|
<br />788
<br />
<br />ABBS: INVESTIGATION OF PROBABLE MAXIMUM PRECIPITATION ASSUMPTIONS
<br />
<br />I
<br />
<br />"
<br />
<br />
<br />......_. 325
<br />
<br />..................... ................. 34S
<br />
<br />14l!:E
<br />
<br />,...
<br />
<br />,...
<br />
<br />mE
<br />
<br />Figuri~ 1. Map -showing locations mentioned in the text.
<br />Shading indicates terrain above 250 m with a contour interval
<br />of 500 m,
<br />
<br />front over the mountainous terrain of southeastern Australia
<br />where it gave a good distribution of the rainfall [Ryan and
<br />A'th.~, lQQ1J. Further OhRl:lVlltional and modelin~ studies hy
<br />Ahhs dnd Jensen [19921 shown that the mode:} rcalistic<llly cal~
<br />culatcs ooth the in-cloud structures of the clouds that form
<br />over 1 hose mountains and the resulting pre:cipitation. These
<br />results show that for this" simulation, RAMS correctly models
<br />the atmospheric portion of the water cycle for this storm.
<br />
<br />3. Case Study
<br />
<br />10 our overall study, four case studies have been completed
<br />IAbbs and Ryan, I997}. These comprised (I) an cast coast low,
<br />(2) an upper-level cutoff low (both repr."entative of the
<br />GSAIvI), (3) a northwest Australian tropical cyclone, and (4) a
<br />northeast Australian tropical cyclone (both representative of
<br />the GTSM), In this paper we will only present the results from
<br />case study I (the east coast low of August 5- 8, 1986), This case
<br />has been chosen to illustrate the techniques used and to de-
<br />scribe the conclusions that were common to all case studies.
<br />The conclusions, based on the results from the four case stud-
<br />ies, <l:re presented in section 4. The results of the other case
<br />studies (and hence the evidence confirming the conclusions
<br />presented here) are described by Abbs and 19'an [1997],
<br />FOor each of these cases we investigated the effects that
<br />increases in the moisture availability have on the precipitation
<br />produced by the storm. on the precipitation efficiency of the
<br />storm, and on the DDA analyses for the storm. We have also
<br />investigated the effect of terrain on the distribution of the
<br />precipitaIion produced by Ihe east coast low and upper-level
<br />cutofr low. The extreme storms, representative of the GSAM,
<br />chosen for the study are the east coast low of August 5-8,1986,
<br />and the upper-level cutoff low of April 28 to May 1, 1988, BoIh
<br />of Ihese evenIS caused significant spill from the Wanagamba
<br />Dam (A. Dodds, private communication, 1997). See Figure 1
<br />for locations mentioned in the text.
<br />East coast lows have heen identified as the major cause of
<br />Ilood-producing rains on the east coast of Australia. These
<br />intense extratropical cyclones are characterized by heavy rain-
<br />
<br />...
<br />
<br />fall and strong winds, often of Iropical cyclone strength, Hol-
<br />land et ai, [I987J idenIified Ihree Iypes of easI coast lows that
<br />occur typically in autumn or winter with an average frequency
<br />of one or two per year. Because east coast lows arc agngrnlly nf
<br />,ub.yooptlc .cnlo nod develop rapidly, Ihey are difficult to
<br />forecast, and operational quantitative prediction forecasts
<br />have been poor. Leslie et al. [1987} investigateu the predict-
<br />ahility of easl coast lows and showed that the initial develop-
<br />ment of these systems could he forecast in a numerical model
<br />.of l50-km horizontal resolution. However, they found that
<br />higher resolution is required to capture fully the intensity.
<br />structure, and Irack of the system, In later S!udies. Hess [1990],
<br />Melnnes and Hess [I992J, and Golding and Leslie [1993J dem-
<br />onstrated the sensitivity of model results to improvements in
<br />both the model resolution and physical parameterizatioos.
<br />The meteorology of thg formAtion Rod development or the
<br />east coast low of August 5-8, 1986, has been described in detail
<br />by both Lynclt [1987J and Bureau of Meteorology [1987J. This
<br />event produced the worsI flooding 10 occur in Sydney for more
<br />Ihan a century; the heavy rain exIended to the Central Table-
<br />lands and IIIawarra districIs and caused major flooding of the
<br />principal river systems of Ihe Sydney basio over this period,
<br />Tl1e Sydney rainfall for the 24 hours ending 0900 L T on August
<br />6 was 328 mm [Bureau of Meteorology. 1987J, wiIh the most
<br />intense phase of Ihe rainfall occurring on August 5, The rain-
<br />fall was associated with a low-pressure system that had formed
<br />off the coast near Port Macquaric during tho oYl:nlt1g uf Au-
<br />gust 4, then moved SSW to near Norah Head-hy 0900 on the
<br />August 5. The low remained in this approximate area for the
<br />next 18 hours. Intcnsity-frequency-duration curves for Sydney
<br />Central showed that for short durations the intensities were
<br />not exceptional, but that for durations beyond 8 hours the
<br />average intensity exceeded the 1 in lOO.year event [Bureau of
<br />Meteorology. 19871,
<br />
<br />3.1. Model Initialization
<br />
<br />In the simulations discussed here the numerical model has
<br />been initialized using European Centre for Medium-Range
<br />Weather Forecasts (ECMWF) analyses that are available at a
<br />resolution of 2,SO latitude aod longitude for the entire globe,
<br />These analyses have been interpolated horizontally and verti-
<br />cally to the coarsest mesh; Ihey also provide the temporal
<br />forcing on Ihe lateral boundaries of the coarsest mesh. Three
<br />levels of interactive grid nesting were used, the coarsest having
<br />a horizonIal grid spacing of -60 km and the finest having a grid
<br />spacing of -7 \un. Ideally, a higher resolution would have been
<br />desirable, but computing constraints made this difficult to
<br />achieve. The grid dimensions used for this simulation arc 50 x
<br />64 for the coarseS! mesh, 42 X 50 for Ihe iotermediaIe mesh,
<br />and 86 x 98 for the finest mesh. There were 30 levels in the
<br />vertical, with the bottom level at -50 m and the top of the
<br />model domain at 24 km,
<br />The terrain used on all meshes was interpolated from a
<br />1/40th-degree data set. The sea surface temperatures (SST)
<br />were obtained from the ECMWF analyses and enhanced in the
<br />waters off southeasIern Australia wiIh data from the weekly
<br />SST analyses of the Royal Australian Navy. In these simula-
<br />tions the microphysics parameterization was activated on a\l
<br />grids, In addition to Ihe microphysics scheme, the convective
<br />parameterizaIion scheme developed hy Frank and Cohen
<br />[1985, 1987J was used on the two finest meshes. Companion
<br />simulations [Abbs and Lee, 1997J made as part of our study
<br />showed that for the cases presenIed it was necessary to use a
<br />
<br />!
<br />I
<br />I
<br />
|