices of hydrologic processes on streamflow are first estimated using the xSSA method (Fig. 3; x-axis), after which they are predicted through basin-specific characteristics only (y-axis) and using the relations in Table 1. Each panel shows the results for one hydrologic process, i.e.: (a) infiltration, (b) quickflow, (c) evaporation, (d) snow balance, (e) convolution of surface runoff, (f) potential melt, (g) percolation, (h) rain-snow partitioning, and (i) precipitation correction. The mean and standard deviation of the Adjusted Coefficient of Determination \({R}_{adj}^{2}\) and the mean absolute error (MAE) are reported. Both indexes are derived between (i) xSSA-derived indices and (ii) the predicted indices and added as labels for calibration (cal) and validation (val) sets. The color indicates the density of samples. The samples shown here are the samples of the 100 validation experiments.
