1. Introduction
Severe weather in the semiarid–arid climate of the southwestern United States is principally the result of organized convective events that take place during the late summer months and particularly at lower elevations (e.g., Schmidli 1986; McCollum et al. 1995; Smith and Gall 1989; Wall et al. 2012). These more severe convective events, as well as warm-season rainfall in general in northwestern Mexico and much of the southwestern United States, occur during the North American monsoon (NAM; Adams and Comrie 1997). The NAM provides on the order of 30%–50% of annual precipitation in the southwestern United States and an even greater portion in northwestern Mexico (Douglas et al. 1993). Although organized convective events supply much-needed precipitation, they are also associated with severe weather hazards, including hail, lightning, dust storms (haboobs), microbursts, and flash flooding (e.g., McCollum et al. 1995; Ray et al. 2007). How severe weather events may change in the future is a subject of rapidly growing interest and research that is beginning to be addressed in the context of high-resolution regional atmospheric modeling with explicitly resolved convection (e.g., Trapp et al. 2011). To properly represent organized mesoscale convective systems in the NAM region, it is necessary to utilize a convective-permitting model with a spatial scale on the order of 1 km (e.g., Li et al. 2008; Newman and Johnson 2012). As a consequence, objective methods for identifying atmospheric conditions suitable for the generation of severe weather become necessary to efficiently simulate them in long-term simulations with regional climate models.
Deep convective activity and its suppression during the NAM in the Southwest occur in association with “burst” and “break” periods (Carleton 1986; Adams and Souza 2009). Monsoon bursts typically last for several days, whereas the breaks range in length from several days to several weeks (Maddox et al. 1995). Burst and break conditions result from shifts in the position of the monsoon ridge, typically considered at 500 hPa and climatologically centered over the Four Corners region in the Southwest during the period from late July to early August (Carleton 1986; Maddox et al. 1995). A northward shift in the monsoon ridge can facilitate more convectively active burst periods through a combination of 1) increased moisture advection, 2) decreased atmospheric stability under weakened subsidence, and 3) the passage of transient disturbances that provide favorable vertical wind shear profiles, with wind speed generally increasing from an easterly direction from the surface to upper levels, and/or synoptic-scale upward vertical motion to facilitate convective organization (Maddox et al. 1995; Douglas and Englehart 2007; Bieda et al. 2009; Finch and Johnson 2010; Newman and Johnson 2012).
Objective techniques for identifying large-scale synoptic flow patterns have not yet been developed for the NAM region using available long-term gridded datasets—for example, atmospheric reanalysis products. Maddox et al. (1995) do present an effective and subjective classification of synoptic severe weather patterns for the NAM in the Southwest through compositing of daily weather maps. They specifically identify three midlevel (500 hPa) synoptic patterns in their Fig. 3, shown as Fig. 1 here. The first two patterns reflect a northeasterly or northwesterly displacement of the monsoon ridge from its climatological position. As discussed in Bieda et al. (2009), these types of monsoon-ridge displacements are also typically associated with the passage of synoptic disturbances, namely, a westward-moving, transient inverted trough along the southern flank of the ridge, oftentimes resulting in the formation of a propagating mesoscale convective system. The third pattern reflects a transitional pattern to stronger westerly flow aloft with the passage of an upper-level trough and surface cold front. This type of pattern occurs more frequently toward the seasonal withdrawal of the monsoon and can even be associated with the occurrence of weak supercell-type convection in Arizona (Blanchard 2011). In all three of the Maddox et al. severe weather types, increases in low-level moisture over southern Arizona occur, leading to greater thermodynamic instability. A very important caveat to note with respect to the Maddox et al. monsoon-related severe weather types is that they do not account for extreme weather events that are due to the direct passage of remnant tropical cyclones. These types of events would tend to occur more in late summer and early autumn in the southwestern United States (e.g., Wood and Ritchie 2013) and are associated with some of the most extreme precipitation events in the region. To be clear, the methodological approaches to identify severe weather events described in this study best characterize organized convection, namely, westward-propagating mesoscale convective systems, that would occur in the most moist and unstable atmospheric environments in southern Arizona.
Characteristic 500-hPa streamlines for severe-weather-pattern types I, II, and III. Original source: Maddox et al. (1995).
Citation: Journal of Applied Meteorology and Climatology 55, 11; 10.1175/JAMC-D-16-0075.1
We take advantage of existing long-term datasets, both observational and gridded data products, to objectively characterize modes of severe weather for the NAM in the Southwest. In an approach that differs from that of Maddox et al. (1995), who use reported severe weather events to derive their composites, we invert their method. Given their emphasis on low-level moisture and resulting atmospheric instability, we consider first thermodynamic criteria—to be specific, convective available potential energy (CAPE) and integrated precipitable water vapor (PWV)—that are derived from daily radiosonde observations as the necessary precondition for identification of severe-weather-event days. From these “thermodynamically favorable” days, we apply empirical orthogonal function analysis to gridded daily 500-hPa heights from an atmospheric reanalysis to extract the dominant modes of variability. Our hypothesis is that Maddox et al. (1995) synoptic classifications should be objectively retrieved in both paradigms and that there should be clear evidence of organized convective systems on the identified severe-weather-event days. The work that is presented here represents our first step toward developing an objective methodological approach to severe-weather-event identification that begins with the use of historical radiosonde data. In subsequent work, we will then adapt it to gridded data from long-term regional atmospheric reanalysis and long-term simulations with regional climate models.
This paper is organized as follows. Section 2 describes the datasets used in the analysis, including upper-air and precipitation data. The methodological approach to objectively select the severe event days on the basis of thermodynamic parameters and to identify dominant atmospheric circulation patterns using matrix methods is outlined in section 3. Results in section 4 show the thermodynamic characteristics of severe-weather-event days, the derived severe-weather-event modes, and the relationship between these modes and monsoon precipitation. Section 5 presents conclusions, discussion, and future research directions and, in particular, how the methodological approach may be applied in the near future to analysis of simulation data from regional climate models.
2. Data sources
a. Tucson sounding and atmospheric reanalysis data
To characterize the thermodynamic environment in Arizona during the period of the NAM in the Southwest, we utilize the 1200 UTC radiosonde data from Tucson, Arizona (TUS; 32.1°N, 110.9°W; http://esrl.noaa.gov/raobs/) that are available from the NOAA/ESRL radiosonde database, consistent with the analysis of Maddox et al. (1995) and Adams and Souza (2009). The motivations for using only the 1200 UTC sounding are discussed later in the next section. The time period considered for the analysis is from 1993 to 2010. To define large-scale atmospheric circulation characteristics, we use the North American Regional Reanalysis (NARR; Mesinger et al. 2006).
b. Stage IV precipitation product
To verify the development and propagation of severe convective events, the combined stage IV radar–gauge precipitation product, as described by Zhang et al. (2011), is used for the period 2002–10. This dataset is a multisensor approach that integrates radar data with surface observations and has been employed in previous studies of NAM convection—for example, in previous work by the authors (Minjarez-Sosa et al. 2012). Stage IV is an improvement from stage III data, with manual quality control (QC) from 12 River Forecast Centers across the United States. Stage IV data have a 4-km resolution on polar stereographic grids on an hourly basis. Accumulated precipitation is recorded with QC for 6- and 24-hourly quantities. The data availability of the stage IV products is the limiting factor in the study period for considering the associated precipitation with the severe-weather-event modes.
3. Analysis methods
a. Convective available potential energy
CAPE is the first-order factor in forecasting NAM thunderstorms and has a clear relationship with precipitation occurrence (Adams and Souza 2009). CAPE is the positive buoyant energy of a given parcel of air, integrated vertically through the atmosphere (Moncrieff and Miller 1976). Methodological considerations in the calculation of CAPE include level of parcel initiation, thermodynamic path (reversible or pseudoadiabatic), virtual temperature effects, inclusion of latent heat of fusion, and water loading (Doswell and Rasmussen 1994; Emanuel 1994; Adams and Souza 2009). Much of the prior work in this area has focused on the characterization of convection in the central United States, but the traditional methodological approaches for CAPE computation that are appropriate for that region may not be appropriate for the NAM. In the central United States, convective inhibition (CIN) can allow for substantial values of CAPE that are unrealized. For the NAM, however, CIN associated with elevated temperature inversions normally plays only a limited role, because of the substantial amount of surface heating during the day, in comparison with the situation on the U.S. central plains (Bright and Mullen 2002). The intense insolation works to erase any near-surface or lower-atmospheric temperature inversions, and a deep adiabatic layer is created.
The study of Adams and Souza (2009) provides a detailed methodological approach for CAPE computation during the NAM. They demonstrated that the largest possible CAPE values for a given monsoon day are most highly correlated with widespread precipitation events across south-central Arizona (see their Fig. 1). There are two general considerations for computation of CAPE that account for the uniqueness of the NAM. The first is the timing of convection. The 0000 UTC (1700 LST) sounding may be unrepresentative of synoptic-scale characteristics because prior or simultaneous convective events can contaminate these soundings, particularly the near-surface temperature characteristics. By this reasoning, the morning soundings at 1200 UTC (0500 LST) are most likely to be uncontaminated by any convective precipitation, although these authors also eliminated contaminated morning soundings. The morning sounding is then modified for appropriate afternoon maximum temperature and boundary layer moisture, for a well-mixed layer, to calculate a more representative CAPE. To compute CAPE, the average of the mixing ratio in the lowest 50-hPa layer of the sounding and the warmest daily surface temperature are utilized. Likewise, parcel ascent is assumed to be pseudoadiabatic. This calculation yields the most unstable CAPE (MUCAPE).
b. Precipitable water
As noted above, severe weather has been directly linked to increases in low-level moisture from the surface to 700 hPa during the NAM (Maddox et al. 1995; Wall et al. 2012). The absence of this low-level moisture can stifle convection and inhibit precipitation (Adams and Souza 2009). Low-level moisture is accounted indirectly in the calculation of CAPE, but CAPE fails to depict accurately the overall influence of moisture in convective events (e.g., precipitation efficiency or downdraft strength). To account for atmospheric moisture content, we consider PWV as an additional thermodynamic parameter to CAPE. In monsoonal regions and in tropical continental regions in general, PWV provides a consistent positive correlation with precipitation (Bretherton et al. 2004; Neelin et al. 2009; Janiga and Thorncroft 2014) along with the onset and demise of the monsoon (Lu et al. 2009). PWV and its temporal evolution have also been shown to have a strong positive relationship with precipitation in southern Arizona and Mexico, using both observations (Kursinski et al. 2008a; Serra et al. 2016) and sensitivity experiments within a regional atmospheric model: the Weather Research and Forecasting (WRF) Model (Kursinski et al. 2008b). PWV has been employed in the forecasting of monsoon thunderstorms by both research scientists and operational meteorologists at the National Weather Service (Lu et al. 2009). The exact thresholds relating to severe weather development have not been specifically defined in the literature, permitting some necessary geographic subjectivity among operational forecasters. The value of PWV associated with organized convection generally is above 25 mm (i.e., approximately 1 in.) in Tucson, per the operational practice of the National Weather Service Office there.
c. Objectively classifying 500-hPa height patterns for severe-weather-event days
The days during the NAM with relatively high CAPE and PWV values, as determined from an analysis that uses joint frequency distribution described in section 4a, are regarded as favorable for the most severe thunderstorms. We call these, henceforth, the monsoon-related severe-weather-event days that are favorable for thunderstorm generation from a purely atmospheric thermodynamic standpoint. These monsoon-related severe-weather-event days are then analyzed with respect to their synoptic-scale atmospheric circulation patterns, utilizing daily 500-hPa geopotential height from NARR data as the proxy for monsoon-ridge positioning and strength. NWS operational forecasters in the Southwest normally evaluate CAPE and the magnitude of vertical wind shear when classifying the potential for convection. Although shear is not explicitly assessed in this analysis, a favorable vertical wind shear profile, typically associated with a transient inverted trough, does facilitate convective organization and severe weather during the NAM (e.g., Bieda et al. 2009; Newman and Johnson 2012).
The 500-hPa geopotential-height fields of the identified severe-weather event-days are objectively analyzed using empirical orthogonal function (EOF) analysis. EOF analysis, also known as principal component analysis, is a matrix-methods approach to find the spatial patterns that explain the most variance with respect to space and time. Considering a space–time field, EOF analysis produces 1) maps of the dominant spatial patterns (EOFs), 2) time series of these modes (principal components), and 3) the percent of explained variance of the modes (eigenvalues). Use of EOF analysis in atmospheric-science-related applications has been very well established in the literature (e.g., Kutzbach 1967; Prohaska 1976; Bretherton et al. 1992; Wallace et al. 1992). The EOF analysis here is performed using singular value decomposition (SVD). To prepare the data prior to SVD, considering only the identified monsoon-related severe-weather-event days, the mean values of 500-hPa geopotential height at each grid point are removed and the grid points are weighted on the basis of the square root of the cosine of latitude. This analysis produces maps of 500-hPa height anomaly patterns. The synoptic patterns associated with the mean 500-hPa height of the monsoon-related severe-weather-event days and dominant modes of the severe-weather-event days may then be compared with the prior subjectively determined results of Maddox et al. (1995).
For the EOF analysis procedure just described, the analysis domain encompasses the entire NARR domain. The entire NARR domain is used so as to be as consistent as possible with what has been shown in previous work by the authors in Castro et al. (2007b) in their analysis of interannual variability of 500-hPa height anomalies during the North American warm season. In their Figs. 5 and 6, they basically show that the height anomaly patterns that govern the variation in monsoon-ridge positioning in western North America are part of quasi-stationary Rossby wave teleconnection patterns that are hemispheric in nature. The existence of quasi-stationary Rossby wave train atmospheric teleconnections that affect North American warm-season climate has also been more recently confirmed by analysis of dominant modes of U.S. warm-season precipitation (Ciancarelli et al. 2014) and EOF analysis of 500- and 200-hPa geopotential-height anomalies over the entire Northern Hemisphere (e.g., Ding et al. 2011). These two latter works demonstrate that the variation in monsoon-ridge positioning is most closely related to a wave train that emanates from the western tropical Pacific and Indian Oceans monsoon region, referred to as the western Pacific–North America pattern. A standard (unrotated) EOF analysis on 500-hPa geopotential-height anomalies within a relatively large (from continental to hemispheric) domain is therefore a robust way to capture these types of atmospheric teleconnections and is consistent with the aforementioned prior literature. The EOF analysis technique has also been widely used in atmospheric data analysis since the seminal work of Wallace and Gutzler (1981), one of the first investigations that described the nature of Northern Hemisphere winter-season atmospheric teleconnections. Rotation of EOFs is less desirable to perform in our analysis because it would tend to create more localized centers of action in the spatial loading pattern (e.g., Richman 1986) that are less appropriate for capturing the hemispheric-scale nature of the quasi-stationary Rossby wave trains. Some brief discussion of the deleterious effects of EOF rotation and reduction of domain size on the characterization of 500-hPa height anomaly patterns is included in section 4.
The daily-precipitation climatological analysis of monsoon-related severe-weather-event days is calculated from the stage IV product for the period 1993–2010. To determine precipitation patterns associated with the dominant 500-hPa geopotential-height modes determined with EOF analysis, we use a composite-analysis technique. The composite values are based on the highest values of the normalized principal component time series of severe-weather-event days for the given mode. To be specific, we use the top 15% of events in the principal component time series. This approach yields two datasets consisting of approximately 60 severe weather events each, corresponding to type-I and type-II patterns. There are thus 3–4 events per year for each type, on average. The precipitation associated with a dominant mode of 500-hPa height anomalies is displayed on the basis of the composite precipitation of the highest 15% of days of a given mode minus the average daily-precipitation climatological value of all monsoon-related severe weather events. We found that displaying the precipitation patterns in this way emphasizes the expected presence of organized convective precipitation more downwind of elevated regions for days on which the 500-hPa ridge configuration should be most conducive for severe weather, as will be shown later. We finally consider the propagation characteristics of organized convection (speed and direction) with the stage IV product during sample severe weather events that are associated strongly with the derived 500-hPa height modes, utilizing Hovmöller diagrams.
4. Results
a. Severe-weather-event days based on Tucson MUCAPE and PWV
Thermodynamically favorable monsoon-related severe-weather-event days are identified using MUCAPE and PWV for the Tucson sounding during the period of June–September for the years 1993–2010. The histograms of these distributions are shown respectively in Figs. 2 and 3. We do not employ theoretical distribution fits to these data, as this is not really essential to our objective of determining the severe-weather-event threshold points. Nonetheless, there are some important distributional characteristics of these raw histograms that should be at least qualitatively described. The MUCAPE histogram is clearly positively exponentially distributed as a result of a high frequency of MUCAPE values at or near zero. Such days with virtually no appreciable atmospheric instability occur either prior to the onset of the monsoon (mostly during June) or during monsoon breaks. These are very hot, dry periods when the monsoon ridge is typically located directly over southern Arizona. According to the National Weather Service, a 500-hPa geopotential height above 5950 m over Arizona is indicative of a strong monsoon ridge (http://www.wrh.noaa.gov/twc/monsoon/monsoon_progression.php). For days with some appreciable MUCAPE, exceeding 500 J kg−1, there is a fairly uniform frequency of occurrence within the MUCAPE bins up to approximately the 70th percentile, to ~3000 J kg−1. Maximum MUCAPE in Tucson is in the range of 6000 J kg−1. This type of extreme MUCAPE event occurs on the order of 10 times during the 1993–2010 period, or roughly just one event in a given monsoon year. It should be emphasized that MUCAPE values from the modified morning soundings tend to be on the high end. In a typical case, boundary layer ventilation, which is not accounted for in Adams and Souza (2009), would tend to decrease near-surface moisture during the course of the day, thereby reducing CAPE, all else being equal.
Histogram of derived MUCAPE (J kg−1) values as based on raw observational rawinsonde data during June–September for Tucson (1993–2010).
Citation: Journal of Applied Meteorology and Climatology 55, 11; 10.1175/JAMC-D-16-0075.1
As in Fig. 2, but for integrated precipitable water (mm).
Citation: Journal of Applied Meteorology and Climatology 55, 11; 10.1175/JAMC-D-16-0075.1
PWV, as shown in Fig. 3, tends to be more normally distributed, albeit slightly negatively skewed. The 25-mm nominal operational forecast threshold for a convectively active monsoon day is roughly in the middle of the distribution. Even though Tucson is a semiarid climate, on rare occasions PWV values can exceed 50 mm, values that are reminiscent of a deep tropical environment such as the Amazon rain forest (Adams et al. 2013). Documented severe-flooding events during the monsoon in Arizona—for example, the Sabino Canyon flood in Tucson—have values of PWV on this order (Magirl et al. 2007; Griffiths et al. 2009).
A scatterplot of MUCAPE versus PWV for Tucson is shown in Fig. 4. The solid black lines indicates the upper-70th-percentile threshold for both MUCAPE and PWV. Days falling into the category of the upper-right quadrant (i.e., above the 70th percentile in both MUCAPE and PWV) are hence defined, in an objective way, as thermodynamically favorable severe-weather-event days. For the 1993–2010 period, 404 days met these criteria, with MUCAPE exceeding approximately 3000 J kg−1 and PWV exceeding 45 mm. As expected (see Fig. 4), a strong relationship between CAPE and PWV exists, their correlation coefficient being 0.86.
Scatterplot of CAPE (J kg−1) vs precipitable water (mm) in Tucson, considering all days during the monsoon (June–September), derived from original raw radiosonde upper-air-sounding data (1993–2010). The 70% threshold levels for PWV and CAPE, respectively, to select severe-weather-event days are indicated (horizontal and vertical black lines).
Citation: Journal of Applied Meteorology and Climatology 55, 11; 10.1175/JAMC-D-16-0075.1
The identified severe-weather-event days were compared with NWS “Storm-Based Warning” verification data from the Iowa Mesonet website (http://mesonet.agron.iastate.edu/cow/) for 2005–10 for the region of southern Arizona, basically from the latitude of the city of Phoenix and points farther south. These reports are based on NWS warnings and public reports of occurrences of severe weather phenomena, such as damaging winds, large hail, or flash flooding. Roughly 58% of the convective-event days objectively identified using our method corresponded to reported severe weather phenomena between 2005 and 2010. While these reports provide an independent source of data to verify the effectiveness of the objective approach to classify severe monsoon weather, the reporting of severe weather is highly dependent on population density and participation of the general public. Areas with limited or no population in southern Arizona, mostly outside the Tucson and Phoenix metropolitan areas, are generally under–accounted for in severe weather reporting. Despite these caveats, a comparison of severe weather reports with the objectively identified severe-weather-event days is an absolutely necessary step in the inversion of the methodological approach of Maddox et al. (1995), as mentioned in the introduction. We suspect that at least some of the severe weather events that are unaccounted for may be associated with recurving eastern Pacific tropical cyclones, as previously mentioned in the introduction. Although not explicitly investigated here, objective-analysis techniques have been used in other studies to characterize tropical cyclone–related precipitation in the southwestern United States (e.g., Wood and Ritchie 2013). Although these kinds of events are associated with extreme, more widespread precipitation, the atmosphere is actually relatively stable, as compared with the more typical convectively active monsoon day with well-organized mesoscale convective systems.
b. 500-hPa monsoon-ridge patterns of severe-weather-event days
As previously mentioned, the analysis of 500-hPa height patterns only accounts for the thermodynamically favorable severe-weather-event days and not for all of the days during June–September. So our a priori expectation is that the mean of the 500-hPa height field of the severe-weather-event days would be in a favorable configuration to facilitate monsoon precipitation, centered north of southern Arizona. The 500-hPa height anomalies associated with the first few EOFs can then be added to this mean field. The corresponding principal component (PC) time series would then show the relative strength of the individual modes for a particular day. To keep the methodological approach as objective as possible, we do not employ any subjective recombination of the individual EOFs. If the Maddox et al. severe weather modes are truly the dominant 500-hPa modes in the data that can be objectively retrieved by a matrix-methods approach, then their presence in the dominant EOFs should be clear and unambiguous. We note that the individual EOFs may not necessarily be statistically significant by standard tests of eigenvalue separation. Our goal here is simply to objectively recover as well as possible the 500-hPa height modes that have already been deemed physically important and are used in operational practice.
The mean of the 500-hPa heights and anomalies for the severe-weather-event days is shown in Fig. 5, along with the average precipitation anomaly from the stage IV product. The average synoptic pattern of all thermodynamically favorable monsoon-related severe-weather-event days, as expected, shows a strong monsoon ridge, centered slightly north of its favored climatological position in the Four Corners region in late summer. The highest local 500-hPa geopotential-height anomalies are located just to the north of the center of the monsoon ridge. The average precipitation (bottom of Fig. 5) reflects an active monsoon pattern over the Southwest, with precipitation generally maximized on the peaks of the terrain, for example, the Mogollon Rim in Arizona, with maximum values on the order of 7 mm day−1. This constitutes the baseline climatological map for severe weather events.
(top) 500-hPa geopotential-height (m) anomaly (shaded) relative to the total climatological mean (contour) for thermodynamically favorable severe weather events during 1993–2010, as identified by Tucson sounding data. (bottom) Average precipitation (mm day−1) from the stage IV product for the same thermodynamically favorable severe weather events but during 2002–10. Elevation terrain is indicated as contours at intervals of 1000 m. Regions over 2000 m in elevation are shown in hatching.
Citation: Journal of Applied Meteorology and Climatology 55, 11; 10.1175/JAMC-D-16-0075.1
The 500-hPa height anomalies of the first five dominant EOF modes of the severe-weather-event days are shown in Fig. 6. Taken together, the first five modes explain about 50% of the 500-hPa height variance, albeit again not necessarily statistically distinct by standard tests of eigenvalue separation (e.g., North et al. 1982). We assert that at least the first two modes are physically important because they 1) correspond approximately to Maddox et al. type-I and type-II severe monsoon weather patterns and 2) are very similar to height anomaly patterns that are related to the interannual variation in monsoon-ridge positioning, as previously described in Castro et al. (2007b). The Maddox et al. type-I severe weather pattern in the first EOF (EOF1) is related to a northward and eastward displacement of the ridge into the central United States. This first mode of monsoon-ridge displacement is related to decadal variability in Pacific sea surface temperatures and the occurrence of wet and dry periods in the central United States, as shown in Fig. 6 of Castro et al. (2007b). The Maddox et al. type-II severe weather pattern in EOF2 is related more to a northward and westward displacement of the ridge into Nevada and Utah. This second mode of monsoon-ridge displacement is related to ENSO-like variability in Pacific SSTs, as shown in Fig. 5 of Castro et al. (2007b). At least on an interannual basis, both of these dominant monsoon-ridge modes are related to quasi-stationary Rossby wave trains that emanate from the western tropical Pacific Ocean. Here we are showing that these same types of wave trains also appear in association with the dominant modes of the severe-weather-event days.
Regression between 500-hPa height anomalies (m) and the first five EOFs associated with the selected thermodynamically favorable severe weather events for 1993–2010. The variance explained for each EOF mode is indicated as a percentage.
Citation: Journal of Applied Meteorology and Climatology 55, 11; 10.1175/JAMC-D-16-0075.1
The 500-hPa height patterns that are associated with both of these modes are shown in Fig. 7, closely matching the original figure from Maddox et al. (1995) in our Fig. 1. The type-I pattern can be constructed by adding the mean 500-hPa height field of all severe weather events plus the height anomaly field of EOF1. The type-II pattern can by constructed by adding the same mean 500-hPa height field of all severe weather events plus the height anomaly field of EOF2. As previously stated, the most positive phase is defined as the composite of the highest 15% of the averaged, normalized PC values for the pattern loadings in Fig. 6. The type-III transitional pattern, however, cannot be reconstructed from combination of the dominant 500-hPa height modes, probably owing to the vagaries in the tracks of upper-level lows and their associated cold fronts. One possible alternative method to objectively consider type-III patterns would be to track upper-level lows during the latter part of the monsoon, similar to what we did previously for transient inverted troughs in Bieda et al. (2009). These same analyses were repeated using rotation of EOFs, with a Varimax method, and a smaller domain that is more confined to the contiguous United States (results not shown). To briefly summarize, EOF rotation produces dominant 500-hPa height anomaly modes that have stronger centers of action but less correspondence to the hemispheric-scale quasi-stationary Rossby wave trains and type-I and type-II severe-weather-event patterns. With the reduction in domain size, we were not able to obtain as clearly results that are comparable to the aforementioned figures in Castro et al. (2007b).
Objectively determined severe-weather-event 500-hPa geopotential-height patterns (m) that correspond to the (top) type-I and (bottom) type-II modes of Maddox et al. (1995). The modes are constructed by considering the mean of the 500-hPa geopotential height of all severe-weather-event days plus the average of the 500-hPa height anomalies for events that project most strongly (top 15%) on the positive phase of severe weather events of the (top) EOF1 and (bottom) EOF2 modes.
Citation: Journal of Applied Meteorology and Climatology 55, 11; 10.1175/JAMC-D-16-0075.1
Let us conceptually review why these type-I and type-II patterns would tend to be associated with monsoon-related severe weather in Arizona. First, these orientations of the monsoon ridge would favor enhanced easterly flow at upper levels with a more southeasterly component for type I and a more northeasterly component for type II. In both cases, westerly propagating upper-level disturbances (inverted troughs) are favored as well as weakened subsidence over Arizona (e.g., Bieda et al. 2009). Likewise, in both cases, a concomitant increase is observed in low- to midlevel moisture from the eastern Pacific and Gulf of California, as well as convectively mixed moisture over western Mexico (Maddox et al. 1995). This increase in low-level moisture, a decrease subsidence, and more favorable wind shear profiles may act in concert to increase the likelihood for organized convective events that may propagate westward/southwestward off the higher terrain in central and southeastern Arizona. Additional features that may enhance possibilities for severe weather are outflow boundaries from decaying mesoscale convective systems in northern Mexico as well as deeper gulf surges that have traveled the entire length of the Gulf of California (e.g., Zehnder 2004; Newman and Johnson 2012). These convective outflows from mesoscale convective systems and synoptic-scale features (i.e., inverted troughs, easterly waves, and monsoon-ridge positioning) appear to be linked (e.g., Mejía et al. 2016). As noted by Wallace et al. (1999) and supported by Mejía et al. (2016), however, the relationship between gulf surges and the incidence of severe weather in Arizona is not straightforward. Mejía et al. (2016) state that during the strongest gulf surges operational forecasters in Phoenix have observed that the likelihood of thunderstorms may actually be reduced. Although a surge brings in more humid air that may increase potential instability, it may cool the boundary layer and thus require more daytime heating and deeper lifting to release the instability.
c. Severe-weather-event-day observed precipitation
To capture the precipitation associated with these synoptic patterns, the total daily precipitation was averaged over a 3-day time frame for each burst event for the dates with the highest positive z scores (above 85%, or top 15%) relating to the associated combined PC time series. The precipitation anomaly patterns are shown in Fig. 8. The anomaly is constructed as the average precipitation for the top 15% of the events in a given monsoon-related severe weather mode minus the average precipitation of all the identified severe-weather-event days. The top events for both type-I and type-II modes exhibit even more enhanced precipitation across all of Arizona and New Mexico, exceeding >1 mm day−1 above the average of all identified severe weather events. The pattern of precipitation associated with the average of all severe-weather-event days in Fig. 5 would indicate that maximum precipitation values centered on the peaks of the terrain, where monsoon precipitation is most climatologically favored to occur, but the top precipitation events within this set for the strongest expressions of the type-I and type-II modes show maximum positive precipitation anomalies (above the severe-weather-event average) tending also to occur to the south and west of the Mogollon Rim in Arizona, where organized convection plays a relatively more important role in monsoon precipitation (e.g., Hales 1977; Castro et al. 2007a).
Precipitation anomalies (mm day−1) associated with the objectively defined type-I and type-II severe-weather-event days from mode reconstruction. Anomalies were constructed by averaging precipitation for events that project most strongly on combined EOF modes (top 15%) minus the average precipitation for all thermodynamically favorable severe weather events.
Citation: Journal of Applied Meteorology and Climatology 55, 11; 10.1175/JAMC-D-16-0075.1
To examine whether these modes are associated with organized, propagating convection, we constructed Hovmöller diagrams of the hourly stage IV derived precipitation for each year. To provide an illustrative example, we show the Hovmöller diagram (Fig. 9) over the longitude of Arizona averaged across the latitude band 31.5°–37°N of the stage IV product for the monsoon of 2006. The period of 2006–08 consists of several consecutive years of active, wet monsoons in Arizona, per the records maintained by the Tucson office of the National Weather Service and the stage IV product averaged over Arizona in Fig. 10. The solid horizontal lines on the plot in Fig. 9 indicate days that were identified as thermodynamically favorable severe-weather-event days. The purple horizontal lines indicate days that fall in the top 15% of days for the type-I mode; the light-orange horizontal lines indicate days that fall in the top 15% of days for the type-II mode. Periods with the type-I and type-II modes being most apparent have propagating convection all the way to the Colorado River valley (located at approximately 114°W). Note that there is a gap in stage IV data from early to mid-September of this year.
Hovmöller diagram (in the west–east direction) of daily precipitation during 2006 over the longitudinal extent of Arizona. Precipitation is averaged over the latitude band 31.5°–37°N.
Citation: Journal of Applied Meteorology and Climatology 55, 11; 10.1175/JAMC-D-16-0075.1
Average monsoon precipitation (June–September average; mm day−1) in Arizona for the thermodynamically favorable severe weather events (CAPE–PWV) selected cases for 2002–10, from the stage IV product. The number of total CAPE–PWV cases for each year is shown at the top.
Citation: Journal of Applied Meteorology and Climatology 55, 11; 10.1175/JAMC-D-16-0075.1
Characteristics of organized, propagating convection are examined for two specific events, one each for the type-I and type-II modes. The identified event for the type-I mode is 27–31 July 2006. Figure 11 shows the Hovmöller diagrams for this period in Arizona in the meridional direction, averaged over latitude band 31.5°–37°N, and the zonal direction, averaged over longitude 116°–110°W. For this event, organized convective systems each day are generally propagating to the south and west from their point of origin on the Mogollon Rim (around 110°W). A sample approximate propagation velocity for the convective system that occurred from 0000 to 1200 UTC 29 July is calculated from the yellow dots on the plot. The convective system on this day is propagating at ~12 m s−1 (26 mi h−1) to the south and west. The identified event for the type-II mode is 3–7 July 2006. Figure 12 shows the Hovmöller diagrams for this day. Although the westward propagation speed for the type-II event is very similar to that of the type-I event, in this case the organized convection identified during the period of 0000–1200 UTC 5 July is generally propagating in a northwesterly direction from its point of initiation, at nominally the same speed of the other type-I event. Note that our main objective here in these examples is solely to illustrate that, in fact, organized, westward-propagating convection in Arizona is present in the strongest type-I or type-II objectively identified events in a sample active monsoon year, supporting our original hypothesis. We are not necessarily inferring a climatologically preferred direction for convective propagation in type-I and type-II severe weather events on the basis of just these two examples presented, although we acknowledge further study on the stage IV data is probably warranted.
Hovmöller diagrams of precipitation rate in Arizona from stage IV data from 1200 UTC 27 Jul to 1200 UTC 31 Jul 2006, as an example of a type-I severe weather event. Precipitation is averaged (left) latitudinally and (right) longitudinally, with bounds as indicated. A sample phase speed (u or υ component) of propagating convection (m s−1) that reaches the Colorado River valley is included, calculated from the slope of the solid black line.
Citation: Journal of Applied Meteorology and Climatology 55, 11; 10.1175/JAMC-D-16-0075.1
As in Fig. 11, but for a type-II severe weather event from 1200 UTC 3 Jul to 1200 UTC 7 Jul 2006.
Citation: Journal of Applied Meteorology and Climatology 55, 11; 10.1175/JAMC-D-16-0075.1
5. Conclusions, discussion, and future research directions
We emphasize that this study represents an initial step in a longer-term research process to objectively identify favorable NAM severe-weather-event days in the context of real-time operational forecasting and analysis of regional climate model data. We characterize the thermodynamic environment on the basis of the level of atmospheric instability (MUCAPE) and moisture (PWV) during June–September, as based on the observed Tucson sounding data (1993–2010). Maddox et al. (1995) similarly used the Tucson sounding data. Favorable thermodynamic conditions are the first-order criterion that an operational weather forecaster in the southwestern United States would use to forecast monsoon thunderstorms. We use a threshold of the top 30% of MUCAPE and PWV days in a joint probability distribution to define the thermodynamically favorable severe-weather-event days. The MUCAPE and PWV threshold values are above the nominal operational thresholds used to flag potentially severe monsoon weather days in Arizona. The severe-weather-event days also correspond to nearly 60% of the observed severe weather reports and NWS warnings in southern Arizona during a 5-yr period.
An EOF analysis was performed on daily 500-hPa geopotential height to objectively define the synoptic-scale patterns that are associated with severe weather events, with respect to the dates of the thermodynamically favorable, severe-weather-event days. The first two dominant modes are essentially the Maddox et al. (1995) type-I and type-II monsoon-related severe weather patterns in Arizona. They generally reflect a monsoon ridge that is centered over the south-central United States in eastern New Mexico, Texas, and Oklahoma (type I) or over the Great Basin in Utah and Nevada (type II). Both of these monsoon-ridge orientations represent a general displacement of the monsoon ridge to the northeast or northwest, respectively, from its climatological position over the Four Corners area of the southwestern United States in mid-to-late summer. These severe weather synoptic patterns frequently result in large convective outbreaks during the so-called burst periods of the monsoon. On the days for which these patterns are most apparent, in the hourly stage IV data there is clear evidence of organized, propagating convection that traverses from east to west across Arizona. The method that was developed here utilizes an objective approach to identify and classify the severe-weather-event days on the basis of the most thermodynamically favorable conditions and how the 500-hPa geopotential-height anomaly patterns project on the Maddox et al. type-I and type-II modes. The fact that our objectively based approach essentially recovers the two monsoon-related severe-weather-event modes that were originally identified by B. Maddox, who detected them by conducting a subjective analysis of many synoptic weather maps over a series of years, is a true testament to his skill as a synoptic meteorologist.
In the framework of operational forecasting, we envision that the “climatology” that we have developed here could potentially be further developed into an objective, statistically based algorithm to flag automatically the thermodynamically favorable severe-weather-event days in the southwestern United States and to classify the monsoon ridge as a type-I or type-II pattern. For short-term forecasting in Tucson, Arizona (where the radiosonde data used in this study were derived), such an automated algorithm would operate in the following sequence of steps: MUCAPE and PWV would be computed from the morning (1200 UTC) sounding data. The day would be flagged as thermodynamically favorable for severe weather if these values fall in the top 30% of a climatological joint distribution. If the day is flagged as thermodynamically favorable, an index score for the type-I and type-II severe weather modes can then be computed by projecting the observed (analysis) map for 500-hPa height anomaly onto the first two EOFs from the historical severe-weather-event climatology. Such a simple algorithm could be very quickly executed, immediately after the morning sounding data are processed, in time to inform the discussion for the daily weather forecast. An index score for the Maddox severe weather pattern types would be very similar in principle to the use of existing operational convective indices (K index, Showalter index, etc.) to provide local operational forecast guidance and thus would help to standardize operational forecast procedures during monsoon-related severe weather events. There is also no reason why such an algorithm could not be used to identify the potential skill for objectively forecasting monsoon-related severe weather events beyond a 1-day forecast, although we strongly suggest that rigorous testing on previous forecast (hindcast) data would be necessary prior to any operational forecast use. Although all of the analyses in this work and that of Maddox et al. (1995) have been based on the use of just the Tucson sounding, the thermodynamically favorable days for severe weather during the monsoon are spatially coherent throughout the Southwest (Jares et al. 2014). So the same analysis procedures as were used here could be applied to other radiosonde observing sites in the region (e.g., Flagstaff, Arizona, or Las Vegas, Nevada), and we suspect that very similar results would be obtained. Jares et al. (2014) also document the spatial patterns of CAPE and PWV that are associated with severe-weather-event days in the southwestern United States in the context of a long-term simulation by a regional climate model that dynamically downscales an atmospheric reanalysis (Castro et al. 2012).
What further work can be done to establish firmer physical links between these thermodynamically favorable days and severe monsoon weather? We mentioned in the introduction that these monsoon-related severe weather 500-hPa height patterns are frequently associated with the passage of inverted troughs. Although this work does not explicitly relate inverted troughs to the monsoon-ridge positioning or monsoon precipitation, this has been done in other studies by the authors (Bieda et al. 2009; Lahmers et al. 2016). The passage of inverted troughs would facilitate convective organization by favorable vertical wind shear profile and/or synoptic-scale (i.e., quasigeostrophic) upward vertical motion (Finch and Johnson 2010; Newman and Johnson 2012). The modeling study by Newman and Johnson (2012) suggests that it is more the mesoscale effects of the vertical wind shear that facilitate convective organization for a case of development of a mesoscale convective system during the North American Monsoon Experiment. In a companion work by the authors (Lahmers et al. 2016), objective identification of inverted troughs that track across the NAM domain, done by using near-tropopause characterization of potential vorticity anomalies, presents additional information on the nature of convective organization within the stage IV data. Lahmers et al. (2016) and Bieda et al. (2009) found that the passage of IVs through the Southwest is closely linked to the occurrence of organized, propagating, convection, similar to what we have found here in association with the type-I and type-II severe-weather-event modes. We used Tucson sounding data and inverted-trough days that were identified in the long-term dynamically downscaled reanalysis used in Lahmers et al. (2016) to assess the correspondence between 1) the thermodynamically favorable severe-weather-event days (as defined in this study) with and without inverted troughs and 2) easterly upper-level winds (at 400 hPa) and easterly shear (from the surface to 450 hPa). The results in Table 1, shown as percentages and for the moving time windows that they used, indicate that the thermodynamically favorable severe-weather-event days with inverted troughs in the southwestern United States tend to occur more in a regime of upper-level easterly flow and/or easterly wind shear, as compared with no-trough days. This is especially true during the onset period of the monsoon (from late June to early July). We therefore suggest that the additional development of an idealized vertical wind shear profile associated with severe-weather-event days would provide helpful guidance in forecasting convective outbreaks during the NAM, from either in situ sounding data or numerical weather prediction model output (Pytlak et al. 2005).
Percentage of thermodynamically favorable days coinciding with easterly flow (at 450 hPa), with easterly shear (from the surface to 400 hPa), and with both characteristics in Tucson sounding data (1950–2010) for all thermodynamically favorable days, no-trough days, and trough days. The inverted troughs were objectively identified from a long-term dynamically downscaled atmospheric reanalysis over the contiguous United States in a companion publication by the authors (Lahmers et al. 2016).
The monsoon ridge positioning reflects mainly easterly transport of moisture from the Gulf of Mexico at mid- and upper levels of the atmosphere, specifically above 700 hPa (e.g., Schmitz and Mullen 1996; Hu and Dominguez 2015). While the Gulf of Mexico does play a role in increasing the upper-air moisture, increased moisture and thermodynamic instability due to low-level moisture transport from the Gulf of California, eastern tropical Pacific, and Mexico itself are at least equally important in Arizona (Adams and Comrie 1997; Maddox et al. 1995; Hu and Dominguez 2015). To be specific, the mechanism of the gulf surge (e.g., Zehnder 2004) transports low-level moisture into southern Arizona, west of the Mogollon Rim. Methods for objectively identifying gulf-surge events have shown that they do coincide with episodes of enhanced monsoon precipitation over the Southwest (e.g., Bordoni and Stevens 2006). Severe thunderstorms in Arizona associated with high levels of instability and atmospheric moisture often occur in the absence of a gulf surge, however (Wallace et al. 1999). Mejía et al. (2016) recently related the occurrence of gulf surges to a satellite-derived climatology of mesoscaled convective systems in the North American monsoon region. They found that mesoscale convective system activity enhances the offshore flow along the eastern Gulf of California coast, which then enhances the Gulf of California low-level jet. Satellite composites show that gulf surges are associated with enhanced convective activity that propagates along the coastal plain of the gulf.
It is also important to recognize that the actual evolution of severe weather during the NAM in the southwestern United States is intimately linked to physical processes on the mesoscale that are difficult, if not impossible, to quantify in the context of long-term historical observational data or a retrospective regional atmospheric reanalysis, such as the NARR product used here. Exactly how and where convection is initiated on the terrain during the day depends on reaching the convective temperature. The atmosphere may be “too wet to rain”—a colloquial term used by local operational forecasters to refer to days with relatively high amounts of atmospheric moisture and extensive morning cloudiness. The morning clouds, typically debris clouds from convection on the previous day, suppress the amount of solar radiation that reaches the surface so that the convective temperature is not able to be reached during the afternoon (Adams and Souza 2009). Once NAM convection develops and propagates away from the terrain, cold-pool dynamics and propagation of outflow boundaries become important to initiation of new regions of convection (e.g., Corfidi 2003; Nesbitt et al. 2008). Some of the most severe storms are caused by convergence of multiple outflow boundaries from different thunderstorms. So even when the atmosphere is highly thermodynamically conducive to convection on the synoptic scale, organized convection would be highly subject to the timing of convection and the subsequent outflow boundary formation.
Because the convective organization process that was just described is fundamentally on the meso-γ scale (on the order of 1 km), we would assert that numerical atmospheric modeling at this scale is absolutely necessary to study the climatology of severe monsoon thunderstorms. Convective-permitting modeling on the meso-γ scale is necessary to represent features associated with MCSs, including squall lines, gust fronts, and stratiform precipitation (e.g., Klemp 2006). Our upcoming work will demonstrate the value added by convective-permitting modeling by simulating thermodynamically favorable severe-weather-event days as identified in sources of data from regional climate models. These sources will include a retrospective dynamically downscaled long-term atmospheric reanalysis and dynamically downscaled data from models of global climate change projection that were used in phases 3 and 5 of the Coupled Model Intercomparison Project. Our ultimate objective will be to assess how monsoon thunderstorms may be changing in association with anthropogenic climate change.
Acknowledgments
This research is supported by the Strategic Environmental Research and Development Program (SERDP), Project RC-2205, through the U.S. Departments of Defense and Energy and the U.S. Environmental Protection Agency. Additional support for contributions by Dr. David Adams was provided through the Universidad Nacional Autónoma de México Programa de Apoyo a Proyectos de Investigatión e Innovación Tecnológica(UNAM PAPIIT; Grant IA100916). The authors are grateful for the assistance of Dr. Erick Rivera, Ms. Jennifer Stutler, Ms. Megan Jares, and Mr. Timothy Lahmers. The methodological approach reflects valuable input from operational forecasters at the Tucson NWS WFO. Comments and suggestions from three anonymous reviewers substantially improved the quality of the typescript.
REFERENCES
Adams, D. K., and A. C. Comrie, 1997: The North American monsoon. Bull. Amer. Meteor. Soc., 78, 2197–2213, doi:10.1175/1520-0477(1997)078<2197:TNAM>2.0.CO;2.
Adams, D. K., and E. P. Souza, 2009: CAPE and convective events in the Southwest during the North American monsoon. Mon. Wea. Rev., 137, 83–98, doi:10.1175/2008MWR2502.1.
Adams, D. K., S. I. Gutman, K. L. Holub, and D. S. Pereira, 2013: GNSS observations of deep convective time scales in the Amazon. Geophys. Res. Lett., 40, 2818–2823, doi:10.1002/grl.50573.
Bieda, S. W., III, C. L. Castro, S. L. Mullen, A. C. Comrie, and E. Pytlak, 2009: The relationship of transient upper-level troughs to variability of the North American monsoon system. J. Climate, 22, 4213–4227, doi:10.1175/2009JCLI2487.1.
Blanchard, D. O., 2011: Supercells in environments with atypical hodographs. Wea. Forecasting, 26, 1075–1083, doi:10.1175/WAF-D-11-00012.1.
Bordoni, S., and B. Stevens, 2006: Principal component analysis of the summertime winds over the Gulf of California: A gulf surge index. Mon. Wea. Rev., 134, 3395–3414, doi:10.1175/MWR3253.1.
Bretherton, C. S., C. Smith, and J. M. Wallace, 1992: An intercomparison of methods for finding coupled patterns in climate data. J. Climate, 5, 541–560, doi:10.1175/1520-0442(1992)005<0541:AIOMFF>2.0.CO;2.
Bretherton, C. S., M. E. Peters, and L. E. Back, 2004: Relationships between water vapor path and precipitation over the tropical oceans. J. Climate, 17, 1517–1528, doi:10.1175/1520-0442(2004)017<1517:RBWVPA>2.0.CO;2.
Bright, D. R., and S. L. Mullen, 2002: The sensitivity of the numerical simulation of the Southwest monsoon boundary layer to the choice of PBL turbulence parameterization in MM5. Wea. Forecasting, 17, 99–114, doi:10.1175/1520-0434(2002)017<0099:TSOTNS>2.0.CO;2.
Carleton, A. M., 1986: Synoptic-dynamic character of ‘bursts’ and ‘breaks’ in the south-west U.S. summer precipitation singularity. J. Climatol., 6, 605–623, doi:10.1002/joc.3370060604.
Castro, C. L., R. A. Pielke Sr., and J. O. Adegoke, 2007a: Investigation of the summer climate of the contiguous United States and Mexico using the Regional Atmospheric Modeling System (RAMS). Part I: Model climatology (1950–2002). J. Climate, 20, 3844–3865, doi:10.1175/JCLI4211.1.
Castro, C. L., R. A. Pielke Sr., J. O. Adegoke, S. D. Schubert, and P. J. Pegion, 2007b: Investigation of the summer climate of the contiguous United States and Mexico using the Regional Atmospheric Modeling System (RAMS). Part II: Model climate variability. J. Climate, 20, 3866–3887, doi:10.1175/JCLI4212.1.
Castro, C. L., H.-I. Chang, F. Dominguez, C. Carrillo, J.-K. Schemm, and H.-M. H. Juang, 2012: Can a regional climate model improve the ability to forecast the North American monsoon? J. Climate, 25, 8212–8237, doi:10.1175/JCLI-D-11-00441.1.
Ciancarelli, B., C. L. Castro, C. Woodhouse. F. Dominguez, H.-I. Chang, C. Carrillo, and D. Griffin, 2014: Dominant patterns of US warm season precipitation variability in a fine resolution observational record, with focus on the Southwest. Int. J. Climatol., 34, 687–707, doi:10.1002/joc.3716.
Corfidi, S. F., 2003: Cold pools and MCS propagation: Forecasting the motion of downwind-developing MCSs. Wea. Forecasting, 18, 997–1017, doi:10.1175/1520-0434(2003)018<0997:CPAMPF>2.0.CO;2.
Ding, Q., B. Wang, J. M. Wallace, and G. Branstator, 2011: Tropical–extratropical teleconnections in boreal summer: Observed interannual variability. J. Climate, 24, 1878–1896, doi:10.1175/2011JCLI3621.1.
Doswell, C. A., and E. N. Rasmussen, 1994: The effect of neglecting the virtual temperature correction on CAPE calculations. Wea. Forecasting, 9, 625–629, doi:10.1175/1520-0434(1994)009<0625:TEONTV>2.0.CO;2.
Douglas, A. V., and P. J. Englehart, 2007: A climatological perspective of transient synoptic features during NAME 2004. J. Climate, 20, 1947–1954, doi:10.1175/JCLI4095.1.
Douglas, M. W., R. A. Maddox, K. Howard, and S. Reyes, 1993: The Mexican monsoon. J. Climate, 6, 1665–1677, doi:10.1175/1520-0442(1993)006<1665:TMM>2.0.CO;2.
Emanuel, K. A., 1994: Atmospheric Convection. Oxford University Press, 592 pp.
Finch, Z. O., and R. H. Johnson, 2010: Observational analysis of an upper-level inverted trough during the 2004 North American Monsoon Experiment. Mon. Wea. Rev., 138, 3540–3555, doi:10.1175/2010MWR3369.1.
Griffiths, P. G., C. S. Magirl, R. H. Webb, E. Pytlak, P. A. Troch, and S. W. Lyon, 2009: Spatial distribution and frequency of precipitation during an extreme event: July 2006 mesoscale convective complexes and floods in southeastern Arizona. Water Resour. Res., 45, W07419, doi:10.1029/2008WR007380.
Hales, J. E., Jr., 1977: On the relationship of convective cooling to nocturnal thunderstorms at Phoenix. Mon. Wea. Rev., 105, 1609–1613, doi:10.1175/1520-0493(1977)105<1609:OTROCC>2.0.CO;2.
Hu, H., and F. Dominguez, 2015: Evaluation of oceanic and terrestrial sources of moisture for the North American monsoon using numerical models and precipitation stable isotopes. J. Hydrometeor., 16, 19–35, doi:10.1175/JHM-D-14-0073.1.
Janiga, M. A., and C. D. Thorncroft, 2014: Convection over tropical Africa and the east Atlantic during the West African monsoon: Regional and diurnal variability. J. Climate, 27, 4159–4188, doi:10.1175/JCLI-D-13-00449.1.
Jares, M., C. L. Castro, H. I. Chang, C. Carrillo, J. J. Mazon, J. Stutler, and J. J. Brost, 2014: High resolution WRF simulation and climatological analysis of severe weather events during the North American monsoon. 26th Conf. on Weather Analysis and Forecasting/22nd Conf. on Numerical Weather Prediction, Atlanta, GA, Amer. Meteor. Soc., 151. [Available online at https://ams.confex.com/ams/94Annual/webprogram/Paper237080.html.]
Klemp, J. B., 2006: Advances in the WRF Model for convection-resolving forecasting. Adv. Geosci., 7, 25–29, doi:10.5194/adgeo-7-25-2006.
Kursinski, E. R., and Coauthors, 2008a: Water vapor and surface observations in northwestern Mexico during the 2004 NAME enhanced observing period. Geophys. Res. Lett., 35, L03815, doi:10.1029/2007GL031404.
Kursinski, E. R., D. K. Adams, and M. Leuthold, 2008b: GPS observations of precipitable water and implications for the predictability of precipitation during the North American monsoon. CLIVAR Exchanges, No. 42, International CLIVAR Project Office, Southampton, United Kingdom, 14–21. [Available online at http://www.clivar.org/sites/default/files/documents/Exchanges45.pdf.]
Kutzbach, J. E., 1967: Empirical eigenvectors of sea-level pressure, surface temperature, and precipitation complexes over North America. J. Appl. Meteor., 6, 791–802, doi:10.1175/1520-0450(1967)006<0791:EEOSLP>2.0.CO;2.
Lahmers, T. M., C. L. Castro, D. K. Adams, Y. L. Serra, J. J. Brost, and T. Luong, 2016: Long-term changes in the climatology of transient inverted troughs over the North American monsoon region and their effects on precipitation. J. Climate, 29, 6037–6064, doi:10.1175/JCLI-D-15-0726.1.
Li, J., S. Sorooshian, W. Higgins, X. Gao, B. Imam, and K. Hsu, 2008: Influence of spatial resolution on diurnal variability during the North American monsoon. J. Climate, 21, 3967–3988, doi:10.1175/2008JCLI2022.1.
Lu, E., X. Zeng, Z. Jiang, Y. Wang, and Q. Zhang, 2009: Precipitation and precipitable water: Their temporal-spatial behaviors and use in determining monsoon onset/retreat and monsoon regions. J. Geophys. Res., 114, D23105, doi:10.1029/2009JD012146.
Maddox, R. A., D. M. McCollum, and K. W. Howard, 1995: Large-scale patterns associated with severe summertime thunderstorms over central Arizona. Wea. Forecasting, 10, 763–778, doi:10.1175/1520-0434(1995)010<0763:LSPAWS>2.0.CO;2.
Magirl, C. S., and Coauthors, 2007: Impact of recent extreme Arizona storms. Eos, Trans. Amer. Geophys. Union, 88, 191–193, doi:10.1029/2007EO170003.
McCollum, D. M., R. A. Maddox, and K. W. Howard, 1995: Case study of a severe mesoscale convective system in central Arizona. Wea. Forecasting, 10, 643–665, doi:10.1175/1520-0434(1995)010<0643:CSOASM>2.0.CO;2.
Mejía, J. F., M. W. Douglas, and P. J. Lamb, 2016: Observational investigation of relationships between moisture surges and mesoscale- to large-scale convection during the North American monsoon. Int. J. Climatol., 36, 2555–2569, doi:10.1002/joc.4512.
Mesinger, F., and Coauthors, 2006: North American Regional Reanalysis. Bull. Amer. Meteor. Soc., 87, 343–360, doi:10.1175/BAMS-87-3-343.
Minjarez-Sosa, C. M., C. L. Castro, K. L. Cummins, E. P. Krider, and J. Waissmann, 2012: Toward development of improved QPE in complex terrain using cloud-to-ground lightning data: A case study for the 2005 monsoon in southern Arizona. J. Hydrometeor., 13, 1855–1873, doi:10.1175/JHM-D-11-0129.1.
Moncrieff, M. W., and M. J. Miller, 1976: The dynamics and simulation of tropical cumulonimbus and squall lines. Quart. J. Roy. Meteor. Soc., 102, 373–394, doi:10.1002/qj.49710243208.
Neelin, J. D., O. Peters, and K. Hales, 2009: The transition to strong convection. J. Atmos. Sci., 66, 2367–2384, doi:10.1175/2009JAS2962.1.
Nesbitt, S. W., D. J. Gochis, and T. J. Lang, 2008: The diurnal cycle of clouds and precipitation along the Sierra Madre Occidental observed during NAME-2004: Implications for warm season precipitation estimation in complex terrain. J. Hydrometeor., 9, 728–743, doi:10.1175/2008JHM939.1.
Newman, A., and R. H. Johnson, 2012: Mechanisms for precipitation enhancement in a North American monsoon upper-tropospheric trough. J. Atmos. Sci., 69, 1775–1792, doi:10.1175/JAS-D-11-0223.1.
North, G. R., T. L. Bell, R. F. Cahalan, and F. J. Moeng, 1982: Sampling errors in the estimation of empirical orthogonal functions. Mon. Wea. Rev., 110, 699–706, doi:10.1175/1520-0493(1982)110<0699:SEITEO>2.0.CO;2.
Prohaska, J. T., 1976: A technique for analyzing the linear relationships between two meteorological fields. Mon. Wea. Rev., 104, 1345–1353, doi:10.1175/1520-0493(1976)104<1345:ATFATL>2.0.CO;2.
Pytlak, E., M. Goering, and A. Bennett, 2005: Upper-tropospheric troughs and their interaction with the North American monsoon. 19th Conf. on Hydrology, San Diego, CA, Amer. Meteor. Soc., JP2.3. [Available online at https://ams.confex.com/ams/pdfpapers/85393.pdf.]
Ray, A. J., G. M. Garfin, M. Wilder, M. Vásquez-León, M. Lenart, and A. C. Comrie, 2007: Applications of monsoon research: Opportunities to inform decision making and reduce regional vulnerability. J. Climate, 20, 1608–1627, doi:10.1175/JCLI4098.1.
Richman, M. B., 1986: Rotation of principal components. J. Climatol., 6, 293–335, doi:10.1002/joc.3370060305.
Schmidli, R. J., 1986: Climate of Phoenix, Arizona. NOAA Tech. Memo. NWS WR-177, 146 pp. [Available online at https://www.weather.gov/media/wrh/online_publications/TMs/TM-177.pdf.]
Schmitz, J. T., and S. L. Mullen, 1996: Water vapor transport associated with the summertime North American monsoon as depicted by ECMWF analyses. J. Climate, 9, 1621–1634, doi:10.1175/1520-0442(1996)009<1621:WVTAWT>2.0.CO;2.
Serra, Y. L., and Coauthors, 2016: The North American Monsoon GPS Transect Experiment 2013. Bull. Amer. Meteor. Soc., doi:10.1175/BAMS-D-14-00250.1, in press.
Smith, W. P., and R. L. Gall, 1989: Tropical squall lines of the Arizona monsoon. Mon. Wea. Rev., 117, 1553–1569, doi:10.1175/1520-0493(1989)117<1553:TSLOTA>2.0.CO;2.
Trapp, R. J., E. D. Robinson, M. E. Baldwin, N. S. Diffenbaugh, and B. R. J. Schwedler, 2011: Regional climate of hazardous convective weather through high-resolution dynamical downscaling. Climate Dyn., 37, 677–688, doi:10.1007/s00382-010-0826-y.
Wall, C. L., E. J. Zipser, and C. Liu, 2012: A regional climatology of monsoonal precipitation in the southwestern United States using TRMM. J. Hydrometeor., 13, 310–323, doi:10.1175/JHM-D-11-031.1.
Wallace, C. E., R. A. Maddox, and K. W. Howard, 1999: Summertime convective storm environments in central Arizona: Local observations. Wea. Forecasting, 14, 994–1006, doi:10.1175/1520-0434(1999)014<0994:SCSEIC>2.0.CO;2.
Wallace, J. M., and D. S. Gutzler, 1981: Teleconnections in the geopotential height field during the Northern Hemisphere winter. Mon. Wea. Rev., 109, 784–812, doi:10.1175/1520-0493(1981)109<0784:TITGHF>2.0.CO;2.
Wallace, J. M., C. Smith, and C. S. Bretherton, 1992: Singular value decomposition of wintertime sea surface temperature and 500-mb height anomalies. J. Climate, 5, 561–576, doi:10.1175/1520-0442(1992)005<0561:SVDOWS>2.0.CO;2.
Wood, K. M., and E. A. Ritchie, 2013: An updated climatology of tropical cyclone impacts on the southwestern United States. Mon. Wea. Rev., 141, 4322–4336, doi:10.1175/MWR-D-13-00078.1.
Zehnder, J. A., 2004: Dynamic mechanisms of the gulf surge. J. Geophys. Res., 109, D10107, doi:10.1029/2004JD004616.
Zhang, J., and Coauthors, 2011: National Mosaic and Multi-Sensor QPE (NMQ) system: Description, results, and future plans. Bull. Amer. Meteor. Soc., 92, 1321–1338, doi:10.1175/2011BAMS-D-11-00047.1.