. 2018 Jul 3;5:180080. doi: 10.1038/sdata.2018.80

Table 5. Considered sources of uncertainty and investigation methods, following Westerberg and McMillan (2015, ref. 67).

Variable	Uncertainty sources	Investigation method	Reference
Precipitation	Equipment malfunction	Rainfall data with and without quality control	⁹⁰
	Point measurement error	Normal distribution with standard deviation as a function of rain rate	^48,51
	Rainfall intensity interpolation method, rainfall event definition	Composite cubic spline interpolation, threshold intensities, and bias correction	^{43,49–51,70}
	Spatial interpolation uncertainty	Subsampling from the available rain gauges in the network	⁶⁷
Streamflow	Water level measurement uncertainty, barometric and temperature pressure compensation	Sampling from uniform or normal distributions with standard error of ±5mm or using sensor nominal accuracy	^{67,74,75,79,82,91}
	Rating curve uncertainty, calibration and extrapolation	Constraining uncertainty by using a discharge control structure; voting point likelihood; heteroscedastic maximum likelihood model	^{56,81,83,85,92}
	Streamflow time discretisation	Comparing indices calculated at different time steps	^5,92
	Seasonality in hydrological response	Using total length of time series or divided by season	^38,93
	Hydrological index calculation method, e.g. baseflow separation method	Comparing different algorithms and methods; Correlation analyses; regionalisation	^{34,59,64,65,81,87}