The influence of low-level inflow wind and its high-frequency variability on the spatial characteristics of the eastern margin of the South Pacific Convergence Zone (SPCZ) is examined. Compositing daily and 5-day mean low-level wind, precipitation, and tropospheric moisture data reveals a clear relationship between high-frequency zonal inflow variations and the eastern SPCZ margin, with relaxation of trade wind intensity associated with increased moisture near the mean eastern SPCZ margin and an eastward displacement of the convection. An idealized 2-dimensional model demonstrates that variations in dry air inflow cause tropospheric moisture and precipitation variations akin to those observed. In this prototype, factors affecting the extent of SPCZ variability are also important to the mean margin position. SPCZ margin shifts under natural variability thus offer an observational target against which to evaluate simulated interactions of inflow air mass characteristics with convection.
 The South Pacific Convergence Zone (SPCZ) consists of low-level convergence and strong convective precipitation extending from the equatorial western Pacific near 140°E southeastward to 120°W and 30°S, where it exhibits a northwest–southeast orientation [Kiladis et al., 1989; Vincent, 1994] (for reference, the auxiliary material provides the January SPCZ region climatology). Despite the SPCZ's significance to tropical climate, many questions remain regarding its dynamics, and model simulation of the SPCZ is often problematic [Mechoso et al., 1995; Lin, 2007]. Common model deficiencies include an excessively zonal SPCZ and deep convection penetrating too far eastward, producing a spurious “double ITCZ” [Zhang, 2001]. Such behavior may stem from deficiencies in ocean-atmosphere coupling, including errors in ocean dynamical processes and their effect on surface fluxes or SST gradients, parameterization of low-level marine stratus clouds, and interactions between continental topography and the large-scale circulation [e.g., Yu and Mechoso, 1999; Li et al., 2004; Meehl et al., 2005; Sun et al., 2006; Takahashi and Battisti, 2007; R. Wood and VOCALS Scientific Working Group, VOCALS—Regional experiment: Scientific programme overview, 2006, available at http://www.eol.ucar.edu/projects/vocals].
 Regionally, some of the most pronounced biases occur where the southeasterly trade winds enter the convection zone, i.e., in the transition between the eastern Pacific descent zone and the SPCZ. The significance of ventilation of the margins of convection zones by dry air inflow has been previously examined in model studies of monsoons, ENSO teleconnections and global warming [Chou and Neelin, 2001; Neelin et al., 2003; Neelin and Su, 2005] and of the SPCZ [Takahashi and Battisti, 2007] and through interactions of air mass properties and tropical convection in observations [Parsons et al., 2000; Back and Bretherton, 2006] and idealized cloud-resolving models [Tompkins, 2001; Derbyshire et al., 2004]. Lintner and Neelin  (hereinafter referred to as LN07) developed a prototype for the behavior of inflow convective margins—i.e., margins with a substantial flow component from a drier region toward the margin—and applied this to analyze the impact of El Niño on tropical South America precipitation. In a similar vein, we investigate here the possible impacts of anomalous low-level inflow on the moisture and precipitation characteristics of the SPCZ.
 Inflow wind variations may affect how far east the SPCZ-related convection occurs. For example, in models that produce insufficiently large zonal moisture gradients or that lack appropriate parameterized convective sensitivity to tropospheric moisture, the SPCZ may be anticipated to extend too far eastward. As a first step in addressing such modeling biases, we aim to establish a baseline for observed variations in low-level circulation and their relationship to moisture and precipitation along the eastern SPCZ convective margin at time scales of atmospheric internal variability. Using daily and 5-day mean data permits examination of such relationships in satellite data sets with reasonable statistical confidence and provides a potential model-data comparison independent of the climatology. This can aid in constraining climatological and low-frequency behavior to the extent that these depend on similar moisture-precipitation-inflow mechanisms.
2. Data Sets and Methodology
 We use high-frequency precipitation data from two satellite data sets. The first set, the Climate Prediction Center (CPC) Merged Analysis of Precipitation (CMAP) [Xie and Arkin, 1997], consists of 5-day (pentadal) means at a resolution of 2.5° × 2.5°. The CMAP data analyzed here span 1979–2006. The second is the Special Sensor Microwave Imager (SSM/I) [Hilburn and Wentz, 2007], consisting of twice-daily observations at 0.25° × 0.25° resolution that are aggregated here to daily averages for the period 1989–2006. The SSM/I retrieval also provides total column water vapor. Low-level winds are from the National Center for Environmental Prediction (NCEP) [Kalnay et al., 1996] Reanalysis (daily averages at 2.5° × 2.5° resolution). The NCEP Reanalysis further provides tropospheric specific humidity, used for the pentadal analysis, but with the caveat that this field is dependent on the reanalysis model.
 The data were analyzed using a simple composite approach, conditionally averaging fields on a low-level circulation index, namely the average of 925 mb zonal wind over 140°W–120°W and 20°S–10°S. The results described below were found to be robust to minor variations in the compositing index definition, e.g., similar behavior was obtained with the pressure level varied between 1000 mb and 850 mb. In order to capture the high-frequency behavior, the mean of the compositing index for each January (i.e., either 31 days for the SSM/I analysis or pentads 1–6 for the CMAP analysis) was first removed. Additionally, to minimize the influence of ENSO, which is known to impact the spatial characteristics of the SPCZ [Folland et al., 2002], those Januaries with either strong El Niño (NINO3 ≤ 1K; 1983, 1987, 1992, and 1998) or strong La Niña (NINO3 ≤ −1K; 1985, 1989, 1999, and 2000) conditions were excluded. The compositing index's standard deviation, σ, computed over all days/pentads, was used to define the positive (index ≥ σ) and negative (index ≤ −σ) phases.
3. Composite Analysis Applied to Observations
 As shown in Figure 1a, composite differences of pentadal NCEP reanalysis 850 mb specific humidity (q850; shading) and 925 mb winds (u925 and v925; arrows) suggest that anomalous low-level westerlies near the mean eastern SPCZ margin are associated with enhanced low-level moisture there. These differences extend beyond the area over which the compositing index is defined, e.g., the q850 anomalies extend to the southeast of 120°W into the dry descent zone. Also, the (positive) differences in u925 and v925 are seen to form the northern edge of a low-level cyclonic circulation centered around 125°W and 25°S.
 For the CMAP precipitation, the presence of anomalous westerlies in the averaging region is associated with enhanced precipitation along the eastern SPCZ margin (Figure 1b). Considering the behavior of the 4 mm day−1 contour, a proxy for the SPCZ margin, anomalous westerlies are associated with an eastward shift of the convection; spatially, this shift occurs slightly downwind (i.e., westward) of the region of enhanced q850. Moreover, during periods of anomalous low-level westerlies and an eastward-displaced SPCZ between 20°S–10°S, the southern edge of the SPCZ appears to shift northward, consistent with the low-level circulation and moisture changes occurring there.
 The principal features of the SSM/I daily analysis (Figure 2) agree broadly with the pentadal CMAP analysis. In particular, for anomalous westerly conditions, total column water vapor is elevated along the mean January axis of the eastern SPCZ margin (Figure 2a). As with the CMAP analysis, both the low-level wind and moisture differences extend well beyond the index area. The precipitation field (Figure 2b) exhibits a broad region of positive differences along the mean eastern SPCZ margin.
Figure 3 shows longitudinal cross-sections (averaged over 20°S–10°S) for the fields plotted in Figure 2. These cross-sections highlight the changes in low-level wind, moisture, and precipitation occurring along the eastern SPCZ and support the view that these changes occur principally as the result of a convective margin shift between anomalous westerly and easterly conditions. A peak low-level zonal wind difference of ∼7 m s−1 near 135°W, yields a 10° shift of the eastern SPCZ margin.
4. Analytic Prototype for the Effect of Low-Level Winds on the SPCZ
 As a simple prototype for the eastern SPCZ margin, we adapt the LN07 framework to 2D inflow from an oceanic nonconvecting subsidence region into a convecting region, using parameters consistent with the southeastern Pacific trade wind region. Details of the prototype are provided in the auxiliary material. We note that, in the absence of inflow-induced advective transport, precipitation would be constant over the domain, so the prototype highlights the potential significance of low-level inflow to the SPCZ. To compare the prototype solution to the observations, random, Gaussian-distributed zonal wind perturbations about the mean wind were considered. In the subtropical SPCZ, such perturbations may be generated by frontal passage and troughs associated with Rossby waves in the midlatitude westerly flow (G. N. Kiladis, personal communication, 2008).
 Averages of the prototype's moisture (q) and precipitation (P) fields over 1000 steady-state realizations appear in Figure 4a. Despite its simplicity, the prototype produces a reasonable mean spatial structure for the eastern boundary of the SPCZ. Horizontal advection, which is relatively dry because of the low values of q assumed along the southern and eastern domain boundaries, maintains the nonconvecting southeastern corner in Figure 4a. The transition from nonconvecting to convecting conditions, reached when the inflowing air has moistened sufficiently, yields a switch from large-scale divergent to convergent conditions. This interaction can maintain an angle of the SPCZ eastern margin with a wind component along the margin as the q field adapts until its gradients balance other terms in the moisture equation.
Figure 4b depicts composite differences for q and P using the zonal wind perturbations as the compositing index. Wind perturbations that oppose the mean flow produce an eastward shift in the SPCZ margin, in agreement with the results shown in Figures 1–3. In terms of the convective margins framework, a reduction of the inflow diminishes the import of relatively dry air toward the convection zone. The prototype assumes a Betts and Miller  moisture threshold, qc, for the onset of deep convection, consistent with recent empirical studies [e.g., Bretherton et al., 2004; Peters and Neelin, 2006]. Since q increases by evaporation, reductions of inflow wind allow q to reach qc sooner along the inflow trajectory, inducing the eastward margin shift.
 To underscore the implications for model sensitivity and simulation error, we repeated the analysis in Figures 4a and 4b but with qc lowered by 10% (Figures 4c and 4d). For the mean state (Figure 4c), the lowering of qc results in a reduction of the spatial extent of the dry zone, with the convective margins shifted closer to the southeastern corner of the domain. The variability (Figure 4d) undergoes a corresponding alteration: the width of the anomalies decreases approximately proportional to the shift in the edge of the convection zone (see auxiliary material).
5. Summary and Discussion
 The results presented here demonstrate the marked sensitivity of the eastern SPCZ when conditioned on low-level circulation variations. Simple composite analysis of daily and pentadal wind, precipitation, and tropospheric moisture data indicates a robust relationship between trade wind inflow and tropospheric moisture and precipitation. Consistent with previous convective margins theory, we argue that reduction in easterly trade wind inflow from the large-scale southeast Pacific descent region enhances moisture along the SPCZ by reducing the import of relatively dry air into the convection zone. The enhanced moisture conditions are accompanied by increased precipitation, which can be interpreted as a shift in the eastern SPCZ margin. A simplified prototype supports the plausibility of this interpretation.
 This analysis provides a step toward understanding model precipitation sensitivity biases, i.e., the response of simulated precipitation to climate perturbations. While such sensitivity biases are important in and of themselves, they also provide a means of suggesting revisions to address well-known model biases in the mean SPCZ. In the simple prototype, the position of the mean SPCZ margin and the width of the precipitation anomaly region caused by inflow wind variations are controlled by similar factors. The analysis of SPCZ margin shifts under natural variability thus provides a robust observational target. The corresponding quantities in simulations represent a potentially useful test diagnostic, notably one that illustrates sensitivity of the interaction of convection with the inflow air masses.
 The authors thank C. R. Mechoso and G. N. Kiladis for discussion and J. E. Meyerson for graphical assistance. This work was supported partly by National Oceanic and Atmospheric Administration NA05OAR4311134 and NA08AR4310882 and National Science Foundation ATM-0645200. JDN acknowledges sabbatical support from the J. S. Guggenheim Memorial Foundation and the National Center for Atmospheric Research.