FLOOD03496 - Laserfiche WebLink

Home Browse Search

radar estimates from the nine surrounding polar grid bins, it assigns a radar estimate thaI: is exactly equal to the gauge observation, i.e., it forms a perfect-match (unbiased) gauge-radar pair. This is illustrated in Fig, 8 by the closed circles that fall exactly along the I: I diagonal line. If the gauge observation falls outside of the nine-bin range, then the closest radar estimate of the nine bins to the gauge observation is chosen in forming the gauge-radar pair. The closed circles that are displaced from the diagonal line in Fig. 8 illustrate this situation. This methodology is designed to account for the uncertainties in precise knowledge of the gauge location and the exact radar bin(s) that contribute rain to the gauge (due to horizontal advection ofrain below the beam), For this case. there is a large breadth of radar rainfall estimates in the nine polar bins surrounding most gauges due partly to the convective nature of the rainfall on this day and partly to the increasing beam width with range. Stratiform rainfall events with typically smaller horizontal gradients of rainfall may not exhibit such large 3x3 bin variability. The net impact of this method of forming gauge-radar pairs in the PPS Adjustm,~nt algorithm on the resulting bias estimates appears to be an artificial smoothing out of any true biases that may exist even betJre the data gets passed into the Kalman filter bias estimation procedure. There is a tendency to cause the computed bias estimates to be closer to unilty than they might otherwise be if other methods were used. Table I illustrates this. It shows the storm- total "sample" bias (i.e., the bias based on gauge-radar observations as opposed to the Kalman filter-based bias estimates from the Adjustment algorithm) according to the fonnula B.....I. = L G, I L R; 12