4.3 The CTH and DTU gauges

In section 4.1 it was already shown that during a period of 16 months the CTH and DTU gauges register about 85% and 115% of the driving rain amount measured by the TUE-II gauge, respectively. Correlations of 10-min driving rain intensities of gauges CTH and DTU with gauge TUE-II are depicted in figures 4.1a and b. For these graphs, the measurement points have been selected into sets with wind speeds $U_y$ of 4-5 m s$^{-1}$ and with 6-7 m s$^{-1}$. The many measurement points on the $x$ axis in figure 4.1a indicate that the CTH gauge often measures no rain while the TUE-II gauge measures a positive driving rain amount. Because of this, the linear fit is not successful; the slope tends to be smaller than in case the points on the $x$ axis are left out. The linear fit through the measurement points with $U_y =$ 4-5 m s$^{-1}$ in figure 4.1a has a slope of 0.97 and a correlation coefficient ($r^2$) of 0.97. This ideal correlation is caused by three measurement points with very high driving rain amounts (between 2 and 6 mm h$^{-1}$, see figure 4.2a) for which the readings of CTH and TUE-II were almost the same. Figures 4.3a and 4.4a, where the measurement points have been additionally selected according to horizontal rain intensity, show much scatter and poor correlations.

Given the similarities in monthly driving rain amounts of the CTH gauge compared to those of the TUE-II gauge, and given the large scatter in the 10-min driving rain amount correlations between the two gauges, one is inclined to attribute the observed differences to differences in time response between the two gauges. Because the tipping bucket of the CTH gauge has to be filled completely before it can tip and give a reading, the reading of the CTH gauge can have only certain discrete values (properly said: the intervals are much larger than those of the TUE-II gauge readings). Figure 4.1a (and consequently also figures 4.3a and 4.4a) shows this. One tipping of gauge CTH in 10 minutes represents a driving rain intensity of approximately 0.18 mm h$^{-1}$. No measurement point can exist between a horizontal line at 0.18 mm h$^{-1}$ (one tipping) and the $x$ axis (zero tippings). Measurement points with two, three and more tippings are also forming horizontal lines at equal distances from each other. Compare the CTH results with those of the TUE-II gauge, which has a much lower threshold for 10-min periods (figure 4.1c).

The correlation of 10-min driving rain amounts of the DTU gauge with those of the TUE-II gauge is plotted in figure 4.1b. The slope of the linear fit is about 1, which is quite good. However, the fit through the measurement points with 4-5 m s$^{-1}$ ( $\includegraphics[width=0.5em]{gen/m-kruisje.eps}$) yields a slope of only 0.67. This is due to one measurement point for which a driving rain amount of 5.5 mm h$^{-1}$ according to gauge TUE-II corresponds with 2.3 mm h$^{-1}$ according to gauge DTU (figure 4.2b). Selections of the measurement points in figure 4.1b, according to driving rain intensities, are plotted in figures 4.3b and 4.4b. The latter figure shows quite good correlations between DTU and TUE-II: the slopes are 1.04-1.08, although due to the scatter the correlation coefficients are poor ($r^2=$ 0.39-0.54). The large scatter is probably due to the noise caused by the wind acting on the freely suspended collector. This noise is filtered out by simply averaging the signal over each 10-min clock period (section 3.4.6), which may be too simple a method. Figure 4.3b (selection with $R_{\text{h,c}}=$ 1.0-2.5 mm h$^{-1}$) shows fits with a slope of about 1.40, which indicates that lower driving rain intensities yield higher relatively readings of DTU. It seems therefore that the reading of DTU is now sensitive to the driving rain intensity.