Objective Classification of Precipitating Convective Regimes Using a Weather Radar in Darwin, Australia

Transcription

1 MAY 2009 C A I N E E T A L Objective Classification of Precipitating Convective Regimes Using a Weather Raar in Darwin, Australia SIMON CAINE Monash University, Melbourne, Victoria, Australia CHRISTIAN JAKOB Bureau of Meteorology Research Centre, Melbourne, Victoria, Australia STEVEN SIEMS Monash University, Melbourne, Victoria, Australia PETER MAY Bureau of Meteorology Research Centre, Melbourne, Victoria, Australia (Manuscript receive 17 January 2008, in final form 16 September 2008) ABSTRACT A clustering algorithm was applie to Frequency with Altitue Diagrams (FADs) erive from 4 yr of hourly raar ata to objectively efine four tropical precipitation regimes that occur uring the wet season over Darwin Australia. The precipitation regimes efine are istinguishe in terms of convective intensity, presence of stratiform precipitation, an precipitation coverage. Regime 1 consists of patchy convection of meium intensity an low area coverage, an regime 2 contains strong convection with relatively small area coverage. Regime 3 is compose of weak convection with large area coverage an large stratiform regions, an regime 4 contains strong convection with large area coverage an large stratiform regions. Analysis of the seasonal cycle, iurnal cycle, an regime occurrence as a function of monsoon activity all provie insight into the ifferent physical character of the precipitation regimes. Two of the regimes exhibit a iurnal cycle with a peak in the afternoon, while the other two show a peak in their frequency of occurrence in the early morning. The ifferent character of the regimes is also confirme by the varying contributions that convective an stratiform rainfall make to the overall within-regime precipitation. 1. Introuction Tropical rainfall is largely controlle by convective processes. However, as has been shown in numerous stuies (e.g., Orlanski 1975; Rickenbach an Rutlege 1998; Nesbitt et al. 2000, 2006; Houze et al. 1981; Houze an Churchill 1984; Houze 1987; Mapes an Houze 1993; Mapes 1993; Houze et al. 1980), these processes are often organize on a variety of scales, that can vary from the local to mesoscale (e.g., topographically riven convection) to the large scale in convectively couple Corresponing author aress: Simon Caine, School of Mathematical Sciences, Monash University, 3800 Melbourne, VIC, Australia. simon.caine@sci.monash.eu.au tropical waves (e.g., the Maen Julian oscillation). Because of convective organization, the spatial an temporal structure of tropical precipitation events is far more complex than that of a simple archetypal convective shower. Many convective systems exhibit a large stratiform component (Houze 1997, 1993), ientifiable both by its anvil clou signature an its spatially extensive but lower (in comparison to a convective shower) rain rates. The stratiform component of tropical rainfall constitutes a major fraction of the total both globally an regionally (Schumacher an Houze 2003) an its vertical structure of latent heat release is likely to play an important role in the interaction of convection with large-scale ynamical features of the tropical atmosphere (Lin et al. 2004; Schumacher et al. 2004; Mapes an Lin 2005; Tao et al. 2006). Recently, Mapes et al. DOI: /2008MWR Ó 2009 American Meteorological Society

2 1586 M O N T H L Y W E A T H E R R E V I E W VOLUME 137 (2006) have shown that the archetypal composition of precipitating tropical convective systems, consisting of a shallow an eep convective moe, as well as a stratiform moe, is observable over many scales. This raises the question, if convective systems in the tropics inee occur in istinguishable an repeately occurring regimes, what are the characteristics of those regimes an how are they linke to the well-known tropical circulation features? Jakob an Tseliouis (2003) an Rossow et al. (2005) use a simple clustering algorithm applie to ata from the International Satellite Clou Climatology Project (ISCCP; Schiffer an Rossow 1983), to show that tropical clouiness on a km 2 scale (the size of the ISCCP gri box) appears to be organize into six major regimes: three convectively active (an hence likely precipitating) an three suppresse. By their very nature the regimes ientifie in this fashion are base on the raiative signature of the tropical clou systems, an further work (Jakob et al. 2005) establishe that the ISCCP-base regimes show istinct signatures in their raiative an thermoynamic properties. However, because of a lack of suitable ata no conclusions on the precipitation characteristics of the ISCCP-base regimes coul be rawn. Nevertheless, base on the success of these stuies it is vali to ask the question if a similar regime character can be foun in the precipitation signature of tropical convection. An ieal tool to observe the spatial structure of convective clou systems is the centimeter wavelength raar. Previously Boccippio et al. (2005) use clustering analysis on vertical profiles erive from the Tropical Rainfall Measuring Mission (TRMM) precipitation raar to efine a number of archetypal vertical structures for the entire tropics. The profiles use by Boccippio et al. (2005) were essentially vertical columns with a horizontal iameter of 4 km, consequently very little spatial information entere into their regime efinition. Groun-base raars are routinely eploye by weather services worlwie (Demott an Rutlege 1998; Kawashima et al. 2006) for research an forecasting purposes. The Australian Bureau of Meteorology has operate a polarimetric research C-ban raar (C-Pol; Keenan et al. 1998; May et al. 1999; Carey an Rutlege 2000), in Darwin (Australia) since Darwin is locate in the tropical north of Australia an has been the site of numerous fiel stuies on tropical convection, such as the Australian Monsoon Experiment (AMEX; Hollan et al. 1986; Gunn et al. 1989; Keenan et al. 1989), the Marine Continental Thunerstorm Experiment (MCTEX; Keenan et al. 2000; Carbone et al. 2000), the Darwin Area Wave Experiment (DAWEX; Hamilton et al. 2004; Alexaner et al. 2004; Pautet et al. 2005), an most recently the Tropical Warm Pool International Clou Experiment (TWP-ICE; May et al. 2005). These experiments establishe that with its tropical monsoon climate, characterize by istinct wet, ry, an transition seasons, Darwin constitutes an ieal location to observe many of the variety of convective systems observe in the tropics (Keenan an Carbone 1992; Keenan an Rutlege 1993). During active monsoon spells, convection in the Darwin area has been shown to often resemble maritime convection foun over the tropical oceans (Steiner et al. 1995; Keenan an Carbone 1992). The often topographically riven convective systems observe uring both the transition seasons an in monsoon breaks on the other han are among the most intense observe on the planet (Crook 2001; Zipser et al. 2006) an have been shown to be goo examples of continental convection observe over the tropical continents. For these reasons the Atmospheric Raiation Measurement (ARM) program (Ackerman an Stokes 2003) establishe the thir Tropical Western Pacific (TWP) climate research facility in Darwin. The combination of the availability of many years of raar ata an the ability to observe a variety of convective systems makes Darwin an ieal location to stuy the possible regime character of tropical precipitation on scales of a few hunre kilometers. The aim of this paper is to etermine if such regimes can be objectively efine from raar ata an use to escribe the rainfall precipitation properties of convective systems occurring over long perios of time (e.g., seasons) for the region surrouning Darwin. To establish if tropical rainfall exhibits such regime character this stuy applies a cluster algorithm similar to that use in Jakob an Tseliouis (2003) an Rossow et al. (2005) to Frequency with Altitue iagrams (FADs; Yuter an Houze 1995; Neiman et al. 2005) of raar reflectivity erive from four wet seasons of raar ata. The properties of the regimes ientifie by this algorithm are investigate an their relationship to the main circulation regime in tropical Australia the monsoon is establishe. The relationship of the raar-base regimes to the ISCCP-base tropical clou regimes is also investigate. 2. Data escription an metho The ataset use in this stuy is compose of reflectivity values observe by the scanning C-ban polarimetric (C-Pol) raar locate near Darwin (Keenan et al. 1998). Figure 1 shows a map of the raar omain inicating lan areas (gray) an the maximum coverage of the raar footprint (achieve at 3 km in the vertical) use in this stuy. The raar performs a full volume scan every 10 min. Given the aim of stuying possible

3 MAY 2009 C A I N E E T A L FIG. 1. CAPPI showing the maximum coverage of the Gunn Point raar, near Darwin, Australia. CAPPI is vali at 3 km in the vertical. The lan is shae gray an the oceans are white. precipitation regimes as they occur over long perios of time it was necessary (because of computational limitations) to reuce the volume of information by using only one scan per hour. The hourly ataset use in this investigation is compose of the four wet seasons [November April (NDJFMA)] of 1999/2000, 2001/02, 2002/03, an 2003/04, bringing the total number of sample raar volumes to just over As a first step in calculating FADs (Yuter an Houze 1995) of raar reflectivity for each of the sample volumes, the volume scans were interpolate into a Cartesian space consisting of 40 vertical levels with a gri spacing of 500 m, an 121 gri points in each horizontal irection with a gri spacing of 2.5 km. The change in the horizontal extent of the raar omain with height is taken into account by applying masks of the actual observe area at each vertical level. The histograms forming the basis of the FADs were then create by iteratively searching through all vali reflectivity values at each vertical level an calculating the frequency with which these values fall into reflectivity bins of with 2 BZ. Reflectivity values below 0 BZ are exclue from the analysis as they are consiere at the etection limit of the raar, especially at longer ranges. At each vertical level the relative frequency of occurrence of each reflectivity bin was then calculate by iviing the number in each bin by the total number of observe points efine by the raar mask at that level. Consequently, summation over the frequencies for each vertical level can be thought of as a measure for the fractional coverage with raar signal (or precipitation) at that level. This eliberately eviates from the stanar normalization of Contoure Frequency with Altitue Diagrams (CFADs, the graphical representation of FADs; Yuter an Houze 1995), where each level is normalize by its fractional coverage. This choice was mae to maintain area coverage information in the regime analysis that follows, as this provies a crucial istinction between regimes. For illustrative purposes Fig. 2 shows the mean FAD over all samples use in this stuy. An important summary measure use throughout the stuy is the total volume coverage (TVC) with raar echo. This quantity is calculate by summing the number of raar returns over all levels an reflectivity values then iviing by the total number of observable points. It therefore represents the fractional coverage of the raar signal over the entire three-imensional omain. The TVC for the mean histogram is 0.04, inicating that on average approximately 4% of the volume scanne by the raar has a reflectivity value above the 0-BZ threshol. It can also be seen that, on average, the maximum echotop height reaches approximately 16 km an reflectivity values greater than 40 BZ are observe up to 8 km. A large fraction of the returns have reflectivity values less than 18 BZ. As this is the cutoff threshol for the TRMM raar, Fig. 2 gives some inication of the number of hyrometeors the TRMM raar will not observe. Each of the more than histograms provies an input into the efinition of objectively erive regimes. To focus the clustering algorithm (see below) onto the major regimes, first all (trivial) null cases (i.e., those with little or no raar return anywhere in the volume) were remove from the analysis. For this purpose any histogram that i not have at least 1% coverage in at least one of the vertical levels was efine as a noprecipitation histogram an was remove. This eliminate roughly 5000 histograms from the analysis. While these no-precipitation cases o not enter the regime efinition algorithm, they o nevertheless provie important physical information about the state of the atmosphere. Their high frequency of occurrence an their trivial nature justifies their removal to avoi them ominating the outcome of the cluster algorithm. A K-means clustering algorithm (Anerberg 1973) was applie to the remaining histograms to ientify possible recurring patterns in the FADs that coul be inicative of recurring precipitation regimes. The choice of algorithm is largely riven by its successful application in previous stuies (Jakob an Tseliouis 2003) an efficiency requirements ue to the large sample size use here. As iscusse in Jakob an Tseliouis (2003) an Rossow et al. (2005) a feature of this particular algorithm is that it requires the user to preefine the

4 1588 M O N T H L Y W E A T H E R R E V I E W VOLUME 137 FIG. 2. FAD showing the mean of all time perios use in the stuy ( histograms). The contours show bin counts ivie by total observe gri points. The TVC is shown at the top of the iagram, inicating that on average 4% of the volume scanne by the raar is covere by hyrometeors. number of clusters searche for (i.e., k). This is overcome by repeating the analysis several times with varying values of k. Here 2 # k # 10 are use. The basic concept of the iterative K-means algorithm is as follows: ranomly choose k see vectors from the ataset as initial cluster centrois, assign each input histogram to one cluster by fining the minimum Eucliean istance to any of the cluster centrois, calculate new cluster centrois by averaging over all histograms assigne to a cluster, use the newly calculate centrois as see vectors, an iterate the algorithm. With each iteration the ifference between the ol an new centrois ecreases an the clusters are sai to be stable when this ifference is less than a prescribe threshol value. The histograms associate with each cluster are known to be closer to the centroi of this cluster than to those of any of the others. Each regime is then characterize by its centroi or mean histogram. The regimes efine by the clustering algorithm shoul not epen on initial histograms chosen at ranom an tests were performe to ensure this was the case. The avantage of using the K-means algorithm is that it provies an objective metho for efining regimes from large sets of ata. However, as iscusse above, the number of clusters, k, searche for in the algorithm nees to be prescribe. It is therefore necessary to efine a strategy for choosing the optimal set of clusters from the application of the algorithm to values of 2 # k # 10. Here the same quasi-objective metho as in Rossow et al. (2005) is use. For each increase in cluster number the stability of the solution to the (ranomly chosen) initial sees is investigate. Furthermore, the emerging new cluster in the k 1 1 analysis is compare to the existing k clusters of the previous application. The optimal number of regimes in this stuy, K, is then efine as the smallest possible number of clusters for which (i) the algorithm provies stable solutions with respect to initial sees an (ii) the aition of further clusters leas to regimes similar to the alreay existing ones. Here it is foun that K 5 4. Note, that since noprecipitation histograms have been exclue from the analysis to begin with, the total number of physical regimes efine in this stuy is five. Nevertheless, the rest of the paper will continue to refer to four precipitation regimes, treating the no-precipitation regime as a zeroth regime because of its trivial character. The main features of the four major precipitation regimes ientifie will now be escribe. It shoul be note a particular storm event or mesoscale convective system (MCS) may an inee oes fall into multiple precipitation regimes. Transition of a storm event or MCS between regimes occurs when the temporal evolution of the event is capture within the omain of the raar. Transition between regimes may also occur when a storm event or MCS enters an leaves the omain of the raar. 3. Tropical precipitation regimes as ientifie in raar ata The application of a cluster algorithm to the FADs erive from hourly volume scans using the polarimetric C- Pol raar locate near Darwin reveals four major precipitation regimes in that region. To etermine the robustness of the precipitation regimes each wet season was analyze separately. The regimes efine when using iniviual seasons (not shown) were foun to be very similar in structure an composition to the precipitation regimes efine here, although the relative frequency of occurrence (RFO) of the four regimes oes vary from year to year. The mean FADs for each of the four regimes are shown in Fig. 3. The orer of the panels in this figure is in terms of the RFO of each regime, which is shown in the top-left-han corner of each panel. Because of its trivial nature the no-precipitation regime is not plotte; however, for completeness the RFO of this zeroth regime is isplaye on the right-han sie of each panel. It is evient that for the ataset use here hourly ata for four wet seasons (NDJFMA) this regime has the secon highest frequency of occurrence (about 38%), highlighting the well-known fact that even on the scale

5 MAY 2009 C A I N E E T A L FIG. 3. The four precipitation regimes efine by the K-means algorithm. Regimes are orere (most to least) by their RFO. The RFO of the zeroth regime (i.e., the regime containing time perios that have no precipitation over the entire raar omain) is liste for completeness. The contours show bin counts ivie by total observe gri points. of the raar footprint use here (about 300-km iameter) there are many times with no precipitation. Most of those occur in the transition season early an late in the wet-season efinition use here (see section 4). Another important regime characteristic isplaye in Fig. 3 is the TVC of the raar signal introuce in the previous section. The TVC provies a useful first inication for the ifferences between regimes. TVC increases steaily from 2% for the most frequent precipitation regime to 26% for the least frequent regime, inicating that in the Darwin region the tropical atmosphere exhibits regimes with small precipitation coverage significantly more frequently than those with large precipitation coverage. Below follows a brief escription of each of the regimes isplaye in Fig. 3. In escribing the regimes low reflectivities will refer to values between 0 an 20 BZ, meium reflectivities will refer to values between 20 an 40 BZ, an high reflectivities will refer to values between 40 an 60 BZ. Those ranges are inicate in the figures by vertical lines. To further ai conceptualization of the precipitation regimes Constant Altitue Plan Position Inicator (CAPPI) examples of each regime are shown in Fig. 4 for both 2.5 an 10 km in the vertical. These specific examples are representative of each regime an were chosen by fining those histograms with the minimum Eucliean istance from the cluster mean histograms isplaye in Fig. 3. These CAPPI examples confirm that the regimes are well ifferentiate by TVC an the istribution of reflectivity values. Regime 1 is the ominant weather conition occurring over Darwin uring the four wet seasons analyze. It occurs approximately 45% of the time. The precipitation associate with this regime is weak an shallow. It has the lowest frequency of reflectivity values above 40 BZ an a maximum echo-top height up to 2 km shallower when compare with the other regimes. The TVC for this regime is 2%, which is an orer of magnitue smaller than the next closest regime. The low TVC value an absence of a bright ban at the freezing level (approximately 5 km) inicates that the convection has a low spatial/

6 1590 M O N T H L Y W E A T H E R R E V I E W VOLUME 137 FIG. 4. CAPPI examples eeme representative of (top to bottom) the four precipitation regimes. Examples are given for both (left) 2.5 an (right) 10 km in the vertical.

7 MAY 2009 C A I N E E T A L volume coverage an is therefore probably patchy an contains only a small proportion of stratiform precipitation. Average convective an stratiform rain contributions for each regime will be analyze below to further investigate this hypothesis. On average, higher rain rates occur over the continent than over the ocean in this regime (cf. Fig. 5). This is an important feature of regime 1 an will be investigate further below. Regime 2 has a RFO of approximately 10%. The TVC of this regime is an orer of magnitue greater than in regime 1 inicating that the precipitation covers a much larger volume of the raar omain. In this regime the maximum echo-top height extens to just over 16 km an there is a greater frequency of high reflectivity values at low levels, inicating that the convection is of greater intensity than was foun for regime 1. It shoul be note that regime 2 actually has the highest reflectivity values of all four regimes. Another efining feature of regime 2 is the increase occurrence of meium reflectivity values (20 40 BZ) from the lowest level to about 13 km, with the greatest frequency occurring between 6 an 12 km. As was the case in regime 1, the absence of a strong brightban signal near 5 km is inicative of precipitation that is likely to be preominately convective, which will be further investigate in this section. Regimes 3 has a RFO of approximately 4%. A significant increase can be seen in both TVC an echo coverage at low levels. The low occurrence of high reflectivity values below 5 km an the ecrease echotop height inicates that the convective component of this regime is weaker than the convection associate with regime 2. Oceanic convection tens to be shallower an have lower maximum reflectivity than continental convection (Toracinta et al. 2002; Nesbitt et al. 2000), the possibility that regime 3 coul be influence by precipitating systems with oceanic origins will be investigate in section 4. The large increase in frequency of low an meium reflectivity values is likely ue to the presence of significant areas of stratiform precipitation an woul explain the relatively high TVC for this regime; evience of stratiform precipitation can be seen in the form of a bright ban at approximately 5 km for meium reflectivity values. However, as the regime is an average of many time perios an contains both stratiform an convective precipitation, the bright ban is less pronounce than woul normally be the case for purely stratiform states. Further evience of the stratiform component of this regime will be provie in this section. Regime 4 is the rarest of all the regimes an has a RFO of uner 3%. In contrast to regime 3 there is a large occurrence of high reflectivity values below 5 km. The maximum echo-top height for this regime is higher than in regime 3 an is even slightly higher than was foun for regime 2. The sprea of reflectivity values at 15 km is larger than in regime 2 an extens to much higher reflectivity values than those foun in regime 3. Regime 4 has the largest TVC an by far the greatest frequency of meium reflectivity values, inicating the presence of large amounts of stratiform precipitation with greater vertical extent than the stratiform precipitation seen in regime 3. The bright ban in regime 4 is more pronounce than was seen in regime 3 an can be seen up to approximately 46 BZ. Another efining feature of regime 4 is the large frequency of meium reflectivity at the lowest levels, particularly between reflectivity values of 24 an 36 BZ. Figure 5 shows the average rain rate in millimeters per hour for the four regimes. For each time perio use in the stuy (approximately ) the 2.5-km reflectivity ata was converte to rain rate maps using the relationship Z 5 305R 1.36 (Bringi et al. 2001), where Z is the absolute reflectivity an R is the rain rate in millimeters per hour. Average rain-rate maps for each regime were then create by summing the n rain-rate maps associate with each regime an iviing by n. The R total is isplaye in the top-right-han corner of each figure, calculate by summing the rain rates over the entire raar omain an iviing by the number of vali gri points, thereby giving some measure (spatially average) of the rain rate of each regime. The rainfall istribution in Fig. 5a shows that on average more rainfall occurs over lan than over the ocean for regime 1; this is especially true over the Tiwi Islans. As expecte this regime has the lowest rain rate of all the regimes, approximately 0.3 mm h 21 average over the entire omain. Figure 5b supports the assertion that regime 2 contains convection initiate by convergence ue to lan sea breezes; a slightly higher rain rate can be seen orientate parallel to the coastline of the mainlan. Slightly higher rain rates can also be seen over the ocean in the west of the raar omain. Analysis of the iurnal cycle of rain rate for regime 2 (not shown) helps explain the patterns seen in Fig. 5b. During the afternoon precipitation occurs preferentially over lan areas, while uring the night an early morning, precipitation occurs over the ocean. The higher rain rates over the ocean in the west of the omain in Fig. 5b are ue almost exclusively to precipitation occurring at night. Although R total of regime 2 is approximately 4 times that of regime 1 (cf. 1.1 an 0.3), its RFO is approximately 1/4 times that of regime 1 (cf. approximately 11% for regime 2 an 45% for regime 1). Thus regimes 1 an 2 contribute approximately equal amounts of rainfall to the region.

8 1592 M O N T H L Y W E A T H E R R E V I E W VOLUME 137 FIG. 5. Average rainfall rates for the four precipitation regimes in mm h 21. Figure 5c shows that in regime 3 rainfall is fairly constant throughout the omain of the raar, with perhaps a slight increase in rainfall over the ocean to the west. An interesting feature of this regime is that R total is much lower than was foun for regime 2, espite the fact that the volume coverage of hyrometeors is much larger for regime 3 than regime 2. This might be explaine by the fining that the convection associate with regime 3 is weaker than the convection associate with regime 2. Figure 5 shows that in regime 4 high rainfall rates occur preferentially over the ocean. As expecte, R total for this regime is the largest of all four regimes, owing to the large amounts of stratiform precipitation an intense convection foun within this regime. To further test the spatial istribution of the precipitation within the regimes a lan ocean mask was applie an the rain intensities over lan an ocean were calculate for each regime. The lan (ocean) rain intensity is efine as the sum of the rainfall over the lan (ocean) ivie by the number of lan (ocean) points. Table 1 shows that the rain intensity is greater over lan in regime 1. Regime 2 has approximately the same rain intensity over lan an ocean, as iscusse above this regime contains contributions from convection occurring over lan (uring the afternoon) an over the ocean (uring the night). Both regimes 3 an 4 have greater rain intensity over the ocean than the lan. Figure 6 shows the average rain rate for each regime split into convective an stratiform components. The rain fiels measure with the C-Pol raar have been isaggregate into convective an stratiform areas using the algorithm escribe by Steiner et al. (1995). The only moification to this was to utilize the polarimetric rain estimates obtaine with the C-Pol raar using algorithms escribe by Bringi et al. (2001, 2004). These rain estimates were then converte to an effective reflectivity using a simple Z R relation (Z eff 5 305R 1.36 )

9 MAY 2009 C A I N E E T A L TABLE 1. Rain intensity of each regime split into lan an ocean contributions (mm h 21 ). Regime Lan Ocean in orer to apply the Steiner et al. algorithms. This ha the benefit of the classification being less sensitive to variations in measure reflectivity associate with rop size istributions, attenuation of the sample reflectivity, an hail contamination as well as being consistent with the simple mapping of the observe fiel. The convective an stratiform rain was then summe over the raar omain (a 300-km-iameter circle) to calculate the contributions of convective an stratiform rain areas to the total rain accumulation. Figure 6 inicates that regime 1 is preominately a convective regime, with stratiform precipitation contributing 24% to the overall rain rate. Regime 2 is also preominately convective in nature; however, the stratiform component of this regime has increase to 35%. Regime 3 contains the most stratiform precipitation relative to convective precipitation; 64% of the precipitation in regime 3 is stratiform in nature. Regime 4 also contains a large proportion of stratiform precipitation (53%); however, the convective component of this regime is quite large, almost as high as the total rain rate for regime 2. In summary, analysis of the CFADs an convective stratiform rain rates leas to the conclusion that regime 1 contains shallow, patchy, an relatively weak convection. Furthermore, the average rain-rate map for regime 1 an lan ocean rain intensities inicates that the convection associate with regime 1 preferentially occurs over lan areas. Regime 2 was foun to be a relatively strong convective regime that precipitates preferentially over the lan uring the ay an over the ocean at night. The convection in this regime has greater intensity an scale than the convection in regime 1, with a greater proportion of the convection likely initiate by lan sea breezes. The convective nature of this regime was confirme by the relatively large proportion of convective rain as seen in Fig. 6 an absence of a brightban signature in Fig. 3. Regime 3 was foun to contain a mixture of relatively weak convection an large amounts of stratiform precipitation. The temporal evolution of regimes (not shown), inicates that regime 3 is likely to occur from maturing systems associate with regimes 2 an 4. Regime 4 was foun to be a mixture of relatively strong convection with large amounts of stratiform precipitation. Having iscovere some of the basic characteristics of the regimes the next section aims to she further light on their characteristics by stuying their seasonal an iurnal cycles, an their link to the monsoon. 4. Some characteristics of the precipitation regimes The previous section has objectively ientifie four (five if the no precipitation is inclue) major convective regimes in the Darwin region. The four nontrivial FIG. 6. Area-average stratiform an convective rain rates for the four precipitation regimes.

10 1594 M O N T H L Y W E A T H E R R E V I E W VOLUME 137 regimes show characteristic istributions of raar reflectivity both in the horizontal an vertical on the scale of a typical weather raar footprint. While it was possible to link the regimes to known features of tropical convection, their link to large-scale features of the tropics as they affect the Darwin region remain to be eluciate. This section is aime at establishing links between the regimes ientifie in the previous section an the known features of the tropical atmosphere. First their response to the main force moes of the atmosphere, namely the seasonal an iurnal cycles are investigate. This is followe by establishing the regime characteristics as a function of monsoon activity in the Darwin region. a. The seasonal cycle In the Darwin region the seasonal moe is ominate by the Australian summer monsoon, which ominates the region s weather usually from the secon half of December into early March (Drosowsky 1996). Active monsoon perios are interlace with monsoon breaks. The monsoon perio is borere by a builup an ecay perio on the orer of 1 2 months each, while the rest of the year has a istinct ry-season character. Here the signature of the seasonal cycle on the convective regimes is investigate by stuying the seasonal variation of their frequency of occurrence. Figure 7 shows the relative frequency of occurrence of the five regimes (incluing the no-precipitation regime) as a function of month. The no-precipitation regime (regime 0) shows the expecte seasonal cycle with a minimum in January, the month with the largest influence of the Australian summer monsoon, an maxima in the months of March an April, those in which the Darwin region transits into the ry season. The RFO of the patchy convection regime (regime 1), is remarkably constant throughout November February with a ecline in occurrence in March an April. The stronger convective regimes (regimes 2 4) show an increase in their RFO from November to January, a ecline in February an are virtually absent in March an April. The most notable change from November to January occurs for regimes 3 an 4, confirming the impression gaine in the previous section that these regimes probably have a more monsoonal character (Steiner et al. 1995). This provies a further inication that the objectively ientifie precipitation regimes may represent physically relevant states of the atmosphere aroun Darwin, a claim that will be further investigate below. b. The iurnal cycle Figure 8 shows the iurnal cycle of the RFO of the four precipitating regimes. The RFO for each regime has been normalize incluing the no-precipitation regime. The iurnal cycles of the four precipitation regimes appear to be separate into two istinct patterns. Regimes 1 an 2 (top panels) have a peak in their RFO in the late afternoon, while regimes 3 an 4 (bottom panels) show a peak in the late at night an in the early hours of the morning. The two istinct patterns seen in theiurnalcyclesmayinicateaphysicalifferencebetween the precipitation regimes. One possible explanation for this ifference woul be that the ratio of lan to oceanic convection is ifferent for the ifferent regimes (as seen in Table 1). Convection tens to peak uring the late afternoon/early evening over lan, while over the oceans convection tens to peak uring the early morning (Yang an Slingo 2001; Gray an Jacobson 1977; Mapes an Houze 1993). However any conclusion base on the iurnal cycle alone is problematic, Mapes an Houze (1993) have shown that the iurnal cycle of col clou clusters epens on the size of the cluster an Yang an Slingo (2001) showe that the iurnal signal over lan often extens to the ajacent oceans. Furthermore Nesbitt an Zipser (2003) foun that MCS rainfall over lan peaks in the early morning. As regimes 3 an 4 both have high TVC values it is likely that MCS occurring over lan are also associate with regimes 3 an 4. Apart from the late afternoon maximum, regime 1 is equally likely to occur at any other hour of the ay. This may be inicative for the fact that while lan heating strongly influences regime 1, this relatively weak convection may well occur over the ocean an at other times of the ay. Earlier it was postulate that regime 2 has a larger proportion of convection initiale by sea breezes, if this is the case, the minimum in the iurnal cycle for regime 2 (aroun 10 am) may well be ue to the lack of the sea breeze at this time. Regime 2 also isplays a seconary peak in the early hours of the morning, which is likely ue to the precipitation that occurs over the ocean in the western part of the omain. c. Regime occurrence an monsoon activity Another well-known inicator for the character of convection in the Darwin area is the synoptic meteorology of the Australian summer monsoon. A simple but effective efinition of monsoon activity is win irection in the lower troposphere, with westerly wins inicative for an active phase of the monsoon, while easterly wins ominate the break as well as pre- an postmonsoon perios (Drosowsky 1996). Convection tens to be shallower an more representative of that foun over oceans uring perios when a westerly win places Darwin in an oceanic air mass. Easterly win

11 MAY 2009 C A I N E E T A L FIG. 7. Seasonal cycle of the precipitation regimes showing the RFO of each regime, for the months November April. conitions bring air from northeast Australia to the region an the convection is often more representative of that occurring over continents. Here the concept of an east west split in win irection that previously prove useful in other investigations (e.g., Keenan an Carbone 1992) is utilize with the aim of etermining if the win irection has any influence on the occurrence of the four precipitation regimes efine. The purpose of the analysis is not to fin a clear elineation between the regimes, but rather to compare how the precipitation regimes foun using cluster analysis compare to traitional methos. For the analysis, win ata from the Darwin airport at 700 hpa is efine as easterly or westerly if the win irection has any component in those irections. The frequencies of occurrence of win irection within each regime are shown in Fig. 9. It is evient from the figure that there are major ifferences between regimes in their relationship with win irection an hence monsoon activity. Regimes 0 2 show a preominance in their occurrence uring times when the win is coming from the east (break conitions), uring break conitions precipitating clou systems are known to be typical of those with continental origin (Steiner et al. 1995). The no-precipitation regime shows the strongest association with easterly flow. As inicate in Fig. 7 this regime occurs most frequently in the monsoon builup an ecay phases of the season. Those perios are associate with steay easterly flow, which explains the high frequency of occurrence of easterlies when this regime is foun. Although regime 2 is preominately an easterly regime, it contains more westerly flow than regime 1. This may help to explain why regime 2 was foun to have to be a mixture of oceanic an lan convection. In contrast, regimes 3 an 4 are most strongly associate with westerly conitions; westerly flow is typical for active monsoon perios, which have previously been shown to be associate with convection of maritime character (Keenan an Carbone 1992). While simple, analyzing the preominant win irection in each of the precipitation regimes has further supporte the finings of the previous sections, that the precipitation regimes ientifie in this stuy are physically sensible. While probably following intuition, this constitutes a nontrivial fining, as there is no physical a priori information that supports that ivision entering into the cluster algorithm. The fact that the four (five) regimes coul be objectively ientifie an match physical characteristics of convection rather well enables their use for further stuies of convective systems in the Darwin area an possibly beyon. 5. Precipitation versus ISCCP-base clou regimes The previous section has establishe that the convective regimes ientifie in this stuy for the Darwin area exhibit two istinctly ifferent patterns of their iurnal cycles an that the objectively efine regimes

12 1596 M O N T H L Y W E A T H E R R E V I E W VOLUME 137 FIG. 8. Diurnal cycle of the four precipitation regimes (in local time), the RFO has been normalize incluing the zeroth regime. are able to naturally separate when analyzing the synoptic-scale wins. Furthermore, the strength of the convection an the presence an size of the stratiform precipitation component all constitute information containe in the convective regimes. This enables the use of these regimes for further stuies of tropical convection an its interaction with the large-scale circulation. Given this an the fact that this work was partially motivate by the success of fining tropical regimes base on clou properties retrieve from satellites, it appears worthwhile to investigate the relationship of the six ISCCP-base tropical clou regimes of Rossow et al. (2005) with the precipitation-base regimes ientifie here. This will provie some insight into the precipitation structure (in a statistical sense) of the clou regimes, which in turn have alreay been shown to have links to tropical circulation features (Rossow et al. 2005; Höglun 2005). To set the context for further iscussion a brief summary of the six ISCCP weather states is given in this section; for a etaile iscussion of the regimes the reaer is referre to Rossow et al. (2005). Base on their clou signature, three of the six regimes can be escribe as suppresse with respect to the occurrence of eep convection, while the other three have been ientifie as convectively active. Of the active regimes the eep anvil clou (CD) regime exhibits a large coverage with optically thick clous, most likely a mix of thick anvils an convective towers, while the convective cirrus (CC) regime is ominate by a large coverage with cirrus clous of small-to-meium optical thickness. Probably the weakest (in terms of convection) of the three convectively active regimes is the mixe clou (MIX) regime, terme because it likely consists of a mixture of shallow, congestus, an eep convection without significant anvil an cirrus coverage. The suppresse thin cirrus (STC) regime is categorize as a suppresse regime an is ominate by a large coverage with thin cirrus clous. As shown by Rossow et al. (2005) this regime occurs close to convection without containing significant amounts of convective clous in the area over which the ISCCP histogram is calculate. The suppresse shallow clou regime (SSCL) regime has high frequencies of occurrence for clous with high cloutop pressure (low tops) an is interprete as consisting in most part of suppresse shallow clous with a low total clou cover as is typically foun in shallow cumulus clou fiels. The suppresse shallow clou regime with high clou cover (SSCH) is also ominate by low clous, but generally shows a much higher total

13 MAY 2009 C A I N E E T A L FIG. 9. RFO of the four precipitation an zeroth regimes for a given win irection (efine as either easterly or westerly). coverage with clous inicative of stratocumulus clous. Stratocumulus is almost never observe in the Darwin area resulting in an extremely low frequency of occurrence of the SSCH satellite regime in the stuy region. To avoi sampling biases this regime was exclue from further analysis. As in previous stuies (Höglun 2005) one ISCCP weather state per ay is efine for each ISCCP gri box. As Darwin strales several ISCCP gri boxes, sensitivity stuies to the choice of gri point were carrie out, but showe little sensitivity with regar to the main conclusions. Hence, the analysis presente here uses four ISCCP gri boxes each having an area of km 2 locate roughly between S an E. The relationship of the ISCCP-base clou regimes to the precipitation regimes of this stuy, which are ientifie hourly, is investigate by calculating the mean frequencies of occurrence of each precipitation regime on all ays that fall into a particular ISCCP regime. Note that the C-Pol raar etects precipitation-size particles an is unlikely to observe cirrus an nonprecipitating clous. Figure 10 shows the RFO of the five regimes of this stuy for each ISCCP-base clou regime. The clou regimes are sorte as in Rossow et al. (2005) from what is consiere the most convectively active regime (CD) to the most suppresse regime (SSCL). Several noteworthy features are evient in Fig. 10. The occurrence of no precipitation (regime 0) increases steaily from CD to SSCL. This supports the intuitive interpretation use in earlier work efining the ISCCP-base clou regimes as convectively active versus suppresse. Of the suppresse regimes, SSCL clearly shows the least occurrence of eep convective regimes (regimes 2 4) an is mostly characterize by the relatively shallow precipitation regime 1. There is a steay increase in the eep convective regimes (regimes 2 4) for the MIX an CC regimes, mostly at the expense of the no-precipitation regime. Nevertheless, both these clou regimes remain ominate by the regimes with low precipitation coverage (regimes 1 an 2). The only clou regime with a significant occurrence of the regimes with high precipitation coverage (regimes 3 an 4) is the CD regime. In this regime the frequency of occurrence of no precipitation at all has roppe to aroun 10% inicating the frequent occurrence of precipitation in that regime. In earlier stuies (Rossow et al. 2005; Höglun 2005) it was foun that the ISCCP-base CD regime occurs preominantly in active phases (i.e., convergence zones) of large-scale tropical circulation features such as the Maen Julian oscillation (MJO) or the Australian summer monsoon. This an the ientification of the CD regime as the only regime containing high frequencies of regimes 3 an 4 leaves the intriguing possibility that what is thought of as typical maritime precipitating convection is in fact a convective type that is enable an supporte only by larger-scale circulation features. An in-epth stuy into the precipitation characteristics of the ISCCP regimes is beyon the scope of this investigation; however, for completeness a simplistic calculation is performe to etermine the average rain rate per hour of each ISCCP regime. As the rain rates for the ISCCP regimes have been erive from ata obtaine from the region surrouning Darwin, they are not intene to be applie to the tropics as a whole. The rain rate for each of the ISCCP regimes was calculate by multiplying the rain rate for each precipitation regime (Fig. 5) by the frequency with which the raar regimes occur within the given ISCCP regime (Fig. 10). These results are isplaye in Table 2 an show that the rain rates for the ISCCP regimes is highest for the most convective regime CD, an ecreases corresponing to convective intensity, with the suppresse regimes having the lowest rain rates. It is interesting to note that the CC an the MIX regimes both have approximately the same rain rates. Although simplistic, associating an average rain rate with each of the ISCCP regimes can be thought of as a first step in creating/assessing parameterizations for numerical moeling. For example Fig. 10 an Table 2 show that from a precipitation point of view regimes CC an MIX are quite similar; therefore, it may be useful to create parameterizations for tropical weather conitions base on the ISSCP regimes (which represent istinct an ifferent states of the tropical atmosphere that can istinguish the four main classes foun here), using the raar regimes to estimate of the total rainfall an latent heating associate with each regime.

14 1598 M O N T H L Y W E A T H E R R E V I E W VOLUME 137 TABLE 2. Calculate average rain rate of each ISCCP regime (mm h 21 ). CD CC MIX STC SSCL FIG. 10. Comparison between satellite an raar regimes. RFO as a function of raar regime, for the satellite regimes efine by Rossow et al. The SSCH satellite regime has not been inclue because of an extremely low frequency of occurrence. 6. Concluing remarks This stuy has investigate if the three-imensional structure of precipitation in the Darwin area as measure by a scanning polarimetric weather raar can be objectively classifie into regimes. It has furthermore pose the question if those regimes have physical connections to the known features of the tropical atmosphere in the region. A cluster algorithm was applie to two-imensional histograms of reflectivity with height erive from quasi-instantaneous raar volumes. After excluing a regime with virtually no raar returns in the volume (a no-precipitation regime) the analysis reveale the existence of four precipitation regimes in the Darwin region: Regime 1 a patchy convective regime of meium intensity an low area coverage. This regime occurs most frequently in the afternoon an uring break conitions. Regime 2 a strong convective regime with relatively small area coverage. This regime has a peak in the iurnal cycle in the afternoon an a seconary peak in the early morning. Regime 3 a weak convective regime with large area coverage an large stratiform regions. This regime occurs most frequently uring the late night/early morning an uring monsoon conitions. Regime 4 a strong convective regime with large area coverage an large stratiform regions. This regime has a peak in the iurnal cycle in the late night early morning an has links to monsoon conitions. The overall statistics of the occurrence of these regimes once again highlights that even uring convectively active perios such as the Australian summer monsoon, the spatial coverage with precipitation is overall small (cf. Figs. 2 an 3). Linking the regimes ientifie by the algorithm to known features such as the seasonal an iurnal cycle an a simple escription of monsoon activity through win irection has confirme their ifferent physical character. Regimes 1 an 2 show signs of a strongly laninuce iurnal cycle with an afternoon peak an occur throughout the wet season. They are largely associate with wins from the east, typical for monsoon builup an ecay as well as monsoon break conitions, uring which convection is often observe to be of continental character. Regimes 3 an 4 have their largest frequency of occurrence in January the peak of the monsoon season. Their iurnal variation peaks at night or early morning hours an they are largely associate with westerly wins, as is typical for monsoon conitions. Combining the precipitation regimes with tropical clou regimes previously ientifie from ISCCP ata reveals interesting relationships between the two. Arguably there exist probably three (maybe four) broa precipitation regimes across the five ISCCP-base regimes present at Darwin. The suppresse shallow clou regime (SSCL) is ominate by the no-precipitation an the patchy convection regimes. The stronger convection regimes, in particular, the maritime ones o not feature strongly in this clou regime. While still ominate by the patchy convection regimes, the convective cirrus (CC) an mixe (MIX) clou regimes an, to a lesser extent, the suppresse thin cirrus (STC) regime show larger frequencies of occurrence of the stronger precipitation regimes. The fraction of maritime versus continental precipitation regimes shifts steaily to more an more maritime with increasing convective character as ientifie in the clou regimes themselves. By far the most active an most maritime regime is the convective eep anvil clou regime (CD). About 50% of the time in this clou regime one of the three strong convective precipitation regimes is present an this is the only regime where the strong maritime regime occurs with almost equal frequency to its weaker counterpart. The occurrence of no precipitation or patchy convection in this clou regime is significantly reuce compare to any other clou regime, showing the