FLOOD07101 - Laserfiche WebLink

Home Browse Search

792 ABBS: r:;VESTIGATION OF PROBABLE MAXIMUM PRECIPITATION ASSUMPTIONS (a) 150 150 150 Figure 6. Plots of the percentage of the total rainfall that falls as heavy rainfall (i.e., with a rainfall rate >25 mm h~l) for (a) the control simulation. (b) EC86_3, and (c) EC86_5, The re'tllt.< shown orc for grid 3. Black contours are for 10%, and dark shading is for percentages >50%. rainfall. For Ihe casc of the heavy rain[all. precipitation pro- duction is 80%-90% efficient. As the amount of moisture available to the system increases, there is a tendency for PEr to decrease for moderate rainfall. The lower plots of Figures Sa. 5h, and 5c show that as the moisture available to the system is increased, the number of heavy rainfall grid points increases and the heavy rainfall is longer-lived, The percentage of the total rainfall that occurs as heavy rainfall has been calculated for the control simulation. EC86_3, and EC86_S and is shown in Figure 6, For cach simulation the rainfall that falls in the main cloud hand over the sea occurs as heavy rainfall. Over the land, only a small percentage falls as heavy rainfall, but this amount increases as the moisture availability increases. Figures 5 and 6 indicate that long-lived moderate rain processes are important and that the efficiency of these processes may decrease as the moisture availability increases. On the other hand, this may be offset by increases in the percentage of rain falling as heavy rainfall. 3.2.4. Terrain effects. When estimating PMP values for a particular storm, it is usual to attempt to separate out the convergence and topographic components of the storm. The concept underlying this technique is that the convergence com. ponent is due to synoptic-scale atmospheric disturbances such as frontal lifting. The portion of the rainfall that cannot be separated into the convergence component is defined as the topographic component of the storm. In this section, results are presented for Q simullltion wllhnut Ictrltln: thlli ltimuhulun mllY be considered to be the model equivalent of the convergence component for this storm. The convergence component rainfall is then subtracted from the rainfall of the control simulation to provide an estimate of a model-derived topographic component of this storm. The con. vergence and topographic components of the storm are pre- sented in Figure 7. These results show that over the Blue Mount'.lins and IIIa- warra escarpment almost all of the rainfall is due to the topo- graphic component of the storm. However, along the conver- gence line the results are more complicated. The removal of terrain has allowed the entire system to move south faster than in the case of the control simulation. Consequently, the rainfall maximum of 267 mm, for the "no terrain case," occurs farther south than the corresponding maximum of 355 mm for the control simulation, The effect of this difference may be seen in the adjacent, large maximum and minimum values of Figure 7b, A topographic component of -100 mm in this region is more realistic. However, these results illustrate that terrain effects do feed back to synoptic-scale aspects of the storm. 3.2.5. Depth~duration-area analysis. The model~derived precipitation fields have been uscd to calculate DDA curves for the control simulation, EC86_3, and EC86_S, These curves are based on hourly rainfall rates for the finest mesh simulated and calculated using the same method as Minty et ai, [1996]. This melhod differs slightly from the standard method [WMO, 1969J in that multicentercd storms are trcated as if they werc single-centered [Minty et ai" 1996. P, 8J: I I I I , \ I I i The program which determines maximum depth-area curve for each storm simply cOllnts the numher of grid points in hctwccn evenly incrcmented isohycts, from the maximum to the minimum isohyct. and calculates the arithmetic mean rainfall per interval. Area calculalions are based upon the number of grid points counted and the known resolution of the grid. In all cases the hourly rainfalls used in these analyses are from 6 hours of simulation to the completion of the simulation.