Impact of Climate Variability and Climate Change on Rainfall Extremes in Western and Surrounding Areas

Report to the Upper Parramatta River Catchment Trust on Research Activities in Financial Year 2004-05

Debbie Abbs, Bryson Bates, Santosh Aryal

CSIRO Climatic Extremes Research Group

November 2004

1 Introduction

The CSIRO Climatic Extremes research group (CER) is investigating rainfall intensity-frequency-duration (IFD) and depth-area curves for western Sydney and surrounding areas for present day and projected future conditions in collaboration with the Upper Parramatta River Catchment Trust and its partners. The other objectives of the project are:

• Assess the impact of decadal-scale climate fluctuations on rainfall frequency characteristics (subsidiary). • Obtain a “broad brush” understanding of the likely changes in average and extreme rainfall under enhanced greenhouse conditions (core). • Quantify the likely future changes to the rainfall frequency characteristics of the study area due to global warming (core).

2 Study Area The study area has a homogeneous climate roughly bounded by longitudes 149.5º to 153º E; and latitudes 31.5º to 36ºS. The focus is on the coastal drainage areas of this region only (i.e., the SW corner of the region defined above would be ignored completely). The river basins that could be included in the study are shown in Fig. 1:

Figure 1. Location of study area.

3 Scope of the Report

The proposed work plan consists of four components couched within a three-year timeframe. The objectives of Components 1 to 3 are to provide information on rainfall IFD and depth-area curves for durations less than 6 hr for present day and projected future conditions, respectively. This information is critical for flood design applications.

Component 1 will develop a statistical model of extreme rainfall events for the study area, and will provide a list of candidate extreme events that can be explored using high resolution downscaling. The main task undertaken in Component 1 is the statistical analysis of observations at daily rain gauges and recording rain gauges (pluviometers) within and adjacent to the study area. The following steps are involved: 1.1 Clean rainfall data set prepared, 1.2 Spatial model(s) developed and applied to available data for durations of less than 1 hour to 72 hours, 1.3 Initial characterisation of IFD and depth-area curves for present day conditions, 1.4 Comparison of IFD and depth-area curves with those obtained from high resolution dynamical downscaling.

Step 1.1 is a major activity in terms of length of time involved. Section 4 of this report is concerned with the progress made with Step 1.1.

Component 4 of the project is the high resolution dynamical downscaling of extreme rainfall events for the current climate and the climates of 2030 and 2070. This component is being undertaken as a joint project with the AGO. The tasks are: 4.1 Identification of candidate cases, 4.2 High-resolution simulation of extreme rainfall events at a grid spacing of approximately 5km, 4.3 Analysis of downscaled events and preparation of IFD and depth-area curves for durations of less than 1 hour to 72 hours.

Some results from 4.1 are presented in Section 5.

4 Description of Rainfall data Data mainly from the Bureau of Meteorology’s (BOM) rainfall network is used in this project. This network includes over 6000 stations nation-wide, all of which record daily rainfall using standardised equipment and observing protocols. The observations are made at 0900 hours each day, with the 24 hour rainfall total being recorded against the day of observation. Mostly these gauges are operated by volunteers, often at workplaces like post offices, local government offices and farms. As a consequence, the quality of the data is quite variable, even for different time periods at a single station. Some of the data quality issues have been discussed by Lavery et al. (1992) and Viney and Bates (2004). The data quality is affected by a number of reasons including observer’s inconsistencies and exposure changes (changes in the height or structure of the gauge; changes in the windfield associated with growing trees or the construction of nearby buildings). Missing observations in the rainfall records arise from two main sources. The first appears to relate to communication and data management issues. The second source of missing data occurs when observers are absent from the station or otherwise unable to observe the gauge for a period of one or more days. When the observer returns to the gauge it contains rainwater that potentially fell over a period of two or more days. In these circumstances the observer records the accumulation period as well as the rainfall amount, which is entered against the date of observation. Viney and Bates (2004) have discussed the issues related to these “untagged accumulations”.

4.1 Data selection and screening The analysis of rainfall extremes requires rainfall records of high quality. Because we are considering temporal trends in extreme behaviour, we require long records whose quality is consistent in time. Because we are assessing spatial patterns, we require this consistency to extend from site to site. The geographical domain chosen for the rainfall data set encompasses the Upper Paramatta River (UPR) catchment (Figure 1) and adjacent areas within about 50 km of the catchment boundary. In total, 1937 rainfall stations are located within this region. However, many of these stations have quite short or discontinuous records, while others have been decommissioned altogether. This project requires reasonably lengthy and reasonably complete records. Unfortunately, the further back in time one looks, the fewer complete (or nearly complete) records are available, so compromises are needed between completeness and length. It was decided to accept only those gauges that had operated continuously (with allowances for short discontinuities) for more than 50 years until 2003. There are 265 BOM stations that have greater than 50 years of data with less than 10% data missing. Out of these 265 stations 169 stations have records until 2003. They are listed in Table 1 and depicted in Figure 2. This set of stations provides reasonable coverage of the UPR catchment with some gaps in some areas such as the Lithgow region, mountainous regions west of Taree, and areas west of Ulladulla in Morton National Park (Fig. 2).

Out of 169 stations, the following six are the automating weather stations with pluviometer data:

66137 Bankstown Airport AWS 70014 Canberra Airport 61087 Gosford (Narara Research Station) AWS 61250 Paterson (Total AWS) 66062 Sydney (Observatory Hill) 66037 Sydney Airport AMO

We have not analysed rainfall data in detail from these BOM stations, however we have looked at the patched point data (PPD) in more detail. The following section describes the data analysis from the PPD.

4.2 The Patched Point Data Set The PPD is a dataset containing daily rainfall, minimum and maximum temperatures, radiation, evaporation and vapour pressure. It combines original Australian BOM measurements for a particular meteorological station with infilling of any gaps in the record using interpolation methods (see http://www.nrm.qld.gov.au/silo/ppd for further details). In the UPR catchment region there are more than 360 stations with patched data. All these stations have complete set of rainfall and other data either directly observed or interpolated from nearby stations. Out of 169 BOM stations listed in Table 1 only 127 stations are in the PPD set. A preliminary screening of the data from PDP stations showed that 54 stations have fewer than 200 days of missing observations altogether between 1950 and 2003. 41 of these stations also belong to the list of 169 Bureau of Meteorology stations given in Table 1.

4.3 Future Tasks

We will acquire data from remaining 128 (169-41) BOM stations and carry out a detailed analysis to further examine the data quality. The data will be scrutinised for gaps and only those with an acceptable amount of missing data will be used for further processing. We will fill the any gaps in the observed data. The data will also be examined for any trends and inconsistencies. Underlying trends of any station will be compared with other nearby stations to detect if the trends are widespread or localised. The localised trend can be the result of local change in observation practices or setting of the stations. These can be detected by performing the double mass curve analysis.

Table 1 List of rainfall stations with greater than 50 years of rainfall data with less than 10% missing. Percent Years of data Station Start Finish of available No. Station name Longitude Latitude year year record data 55006 Blackville Post Office 150.2350 -31.6414 1886 2004 114.2 96 55014 Curlewis Post Office 150.2682 -31.1168 1904 2004 93.8 94 55017 Premer (Eden Moor) 149.7762 -31.5711 1887 2004 106.7 91 55018 Mullaley (Garrawilla) 149.6456 -31.1711 1884 2004 109.4 91 55024 Gunnedah Scs 150.2687 -31.0261 1948 2004 56.7 100 55025 Willow Tree (Highlands) 150.6730 -31.7928 1915 2004 84.1 94 55036 Caroona (West Mooki) 150.4226 -31.4140 1925 2004 72.4 92 55037 Pine Ridge (Mooki Springs) 150.3986 -31.5077 1886 2004 111.4 94 55038 Mullaley Post Office 149.9114 -31.0976 1899 2004 97.6 93 55039 Spring Ridge (Murrumbah) 150.2090 -31.3753 1922 2004 82.6 100 55043 Willow Tree (Parraweena) 150.4135 -31.7118 1934 2004 67.7 96 55045 Curlewis (Pine Cliff) 150.0270 -31.1804 1903 2004 101.4 100 55046 Pine Ridge (Billabong) 150.4296 -31.5271 1921 2004 76.3 92 55049 Quirindi Post Office 150.6792 -31.5086 1882 2004 118.4 96 55057 Willow Tree (Valais) 150.2856 -31.7731 1881 2004 122.4 99 55062 Werris Creek Post Office 150.6480 -31.3490 1889 2004 108.7 94 55064 Pine Ridge (Windy) 150.3811 -31.6001 1915 2004 88.3 100 55066 Wallabadah (Woodton) 150.8437 -31.6218 1892 2004 106.6 95 55067 Goonoo Goonoo Station 150.9069 -31.3038 1873 2004 127.9 98 55078 Nundle (Benoni) 151.0896 -31.4149 1953 2004 50.8 98 55194 Gowrie North 150.8537 -31.3365 1953 2004 51.1 99 55244 Willow Tree (Cooinda) 150.5736 -31.6352 1950 2004 51.5 94 56077 Walcha (Craigdarroch) 151.5024 -31.0673 1953 2004 50.4 98 56082 Niangala (Ashland) 151.4232 -31.2424 1952 2004 50.9 97 57037 Tia (Rambrah) 151.8014 -31.1117 1953 2004 50.4 98 57050 Brackendale (The Forest) 151.6744 -31.2318 1950 2004 52.9 97 59017 Kempsey (Wide Street) 152.8235 -31.0770 1882 2004 121.4 99 60002 Bulahdelah Post Office 152.2079 -32.4129 1905 2004 94.3 95 60005 Comboyne Post Office 152.4685 -31.6053 1905 2004 97.3 98 60013 Forster Beach Caravan Park 152.5088 -32.1772 1896 2004 105.2 97 60015 Gloucester Post Office 151.9597 -32.0063 1888 2004 109.8 94 Hannam Vale (Hannam Vale 60017 Road) 152.5830 -31.6994 1926 2004 76.2 97 60020 Kendall Post Office 152.7047 -31.6321 1899 2004 100.1 95 60021 Krambach Post Office 152.2598 -32.0511 1910 2004 92.3 98 60023 Harrington (Bowling Club) 152.6828 -31.8817 1887 2003 111.5 96 60027 Lorne (Lorne Road) 152.6146 -31.6561 1938 2004 64.9 97 60030 Taree (Radio Station 2re) 152.4834 -31.8986 1881 2004 118.9 97 60033 Krambach - Bellevue 152.2463 -32.0924 1908 2004 90.8 94 60035 Wauchope (Avondale Street) 152.7315 -31.4595 1890 2004 113.1 99 60036 Wingham (Lanark Close) 152.3444 -31.8620 1888 2004 114.4 99 61002 Blackville (Krui Vale) 150.3410 -31.8449 1885 2004 118.6 99 61007 Bunnan (Milhaven) 150.5836 -32.0332 1900 2004 101.9 98 61010 Clarence Town (Grey St) 151.7813 -32.5879 1895 2004 108.2 99 Cockle Creek (Pasminco 61011 Metals) 151.6267 -32.9456 1900 2003 101.5 98 61016 Denman (Palace Street) 150.6886 -32.3885 1883 2004 120.4 99 61017 Dungog Post Office 151.7582 -32.4022 1897 2004 101.4 95 61024 Gresford Post Office 151.5376 -32.4266 1895 2004 107.2 98 61026 Gundy (Miller St) 150.9971 -32.0119 1887 2004 107.5 92 61031 Raymond Terrace (Kinross) 151.7350 -32.7756 1894 2004 110.2 100 61040 Merriwa (Gummun Place) 150.3577 -32.1388 1882 2004 118.2 97 61046 Morpeth Post Office 151.6285 -32.7254 1884 2004 115.5 96 61050 Sedgefield (Bundajon) 151.2864 -32.4997 1903 2004 98.7 97 61051 Murrurundi Post Office 150.8362 -31.7631 1870 2004 130.9 98 61056 Pokolbin (Ben Ean) 151.2781 -32.7962 1905 2003 92.1 94 61065 Aberdeen (Rossgole) 150.7285 -32.1402 1926 2004 78.6 100 61071 Stroud Post Office 151.9664 -32.4027 1889 2004 112.2 97 61072 (Carrington House) 152.0140 -32.6678 1887 2004 116.8 99 61074 The Entrance (Eloora Street) 151.4956 -33.3544 1943 2004 56.6 92 61075 Merriwa (Tunbridge) 150.2366 -32.2229 1926 2003 76.3 98 Raymond Terrace (Wallaroo 61076 State Forest) 151.8877 -32.6180 1938 2004 61.6 93 61079 Wingen (Murrulla) 150.8809 -31.8679 1877 2004 125.2 98 61082 Wyee Post Office 151.4847 -33.1753 1899 2004 105.3 100 Wyong (Wyong Bowling 61083 Club) 151.4267 -33.2931 1885 2004 114.5 96 61086 Jerrys Plains Post Office 150.9093 -32.4972 1884 2004 118.7 99 Gosford (Narara Research 61087 Station) Aws 151.3290 -33.3949 1917 2004 82.9 95 61089 Scone Scs 150.9272 -32.0632 1950 2004 54.6 100 61093 Ourimbah (Dog Trap Road) 151.3264 -33.3653 1953 2004 50.8 99 61095 (Albano) 151.0932 -32.1939 1932 2004 70.9 99 61096 Paterson Post Office 151.6167 -32.6011 1901 2004 96.4 93 61151 Chichester 151.6830 -32.2426 1942 2004 62.1 99 62003 Mumbil () 149.1006 -32.6673 1951 2004 51.5 96 62005 Cassilis Post Office 149.9800 -32.0067 1870 2004 126.5 94 62009 Cassilis (Dalkeith) 149.9857 -31.9963 1874 2004 121.6 93 62014 Hargraves (The Elders) 149.4720 -32.8016 1913 2004 87.2 95 62015 Merriwa (Merry Vale) 150.2230 -31.9273 1890 2004 113.6 99 62017 Kandos Cement Works 149.9745 -32.8647 1951 2004 53.3 99 62021 Mudgee (George Street) 149.5956 -32.5956 1870 2004 132.7 99 62026 Rylstone (Ilford Rd) 149.9768 -32.8073 1881 2004 114.6 93 62029 Ilford (Tara) 149.8100 -32.9809 1928 2003 75.3 100 62032 Wollar (Barrigan St) 149.9484 -32.3592 1901 2004 101.3 98 Abercrombie (Abercrombie 63000 Bridge) 149.3252 -33.9541 1931 2004 71.4 98 63005 Bathurst Agricultural Station 149.5559 -33.4289 1908 2004 95.6 99 63009 Blackheath (Godson Ave) 150.3023 -33.6199 1898 2004 101.5 95 63030 Oberon (Gingkin) 149.9373 -33.8888 1938 2003 58.3 91 63032 Golspie (Ayrston) 149.6609 -34.2734 1897 2004 106.2 99 63035 Hill End Post Office 149.4146 -33.0362 1880 2004 121.1 97 63039 Katoomba (Narrow Neck Rd) 150.2983 -33.7135 1885 2004 117.9 99 Kurrajong Heights (Bells Line 63043 Of Road) 150.6338 -33.5343 1866 2004 125.5 91 63044 Lawson (Wilson Street) 150.4316 -33.7249 1895 2004 104.1 95 63053 Millthorpe (Inala) 149.1847 -33.4455 1899 2004 101.3 96 63064 O'connell (Stratford) 149.7276 -33.5333 1889 2004 112.7 97 63066 Orange (Mclaughlin St) 149.1110 -33.2741 1889 2004 112.2 98 63076 Sofala Old Post Office 149.6901 -33.0807 1892 2004 112.5 100 63077 Springwood Bowling Club 150.5694 -33.6984 1883 2004 114.9 94 63079 Sunny Corner (Snow Line) 149.9013 -33.3947 1903 2004 99.5 98 63080 Black Springs (Swatchfield) 149.7103 -33.8905 1937 2004 66.9 99 Trunkey Creek (Trunkey 63083 (Black Stump Hote 149.3244 -33.8193 1890 2003 104.2 92 63087 Black Springs Forestry 149.7396 -33.8463 1940 2004 58.3 91 63094 Bigga (Woolbrook) 149.1223 -34.0058 1890 2004 113.7 99 63118 Bilpin (Fern Grove) 150.4892 -33.5156 1895 2004 109.5 100 63271 Tuena (Wyoming) 149.3611 -34.0417 1951 2004 53.7 100 64008 Coonabarabran (Namoi St) 149.2714 -31.2712 1879 2004 125.4 100 64009 Dunedoo Post Office 149.3953 -32.0163 1912 2004 91.9 99 64013 Binnaway (Hawthorne) 149.4270 -31.6417 1886 2004 116.3 98 64015 Mendooran Post Office 149.1206 -31.8236 1886 2004 110.5 93 64025 Coolah (Binnia St) 149.7200 -31.8231 1886 2004 114.3 96 66000 Ashfield Bowling Club 151.1344 -33.8850 1894 2003 99.1 91 66004 Bexley Bowling Club 151.1083 -33.9444 1931 2004 68.9 94 66014 Cronulla South Bowling Club 151.1512 -34.0703 1942 2004 58.8 94 Five Dock (Barnwell Park 66017 Golf Course) 151.1190 -33.8681 1938 2003 62.3 94 66037 Sydney Airport Amo 151.1725 -33.9411 1929 2004 74.8 100 66040 Miranda (Blackwood St) 151.0982 -34.0405 1906 2003 96.4 99 66042 Mosman (Bapaume Road) 151.2428 -33.8194 1895 2004 107.3 98 66045 Newport Bowling Club 151.3187 -33.6572 1931 2004 70.9 97 66058 Sans Souci (The Boulevard) 151.1267 -33.9967 1899 2004 104.1 99 66062 Sydney (Observatory Hill) 151.2050 -33.8607 1858 2004 146.2 100 66157 Pymble (Canisius College) 151.1521 -33.7371 1947 2004 56.7 100 66160 Centennial Park 151.2341 -33.8959 1900 2004 101.8 98 67015 Bringelly (Maryland) 150.7250 -33.9696 1867 2003 132.4 97 67019 Prospect Dam 150.9127 -33.8193 1887 2004 117.5 100 67021 Richmond - Uws Hawkesbury 150.7477 -33.6165 1881 2004 120.3 97 67026 Seven Hills (Collins St) 150.9318 -33.7704 1950 2004 51.2 94 67029 Wallacia Post Office 150.6411 -33.8637 1943 2004 59.4 97 67031 Windsor Bowling Club 150.8152 -33.6100 1897 2004 101.2 94 West Pennant Hills Oratava 67098 Ave 151.0433 -33.7506 1943 2004 58.5 95 68000 Albion Park Post Office 150.7761 -34.5712 1892 2004 102.8 91 68003 Berry Masonic Village 150.6917 -34.7810 1886 2004 112.7 95 68007 Camden (Brownlow Hill) 150.6455 -34.0254 1882 2004 120.6 99 68016 150.8062 -34.2644 1904 2004 97.1 96 68024 Darkes Forest (Kintyre) 150.9111 -34.2289 1894 2004 107.4 97 Gerringong (Mayflower 68027 Village) 150.8211 -34.7472 1895 2004 105.1 96 68028 Helensburgh (Parkes Street) 150.9738 -34.1967 1889 2004 110.1 96 68030 Mittagong (High Range) 150.2999 -34.3738 1945 2004 58.3 98 Jervis Bay (Point 68034 Perpendicular Lighthou 150.8048 -35.0936 1899 2004 104.8 100 68036 Kangaroo Valley (Main Rd) 150.5332 -34.7358 1914 2003 80.8 91 68038 Kiama Bowling Club 150.8519 -34.6750 1897 2003 102.5 96 68045 Moss Vale (Hoskins Street) 150.3768 -34.5444 1870 2004 129.8 97 68048 Nowra Treatment Works 150.6176 -34.8718 1896 2004 106.7 99 68052 Picton Council Depot 150.6145 -34.1685 1880 2004 114.1 92 68062 High Range (Wanganderry) 150.2653 -34.3447 1945 2004 58.7 98 68108 Woonona Popes Rd 150.8999 -34.3420 1929 2004 72.1 96 68204 Sussex Inlet Bowling Club 150.5908 -35.1702 1952 2004 50.7 96 69006 Bettowynd (Condry) 149.7896 -35.7011 1897 2003 105.9 99 69010 Braidwood (Wallace Street) 149.7990 -35.4489 1887 2004 113.6 97 69018 Moruya Heads Pilot Station 150.1532 -35.9093 1875 2004 129.1 100 69052 Batemans Bay - Buckenbowra 150.0508 -35.7343 1943 2004 60.9 99 70000 Ainslie Tyson St 149.1424 -35.2583 1935 2004 68.3 98 70002 Bannaby (Hillasmount) 149.9957 -34.4330 1945 2004 58.2 97 70011 Bungendore Post Office 149.4446 -35.2549 1890 2004 110.5 97 70012 Bungonia (Inverary Park) 149.9709 -34.8996 1883 2004 112.2 92 70014 Canberra Airport 149.2014 -35.3049 1939 2004 65.6 100 70021 Collector (Brookdale) 149.4333 -34.9118 1891 2004 106.8 94 70025 Crookwell Post Office 149.4693 -34.4578 1883 2004 117.2 97 70035 Bungendore (Gidleigh) 149.4727 -35.3083 1886 2004 118.5 100 70036 Lake Bathurst (Gilmour) 149.6582 -35.0131 1931 2004 72.8 100 70043 Gunning Rural Supplies 149.2681 -34.7823 1886 2004 114.9 97 70045 Hall (Lochleigh) 149.0570 -35.1548 1903 2004 100.8 100 70055 Goulburn (Kippilaw) 149.5951 -34.7503 1886 2004 110.5 93 70057 Braidwood (Krawarree) 149.6331 -35.8238 1898 2004 103.5 97 70060 Lower Boro (Attaweenah) 149.7875 -35.1810 1910 2004 93.1 99 70064 Michelago (Soglio) 149.1602 -35.6795 1884 2004 118.5 99 70069 Crookwell (Gundowringa) 149.5740 -34.5438 1945 2004 59.4 100 70071 Goulburn (Pomeroy) 149.5028 -34.6516 1901 2004 98.5 95 70072 Queanbeyan Bowling Club 149.2292 -35.3552 1870 2004 131.6 98 70080 Taralga Post Office 149.8197 -34.4048 1882 2004 115.8 95 70082 Taylors Flat (Callaba) 149.0084 -34.2815 1926 2004 72.2 92 70088 Yarra (Rowe's Lagoon) 149.5201 -34.8948 1896 2004 101.7 94 70111 Biala (Alvison) 149.2497 -34.5612 1938 2004 65.9 99 70119 Big Hill (Glen Dusk) 149.9970 -34.5679 1944 2003 58.2 98

Upper Parramatta River Catchment Rainfall Station -30

-31

-32

-33 e d u t i

Lat -34

-35

-36

-37 148 149 150 151 152 153 154 Longitude

Figure 2: Locations of rainfall stations in the study region. 5 Analysis of extreme rainfall events in the climate model

The daily rainfall output from CSIRO Atmospheric Research Cubic-Conformal Atmospheric Model (CCAM) nested in the Mark 3 Global Climate Model has been analysed to identify extreme rainfall events occurring in the study region and to evaluate the skill of the model in representing the synoptic situations associated with observed present-day extreme precipitation events.

5.1 Observed Extreme Rainfall Bureau of Meteorology rain gauges located east of 149°E between 32° and 37°S have been used to identify extreme rainfall events. The daily rainfall record based on these gauges is for the period 9:00 am on the preceding day to 9:00 am on the day of the record. On a site-by-site basis the daily rainfall record is variable both in length and quality. The stations considered in this analysis were required to have a record length of at least 15 years between 1960 and 2000, and that record was required to be at least 80% complete as identified by the “quality flag” for the station.

Daily rainfall records for stations that meet these criteria were used to create station time series for 1-day 3-day totals. Station subsets were created for 1-day rainfall totals exceeding 85mm and 3-day totals exceeding 225 mm. A total of 637 1-day events that satisfied this criterion were identified over the forty-year period. 300 3-day events were identified.

The focus of this component of the study is on the large-scale extreme rainfall events that result in riverine flooding, rather than the localised events associated with flash flooding. The 1-day and 3-day rainfall subsets were further analysed to identify days which contributed to 3-day totals in excess of 225 mm and in which at least 85 mm of rainfall occurred in the daily record. This analysis produced 321 days, approximately half of the 1-day subset. These days were further analysed and shown to contribute to 66 events in which heavy rainfall occurred on consecutive days. These 321 days are the focus of the present study. The remaining days are single day events, or have contributed to a lighter intensity multi-day event and are not considered here.

The pressure patterns associated with these heavy rainfall days have been analysed to determine the synoptic-scale weather patterns that are conducive to extreme rainfall in the study region. The technique used is known as synoptic typing and follows the method of Yarnal (1993). This is a correlation-based, gridded map-typing technique in which days are grouped based on the Pearson product-moment correlations to establish the degree of similarity between map pairs. Similar fields are identified on the basis of similar spatial structures (i.e. highs and lows in similar positions) with little emphasis on the magnitude of the patterns.

To establish a synoptic climatology compatible with the CCAM output, this technique was first applied to NCEP 00UTC MSLP fields valid for the extreme rainfall days for the study region. These fields were extracted for the 81 points (9×9) in the region between 145 and 165E and 45 and 25S. These fields were further divided into summer (Nov-Apr) and winter (May-Oct) series and the synoptic typing performed on both the summer and winter datasets.

After the typing procedure was completed for the NCEP extreme rainfall dataset, the “key days” produced were used to type the gridded MSLP fields for “extreme” rainfall days extracted from CCAM for the control, 2030 and 2070 climates. The correlation-based classification procedure of the NCEP gridded pressure data identified 5 main synoptic patterns associated with extreme summer rainfall and 6 patterns for extreme winter rainfall.

Summer Figure 3 shows the composite pressure patterns for each of the five summer types. The composite patterns have been obtained by averaging the MSLP patterns for the days that contribute to each synoptic type.

S1 S2

S3 S4

S5

Figure 3: Composited MSLP fields for the synoptic types associated with extreme summer rainfall along the central coast of NSW. Type 1 accounts for 59% of events, type 2 for 14%, type 3 for 10%, type 4 for 3% and type5 for 3% . 11% of summer days are unclassified.

These five synoptic types account for 89% of days considered. The remaining days were unclassified.

Types 1 and 2 are both characterised by a high-pressure system to the south of the continent that acts to produce a large area of onshore south-easterly to north-easterly flow into the study region. In both cases there is a trough in the easterly flow. In the Type 3 cases the high-pressure system has moved further into the Tasman Sea resulting in the transport of warm moist onshore flow into the region. This onshore flow originates over the Coral Sea. The Type 4 pattern is similar to that of Types 1 and 2 showing a tough in the easterly flow. The main feature of the Type 5 pattern is a cut-off low in the Tasman Sea. This pattern is mostly representative of a translating low-pressure system, such as a tropical cyclone or monsoonal depression and is accompanied by an extensive region of very high moisture contents and a closed circulation at 700 hPa. The Type 5 pattern shows a high-pressure system to the south of the continent that produces south-easterly flow into the study region. The Type 6 pattern has a low-pressure system in the Southern Ocean with a trough in the Tasman Sea that produces northeasterly flow in the study region.

Winter Figure 4 shows the composite pressure patterns for each of the 6 winter types.

W1 W2

W3 W4

W5 W6

Figure 4: Composited MSLP fields for the synoptic types associated with extreme winter rainfall in along the Central Coast of NSW. Type 1 accounts for 29% of events, type 2 for 17%, type 3 for 15%, type 4 for 11%, type 5 for 10% and type 6 for 5% . 13% of events are unclassified.

The Type 1 and Type 2 patterns are similar, both characterised by a high-pressure system to the south east of the continent and a deep trough that acts to produce a strong north-easterly to easterly flow into the study region

The main features of the Type 3 and Type 4 patterns are a low in the Tasman Sea. Most cases that contribute to this pattern are mature east coast lows that have moved southwards along the coastline. A small number of events are associated with a mid-latitude low that deepens in the Tasman Sea.

5.2 Extreme Rainfall CCAM

A similar analysis has been performed on the extreme rainfall events selected from CCAM. This analysis has been performed for extreme rainfall events for 3 forty-year climate periods – control, 2030 and 2070. The control climate period is representative of the climate of 1961-2000, 2030 represents 2011-2050 and 2070 represents 2051- 2090. Extreme rainfall events selected from CCAM for this study were chosen to reflect the climatology of the observed heavy rainfall events. This was achieved by determining the number of 1-day and 3-day rainfall events that occurred in each of 4 3-month seasons. Thus, 637 1-day and 300 3-day extreme rainfall events were chosen from the CCAM model and the overlap days identified for further analysis. The synoptic patterns associated with the CCAM summer and winter extreme rainfall days were then determined for the control, 2030 and 2070 climates. The MSLP pressure pattern for each extreme rainfall day was first interpolated from the 0.5°× 0.5° (approximately) CCAM grid to the 2.5°× 2.5° NCEP grid. Each CCAM MSLP grid was correlated with the MSLP grid for the “key days” identified in the synoptic typing of the observational dataset. The CCAM day was classified into the synoptic type with the highest correlation above the threshold of 0.7. This analysis provides a measure of the number of CCAM extreme rainfall days that can be classified according to the observed synoptic types. The results from this analysis are summarised in Tables 1 and 2.

Table 1: The number of summer rain days corresponding to each synoptic class. Values are expressed as a percentage of the total number of rain days selected for the model interval.

CCAM CCAM CCAM Type Description Observations Control 2030 2070 Tasman High 1 59 57 55 44 (coastal northeasterlies) Bass Strait High 2 13 6 17 17 (coastal southeasterlies) Tasman High 3 10 17 13 17 (coastal northerlies) 4 Trough in Tasman Sea 3 4 6 6

5 Cut-off low 3 4 0 0 Other/Unclassified 11 10 8 16

For the summer half year, CCAM has captured the distribution synoptic pattern associated with extreme rainfall along the Central Coast of NSW. However, it should be noted that the comparison with observations was made on fewer days than used in the synoptic typing of the observations. In the observations, approximately 50% of 1- day extreme rainfall events contributed to a 3-day event, while in CCAM this figure was only 18%. Thus, approximately 1/3 of the number of days was available for classification, when compared to the observations.

Table 2: The number of winter rain days corresponding to each synoptic class. Values are expressed as a percentage of the total number of rain days selected for the model interval.

CCAM CCAM CCAM Type Description Observations Control 2030 2070 Tasman High 1 (offshore 29 5 5 0 depression/trough) Tasman High 2 (offshore 17 11 24 23 depression/trough) 3 Low in Tasman Sea 15 0 0 0

4 Coastal Low 11 9 5 14 High over Tasmania 5 10 5 0 6 (coastal southeasterlies) Trough in Tasman 6 5 12 24 17 (coastal northeasterlies) Other/Unclassified 12 58 50 40

The control climate of CCAM contains fewer heavy rainfall days that are due to the 6 standard patterns identified for the observations for the winter case and approximately half of the heavy rainfall days were unable to be classified into the standard types. The dominant pattern in CCAM winter rainfall is Type 2, followed by Type 6. As with the summer case, approximately 1/3 of the number of days was available for classification, when compared to the observations.

5.3 Current and future tasks

Analysis of extreme rainfall days is continuing. Specifically, the analysis is being repeated for all extreme 1-day events, rather than the subset. In addition, the outputs from the climate model are also being typed independently of the observations.

Preparation of the CCAM output for the high resolution modelling is currently in progress. The high resolution modelling tasks are expected to commence by the end of November.

6 References

Lavery, B., Kariko, A., Nicholls, N., 1992: A historical rainfall dataset for . Aust. Meteorol. Mag., 40, 33-39.

Yarnal, B., 1993: Synoptic Climatology in Environmental Analysis: A Primer. Studies in Climatology Series, Belhaven Press, London. 195 pp.

Viney, N.R. and B.C. Bates, 2004: It Never Rains on Sunday: The Prevalence and Implictions of Untagged Multi-day Rainfall Acculations in the Australian High Quality Data Set, Int. J. Climatol., 24, 1171-1192.