1. Introduction
The intertropical convergence zone (ITCZ) is one of the most important components of the climate system. It is associated with the seasonal migration of rainfall in the tropics and the subtropical anticyclones. The ITCZ is characterized by the lower-tropospheric convergence of the northeasterly and southeasterly tropical trade winds and a zonally elongated band of deep convection. On average, air ascending in the deep convection moves poleward in the upper troposphere and descends in the subtropics; the ITCZ is essentially the ascending branch of the Hadley circulation. The precise location of the ITCZ affects the weather conditions over a broad region as it changes the position of convective rainfall at low latitudes and also the suppression of rainfall within the subtropical anticyclones.
The seasonal-mean rainfall from the National Oceanic and Atmospheric Administration (NOAA)/Climate Prediction Center (CPC) morphing technique (CMORPH; Joyce et al. 2004) for the period 1998–2012 is shown in Fig. 1. The time-mean ITCZ locations can be easily inferred by the zonally elongated regions of high rainfall within approximately 15° of the equator in both hemispheres. The day-to-day nature of the ITCZ can be highly variable (Wang and Magnusdottir 2006) because of interactions with monsoon systems and synoptic disturbances (including tropical cyclones). The aim of the current study is to present an objective, automated method for detecting the position and intensity of the ITCZ and to use this method to describe the climatology of these features and their recent trends in a global reanalysis dataset. This work is intended to provide a diagnostic tool for the analysis of climate models and a reanalysis-based benchmark for comparison.
Seasonal averages of CMORPH precipitation (mm day−1) for the period 1998–2012. (SON denotes the season September–November.)
Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00339.1
There is no agreed way to identify the location of the ITCZ (Sadler 1975). Some previous work has used regions of low cloud-top temperature (e.g., Waliser and Gautier 1993) or bands of high precipitation (e.g., Gu et al. 2005), whereas other studies have used dynamical fields, such as relative vorticity (e.g., Chan and Evans 2002), mean sea level pressure, or confluence. Each of these methods has their advantages and disadvantages, but all are able to locate the ITCZ. To the authors’ knowledge, there is only one existing global climatology of the ITCZ: that is the climatology by Waliser and Gautier (1993), who used infrared satellite imagery over a 17-yr period to describe the location of the ITCZ. This study confirmed much of the common knowledge of how the ITCZ location changes through the course of a year. However, it can be difficult to use a method like Waliser and Gautier’s to evaluate the performance of weather or climate models, as cloud-top temperatures may not be a standard output field and considerable manual interpretation may be required. Other studies have used rainfall to define large convergence zones in climate models [e.g., Brown et al. (2011) in examining the South Pacific convergence zone (SPCZ)] with crude linear fitting to join rainfall maxima in seasonal averages. This type of analysis is problematic, since a region of seasonally high precipitation is not exclusively associated with convergence zones and a linear fit of rainfall locations may not be applicable at shorter time scales and at all points on the globe. Additionally, the precise rainfall forecast depends on both the synoptic environment and the model formulation. Here, a more sophisticated objective method of deriving the ITCZ position is presented; this method uses lower-tropospheric convergence, which can be computed from standard dynamic fields from numerical weather prediction (NWP) or climate model output. The technique presented is fully automated and objective allowing large amounts of data (e.g., those produced for climate model intercomparison projects) to be easily processed and quantitatively compared.
2. Methodology
Previous studies (e.g., Sadler 1975) have shown that ITCZs are zonally orientated linear features characterized by convergence in the lower troposphere, with large-scale vigorous ascent and then divergence in the upper troposphere. Over the oceans, ITCZ locations coincide with a mean sea level pressure trough, but this relationship does not hold over landmasses, where the pressure field is dominated by thermally driven features, such as heat lows (e.g., Rácz and Smith 1999). The time-mean position of the ITCZ is well captured by a time averages of many fields including rainfall (see Fig. 1), low-level convergence, vorticity, midtropospheric vertical motion, etc. In this study, the objective ITCZ position is defined in NWP output from the layer mean divergence in the layer between 1000 and 850 hPa. This layer, rather than a single level, is chosen to allow the analysis to be carried out over both land and ocean. Because it is well known that weather systems (e.g., tropical cyclones, easterly waves, and even thunderstorm clusters) have their own divergence pattern, the divergence field used for a particular day is a time mean, as this reduced the impact of these transient features. An infrared satellite image of the tropical Pacific from August 2004 is shown in Fig. 2a with streamlines displaying the layer-mean wind between 1000 and 850 hPa, which are averaged for the proceeding and following 72 h. Across the eastern portion of the ocean basin (east of the dateline) there are scattered cold cloud tops (indicative of deep convection) that are closely aligned with confluent streamlines. West of the dateline, the convection is more organized and lies along a shear line (a streamline with strong cyclonic curvature) that lies west-northwest–east-southeast, with its western end near Taiwan. It is these two large-scale features in particular that would be recognized as ITCZs by synoptic analysts and what is desired to be detected automatically.
Example showing the steps involved in the objective ITCZ technique using ERA-Interim data centered on 1200 UTC 13 Aug 2004. (a) 1000–850-hPa layer-mean streamlines, averaged for 72 h before and after valid time, overlaid on infrared satellite imagery (shaded; K). (b) Divergence (scaled by 105 and colored according to legend) with solid contour showing where the gradient of divergence is equal to zero. (c) Second derivative of divergence (scaled by 1013 and colored according to legend) with solid contour showing where the gradient of divergence is equal to zero in convergent regions only. (d) 850-hPa θw (K; colored according to legend) with a black contour showing where the horizontal gradient of divergence is equal to zero only in convergent regions and where the second gradient of divergence exceeds zero. (e) Objective ITCZ locations (red contours) derived from the fields in (d) with 1000–850-hPa layer-mean streamlines, averaged for 72 h before and after valid time, overlaid on infrared satellite imagery (shaded; K).
Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00339.1
The core diagnostic for the objective detection of ITCZs is the location at which the magnitude of horizontal gradient of divergence is equal to zero in these time- and layer-averaged fields. This contour is overlaid on the 1000–850-hPa time-averaged divergence field (derived from the wind field shown in Fig. 2a), which is shaded in Fig. 2b. The lines where the gradient of divergence is equal to zero lines up precisely with the regions that correspond with the ITCZs determined subjectively by inspection of the infrared imagery and streamlines in Fig. 2a. However, this contour also highlights features that are clearly not ITCZs, as this diagnostic picks out both lines of maximum convergence and divergence in both the tropics and extratropics. Thus, following the techniques of Hewson (1998) and Berry et al. (2011), other fields are used as masks to isolate only the relevant features.
First the divergence field is used to find only those lines in convergent regions; thus, only zero contours (which are convergence lines) where the layer- and time-mean divergence is less than some threshold are retained. Figure 2c shows where the gradient of divergence is equal to zero when those contours in divergent regions are erased. It can be seen that some features that are not convergence maxima are still retained, where there is an inflexion in the horizontal gradient of divergence due to a relative minimum within a convergent region (e.g., near 0° latitude and 160°E). Such features are eliminated using the Laplacian of divergence, which is displayed by the underlying shading in Fig. 2c. The contours where the gradient of divergence is equal to zero are removed when this quantity is less than a threshold value, giving the contour field shown in Fig. 2d. This shows convergence lines and is close to what an analyst might manually draw given the fields in Fig. 2a. By definition, the ITCZ is exclusively found in the tropics, and consequently a thermodynamic mask is added to remove convergence lines within nontropical air masses, where the convergence lines could be associated with features such as midlatitude cyclones. In the example shown, there is a long convergence line near the dateline that passes near Fiji (15°S and 180° longitude) and extends into the extratropics. At the time shown, the southern end of this convergence line intersects with a slow-moving midlatitude cyclone. Therefore, at least part of this convergence line may not conform to the synoptic definition of an ITCZ. The tropics are well defined by a warm moist air mass, so the 850-hPa wet bulb potential temperature (θw) is used as a discriminator between the tropics and extratropics in this study. The 850-hPa θw for the example time is shown by the color shading in Fig. 2d, where it can be seen that the tropical regions are characterized by high values, with relatively little horizontal variation. Consequently, this field is used as a final mask to retain only lines of convergence in the tropical regions.
When these two kinematic and one thermodynamic masks are applied to the zero contour of the horizontal gradient of divergence, a line joining algorithm used by Berry et al. (2011) in the detection of fronts is employed. The masking of the fields is conducted numerically, as opposed to graphically for quantitative results and because the drawing of contours has a dependency on the nature of the plotting program. This algorithm joins points based on their proximity to their nearest neighbor; if the distance between the two points is less than a chosen threshold, the points are joined to form an ITCZ line segment. Only line segments that consist of more than a specified number of individual points and span more than 15° of longitude in total are retained as the interest here is in large-scale features. The thresholds are customizable and must be modified according to the resolution of the input data to produce synoptically sensible results. The objective, numerically defined ITCZ locations for the example time are displayed in Fig. 2e. In this study, when a qualifying ITCZ line segment is encountered, the individual points along it are examined further. The layer-mean wind components at the grid points surrounding the identified ITCZ point are recorded for the computation of composite kinematic fields.
It is recognized that in addition to the ITCZ these diagnostics will also pick out monsoon troughs and other persistent large-scale tropical convergence zones, such as the North African intertropical front (ITF; e.g., Lélé and Lamb 2010). In this study, no attempt it made to separate these phenomena and they are all referred to as the ITCZ. The rationale is they lack precise definitions suitable for an automated scheme; for example, although the ITCZ is commonly defined by confluent easterlies, whereas the monsoon trough is commonly defined by westerlies on its equatorward side, the area over which these westerlies must extend is not clear or necessarily consistent across all regions and seasons. In some instances these phenomena may blend into one another, and thus the distinction might become semantic rather than physical: broad convergence zones in the moist tropics are likely to be associated with deep convection and large-scale ascent.
The ITCZ locations are computed using the Interim European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-Interim). This global dataset spans the period 1979–2010 and is used at 1.5° horizontal resolution. The threshold values used for the background divergence and its Laplacian are set through a subjective comparison with satellite imagery and the ITCZ locations drawn on the unified surface analysis from the U.S. National Weather Service. For the 1.5° ERA-Interim data a five-point smoother was applied to the input wind field and it was determined that a maximum background divergence of −1 × 10−6 s−1 and minimum Laplacian of divergence of 2 × 10−13 s−1 gave the closest match to the manually analyzed fields. The line joining algorithm was then set to scan 3.5° from a candidate ITCZ point for nearby neighbors. To retain only the large-scale features, ITCZ segments in the ERA-Interim must comprise of five or more individual points and the total zonal extent of the segment must exceed 15° of longitude. The 850-hPa θw threshold used to define the tropics is determined by the annual mean of this field from ERA-Interim averaged between 30°N and 30°S, which is 300 K.
As mentioned earlier, it was decided to derive the ITCZ locations based upon a 144-h running average of the 1000–850-hPa layer-mean winds in order to reduce the effect of isolated synoptic events that have expected time scales of approximately 2–5 days. The sensitivity of the results to the choice of averaging period is summarized by Fig. 3, which shows the average ITCZ count for August 2000 computed using different averaging periods. It can be seen that increasing the averaging period has a qualitatively similar effect to increasing a plotting threshold: the regions of highest feature counts become more pronounced, but the overall pattern of relative high and low counts is unchanged. Given that the overall pattern is robust with varying averaging periods (consistent with the persistent nature of the ITCZ) the overall results presented here are not changed substantially if this period is altered. Although the choice is somewhat arbitrary, here a 6-day average is used on the basis that many synoptic features of the tropical troposphere (tropical cyclones, African easterly waves, etc.) have shorter periods.
Sensitivity of ITCZ count for August 2000 produced using running averages of 850–1000-hPa layer-mean winds averaged from (top) 0 to (bottom) ±10 days.
Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00339.1
The ITCZ locations are computed using fixed thresholds for the entire ERA-Interim period and the kinematic fields at the surrounding grid points are retained for further calculations. Seasonal, monthly, and annual averages are used to discuss the climatology of these features and the linear changes during these averaging periods are used to discuss trends. The association between the ITCZ and tropical cyclones is examined in this study using the International Best Track Archive for Climate Stewardship (IBTrACS) tropical cyclone dataset (Knapp et al. 2010).
3. Results
a. Annual means
The ITCZ locations and composite kinematic quantities were computed from the ERA-Interim and the annual averages for the period 1979–2010 are shown in Fig. 4. The annual-mean ITCZ count (Fig. 4, top) shows a pattern this is consistent with results from previous studies (e.g., Waliser and Gautier 1993): well-defined count maxima spanning the Northern Hemisphere Pacific and Atlantic Oceans and a maximum in the southern Indian Ocean and the south Pacific convergence zone (SPCZ). Smaller-scale regional maxima are observed in association with the world’s monsoon systems in North Africa, South America, and Australasia and near the Indian subcontinent. These features are likely best described as monsoon troughs or intertropical fronts.
Annual means derived using the objective ITCZ locations computed using ERA-Interim data for 1979–2010. (top) Count (month−1; shaded), (middle) divergence (s−1; shaded), and (bottom) relative vorticity (s−1; shaded). Black contours in the middle and bottom panels indicate where the count shown in the top panel exceeds 10 month−1.
Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00339.1
Over the oceans, the ITCZs are zonally orientated, with the exception of the SPCZ, which has its characteristic north-northwest–south-southeast tilt. The count values over the oceans are diffused over a larger geographical region in the western parts of basins, suggesting larger variance in the seasonal or day-to-day locations of the ITCZ. The ITCZ is most consistently found in the eastern Pacific, near 90°W, where there is the highest global count constrained within a latitudinally thin band. In the eastern Pacific, there is also evidence of a second ITCZ in the Southern Hemisphere, which has been observed frequently (e.g., Zhang 2001) and is a persistent feature in both general circulation models (e.g., Lin 2007) and “aquaplanet”-type model experiments (e.g., Chao and Chen 2004; Fig. 1). This Southern Hemisphere eastern Pacific ITCZ extends into the central Pacific, where it merges with the SPCZ.
The annual-mean composite divergence along the ITCZ is shown in the middle panel of Fig. 4. Note that at each grid point this field is the average of only the times when the ITCZ is identified and thus is a composite, rather than a simple average. In an annual sense, the strongest convergence occurs when ITCZs are located over the landmasses, particularly West Africa and northern Australia. Over the oceans, it is apparent that when ITCZs exist in the east Pacific and Atlantic there is significantly more convergence (approximately double) than that found when ITCZs occur over other oceanic basins. The corresponding map of composite relative vorticity along the ITCZ is shown in the bottom panel of Fig. 4, where the values in the Southern Hemisphere have been multiplied by −1 for clarity. This demonstrates that the annual-mean composite relative vorticity along the major ITCZ features is cyclonic but the cyclonic vorticity in Atlantic, SPCZ, eastern Pacific, and central Pacific ITCZ maxima is relatively low. High vorticity values are only found in these basins in the rare occasions when the ITCZs are shifted poleward from their climatological position (cf. Fig. 4. top). Compared to these regions, the mean relative vorticity within the regions of high ITCZ counts in the western Pacific and Indian Ocean ITCZ are about 3–5 times greater, suggesting that these may be somewhat different phenomena, such as monsoon troughs (cf. Molinari and Vollaro 2013).
b. Variability
The seasonal cycle of ITCZ count is shown as maps in Fig. 5 with zonal averages for some specific regions in Fig. 6. Over the continents, the ITCZ count and location varies in phase with the climatology of the local monsoon; for example, peak counts occur over North Africa in June–August (JJA) and over Australia during December–February (DJF), with very low counts during the local dry seasons, as might be anticipated. Over the oceans, the ITCZ are more persistent year-round features. In the Atlantic, the ITCZ peak has the largest seasonal shift, as it moves poleward by approximately 10° of latitude between March–May (MAM) and JJA. During the same period, ITCZ features become 21% more frequent. In the eastern Pacific, the Northern Hemisphere ITCZ behaves like that in the Atlantic (albeit with a smaller shift in latitude) and the Southern Hemisphere ITCZ is only truly evident for one season (MAM). Further analysis indicates that this box only exhibits simultaneous ITCZs in both hemispheres on 15% of the total analysis times. The western Pacific northern ITCZ maximum also migrates poleward during summer, although the shift is small (approximately 3°) and the feature count remains consistent throughout the year. The SPCZ count is more interesting as from the graph (Fig. 6b), it is evident that the Southern Hemisphere ITCZ count does not change during the year, but the count spreads across a larger range of latitudes, extending as far poleward as 30°S in DJF and/or MAM. In the Indian Ocean this seasonal perspective reveals that there is also a double ITCZ during most of the year. This basin is dominated by a Southern Hemisphere ITCZ that also migrates poleward during the local summer. The second (Northern Hemisphere) ITCZ is present just poleward of the equator during Austral summer and disappears during boreal summer when convergence zones associated with the Asian monsoon become evident at higher latitudes.
Seasonal cycle of ITCZ (left) count (month−1; shaded) and (right) divergence (s−1; scaled by 105 and shaded).
Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00339.1
Seasonal-mean ITCZ feature counts, averaged in longitude for specific boxes shown in the map.
Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00339.1
The most significant large-scale interseasonal climate mode is the El Niño–Southern Oscillation (ENSO), which is characterized by changes in the areal extent and intensity of low SSTs in the equatorial east Pacific. Given that convection over the oceans is expected to have a relationship with the underlying SST, some variation of the ITCZ locations and/or intensity may be anticipated with changes in the ENSO conditions. This is determined here by examining the difference in ITCZ count and divergence between warm and cold ENSO phases, as defined by the Southern Oscillation index (SOI) compiled by the Australian Bureau of Meteorology. Figure 7 shows these differences between an SOI index of +10 (La Niña) and −10 (El Niño), such that positive values in these plots indicate increased values during La Niña conditions. The most obvious change in the nature of the ITCZ is seen in the Pacific Ocean and is consistent with the changes in the underlying SST. Relative to El Niño conditions, during La Niña the Northern Hemisphere cross-basin ITCZ is displaced poleward by more than 5°. The opposite appears true with the Atlantic ITCZ, where in La Niña conditions the ITCZ is located farther equatorward. Around the Maritime Continent and in the eastern Indian Ocean, La Niña conditions increase the frequency of the ITCZ and the convergence associated with it, as might be expected from rainfall composites (e.g., Ropelewski and Halpert 1987). During La Niña conditions there is a dipole in both count and mean convergence along the SPCZ, centered on its approximate midpoint. This suggests that the western portion of the SPCZ becomes more active, whereas the eastern half becomes less active under La Niña conditions.
(top) Difference in annual-mean ITCZ feature count between positive SOI (>10 units) and negative SOI (<−10 units). (bottom) As in (top), but for the divergence field along objectively identified ITCZ features. The black contour in each shows where the annual-mean count is equal to 10 month−1.
Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00339.1
c. Trends
The trends for annual-mean ITCZ count and divergence are shown in the top panel of Figs. 8 and 9, respectively, as a percentage change relative to their 1979–2009 means. The lower panels in each of these figures show zonal averages of the actual annual means as a function of time across six specific boxes that have interesting trends. Only regions where the trend is significant at the 80% confidence level using a two-sided Student’s t test are plotted in Figs. 8 and 9. As should be expected, the statistically significant trends in both fields are confined to the regions of high annual count. In the Indian Ocean (region i), the ITCZ count has a negative trend over much of the basin; in particular, the maximum in the Northern Hemisphere practically disappears after the mid-1990s. In the southern Indian Ocean, the ITCZ count has a slight negative trend, but the convergence in that region has increased near the count maximum. In terms of total vertical mass flux, these factors will offset one another and may not change the total significantly. At the longitude of the Malay Peninsula (region ii) there are strong trends in both ITCZ count and divergence that indicate a strengthening of the ITCZ along the western coast of Sumatra. It appears that prior to the mid-1980s the ITCZ was essentially present only in the Southern Hemisphere and has become less frequent over the ERA-Interim period. After approximately 1988 there is an abrupt increase in the ITCZ count along the equator and in the Northern Hemisphere, which could be linked to changes in the SOI during this period.
(top) Linear trend (total change; %) of annual-mean ITCZ count for the period 1979–2010 with regions of interest denoted by rectangles labeled i–vi. Trends shown are significant at the 80% confidence level using a two-sided Student’s t test. (bottom) Time–longitude averages of the annual ITCZ counts for the regions i–vi in the top panel.
Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00339.1
As in Fig. 8, but showing the total trend (%) and annual means (10−5 s−1) for the 1000–850-hPa layer-mean divergence.
Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00339.1
A large positive trend in ITCZ count exists between Papua New Guinea and Borneo, just south of the equator. From the time–latitude plots (region iii), it is evident that the majority of this trend is associated with the appearance and intensification of the ITCZ between the equator and 5°S from approximately 1990 onward. In the western Pacific, the count in both the equatorial ITCZ and SPCZ have negative trends on their northern flank and positive trends on their southern flank, suggesting that the whole pattern has shifted to the south during the reanalysis period. The time–latitude plot of the annual-mean count in this box (region iv) supports this, showing a perceptible southward movement of the count maxima in both hemispheres throughout this period. Globally, the largest significant trends in both quantities occurs over the Amazon (region v), where both the number and strength (i.e., divergence) of ITCZ features have increased over the ERA-Interim period. The time–latitude plots indicate that these trends are caused by an abrupt shift in the time-mean location of the ITCZ. It is apparent that around 2003 the ITCZ locations move from being predominantly on or just north of the equator (and over the ocean) to being mostly south of the equator. The divergence along ITCZs along both latitudes decreases during this period but most strongly in the Southern Hemisphere. The physical reason for this change is not clear; thus, this should be investigated further in the future. However, given such an abrupt change, the possibility of changes in the observing system used in the reanalysis should not be discounted. In the eastern tropical Atlantic (region vi) the ITCZ counts have increased near the equator, whereas the composite divergence along these ITCZ has not changed. The time–latitude plot indicates that this positive trend in count arises because of a gradual southward expansion of high ITCZ counts from the Northern Hemisphere.
d. The ITCZ and tropical cyclogenesis
It is well known that the background cyclonic relative vorticity and high-tropospheric humidity of the ITCZ provides a conducive environment for the formation of tropical cyclones (e.g., McBride 1995). Molinari and Vollaro (2013) noted that many previous studies had stated that a large percentage of western North Pacific tropical cyclones (TCs) formed “within the monsoon trough” without a clear definition of the term. These authors wanted to determine more precisely how many TCs developed in the northern west Pacific monsoon trough. Some objective definitions, based on the 850-hPa relative vorticity field, were developed, and they found that 50%–100% of tropical cyclones formed within this region, depending on their precise definition and the state of ENSO. With the objective ITCZ dataset, here a similar question can be posed globally: that is, how close is TC genesis to the ITCZ?
Using the IBTrACS tropical cyclone dataset, the distance between TC genesis and objectively identified ITCZ features is considered. Because the ITCZ identification methodology uses a 6-day running average (±72 h) of the wind field, and the TCs themselves are associated with strong low-level convergence, the distance of a TC genesis location from ITCZ locations derived from wind field centered 3 days previously is considered. This time offset, along with the other tracking criteria, reduces the possibility of the synoptic-scale convergence associated with an isolated TC being wrongly aliased as an ITCZ. The computed mean distance of TC genesis locations from the closest ITCZ is displayed in Fig. 10. As should be expected, the map (Fig. 10a) shows that this distance is low where the ITCZ count (Fig. 4, top) is high, as the ITCZ counts and TC genesis locations are both peaked in the same narrow geographical areas. Farther from the equator the mean distance increases, presumably as a consequence of both the lower ITCZ count and alternative pathways for cyclogenesis (McTaggart-Cowan et al. 2013). The histogram indicates that globally 50% of TCs form within 600 km of an objectively identified ITCZ.
(a) Mean distance (km; shaded) between tropical cyclone genesis location in the IBTrACS dataset and closest objectively identified ITCZ with 1979–2010 total tropical cyclone genesis count (contoured every 5 above 5). (b) Histograms showing the distribution of distances between tropical cyclone genesis locations and nearest ITCZ using the IBTrACS dataset.
Citation: Journal of Climate 27, 5; 10.1175/JCLI-D-13-00339.1
4. Discussion and conclusions
In this study, an objective method has been developed for the purpose of identifying the ITCZ in gridded numerical datasets based upon the time- and layer-averaged horizontal wind field in the lowest part of the troposphere. On day-to-day time scales (Fig. 2), the ITCZ locations are consistent with synoptic experience and expectations and thus have been used here to examine ITCZ behavior and trends over a 30-yr period in the ERA-Interim dataset. It is found that the overall distribution of the ITCZ count is consistent with subjective or satellite analyses (Fig. 4), with elongated zones occurring over the tropical oceans and monsoon regions. Most of these zones move poleward during summer and equatorward in winter. These results are certainly not new, but what is new is that an automated objective methodology has been employed to detect the location of these features and give more detailed information about them. An important result is that a qualitative and repeatable analysis of the impact of ENSO and the recent (30 yr) trends in ITCZ properties has been given. The work presented here provides one of the first automated, objective reference studies of ITCZ features, which can be repeated and quantitatively compared with other gridded datasets, such as other reanalysis products or climate models.
The most persistent ITCZ is found over the eastern Pacific Ocean, where the ITCZ is confined to a narrow latitudinal band year-round, which likely reflects the variability of the underlying sea surface temperatures and the flow around the adjacent subtropical anticylones. The Northern Hemisphere east Pacific ITCZ also possesses the strongest convergence of all the oceanic ITCZ regions during all seasons. The count statistics also pick out a second, weaker and less frequent equatorial ITCZ that extends across the entire Pacific during MAM (see Fig. 5, left), which intersects the SPCZ near the dateline. There are also two ITCZ maxima in the Indian Ocean present throughout the year, although they are most pronounced during the same season, which is reminiscent of double ITCZs that appear in general circulation models when forced with observed sea surface temperatures (e.g., Lin 2007).
The analysis here picks out the SPCZ rather differently from the other tropical convergence zones. Not only does it have a characteristic tilt that means it extends across a range of latitudes, it is the only major ITCZ-type feature that does not shift meridionally during the course of the year. The SPCZ core remains at the same latitude and activity spreads farther poleward during Austral summer. This may highlight some different dynamics associated with the SPCZ, such as the interaction with subtropical fronts (Berry et al. 2011).
The statistically significant annual trends of ITCZ properties are confined to regions of climatologically high ITCZ counts. Over most regions, the convergence within the oceanic ITCZ has increased over the last three decades consistent with the strengthening of the Hadley circulation noted in previous studies (e.g., Stachnik and Schumacher 2011; Mitas and Clement 2005). In the western Pacific, there is a pattern in both ITCZ count and divergence that suggests that the equatorial ITCZ and SPCZ have shifted southward, which is supported by the time–latitude plots in Figs. 8 and 9. Using objective fronts to identify thermal discontinuities across the SPCZ, a similar change in the SPCZ was noted by Berry et al. (2011). The physical mechanism responsible for this shift is unclear: the ITCZ location is likely determined by a combination of the underlying surface (e.g., locations of high and low SSTs) and the large-scale flow in both the tropics and extratropics. Over land, one particularly interesting trend is found in northern South America; here there is a strong positive trend in the ITCZ count and the mean convergence, whereas just offshore there is a negative trend in both, associated with an abrupt shift in the ITCZ around 2003. The physical mechanism driving this is not clear, but it may be associated with observed changes in the sea surface temperatures along the South American coast (see, e.g., Evans and Braun 2012; Lumpkin and Garzoli 2011).
By comparing the objective ITCZ locations with data from the IBTrACS dataset, it was found that more than 50% of TCs formed within 600 km of the ITCZ. From a synoptic perspective, this result is unsurprising as the ITCZ can provide the background cyclonic vorticity required for cyclogenesis and it is found that TC genesis and ITCZ are generally peaked in the same geographical location, consistent with recent work by Molinari and Vollaro (2013). The natural extension of this type of analysis is in the analysis of output from climate models that are either too coarse to resolve TCs or only store data averaged over a period. If these simulations suggest changes in the location or intensity of the ITCZ, this may be used as a proxy to infer general changes in TC genesis and help to test hypotheses of changing TC activity.
The diagnostic methods shown here are simple to apply in any gridded dataset with the necessary lower-tropospheric input fields (i.e., zonal and meridional wind components). Because the ITCZ is an important fixture of the tropical climate, it is recommended that future studies apply these calculations to output from climate models. Doing so will allow the skill of climate models in the tropics in the present day to be objectively determined. Consequently, different models or different versions of the same model can be easily and quantitatively compared. The diagnostics will also permit an automated and consistent assessment of the future state of the ITCZ and forecasts of how conditions could change.
Acknowledgments
The authors wish to thank three anonymous reviewers for their helpful comments. This work is funded by Australian Research Council Grant FS100100081.
REFERENCES
Berry, G., C. Jakob, and M. Reeder, 2011: Recent global trends in atmospheric fronts. Geophys. Res. Lett., 38, L21812, doi:10.1029/2011GL049481.
Brown, J. R., S. B. Power, F. P. Delage, R. A. Colman, A. F. Moise, and B. F. Murphy, 2011: Evaluation of the South Pacific convergence zone in IPCC AR4 climate model simulations of the twentieth century. J. Climate, 24, 1565–1582.
Chan, S. C., and J. L. Evans, 2002: Comparison of the structure of the ITCZ in the west Pacific during the boreal summers of 1989–93 using AMIP simulations and ECMWF reanalysis. J. Climate, 15, 3549–3568.
Chao, W. C., and B. Chen, 2004: Single and double ITCZ in an aqua-planet model with constant sea surface temperature and solar angle. Climate Dyn., 22, 447–459.
Evans, J. L., and A. Braun, 2012: A climatology of subtropical cyclones in the South Atlantic. J. Climate,25, 7328–7340.
Gu, G., R. F. Adler, and A. H. Sobel, 2005: The eastern Pacific ITCZ during the boreal spring. J. Atmos. Sci., 62, 1157–1174.
Hewson, T. D., 1998: Objective fronts. Meteor. Appl., 5, 37–65.
Joyce, R. J., J. E. Janowiak, P. A. Arkin, and P. Xie, 2004: CMORPH: A method that produces global precipitation estimates from passive microwave and infrared data at high spatial and temporal resolution. J. Hydrometeor., 5, 487–503.
Knapp, K. R., M. C. Kruk, D. H. Levinson, H. J. Diamond, and C. J. Neumann, 2010: The International Best Track Archive for Climate Stewardship (IBTrACS). Bull. Amer. Meteor. Soc., 91, 363–376.
Lélé, I., and P. J. Lamb, 2010: Variability of the intertropical front (ITF) and rainfall over the West African Sudan–Sahel zone. J. Climate,23, 3984–4004.
Lin, J.-L., 2007: The double-ITCZ problem in IPCC AR4 coupled GCMs: Ocean–atmosphere feedback analysis. J. Climate, 20, 4497–4525.
Lumpkin, R., and S. Garzoli, 2011: Interannual to decadal changes in the western South Atlantic’s surface circulation. J. Geophys. Res.,116, C01014, doi:10.1029/2010JC006285.
McBride, J. L., 1995: Tropical cyclone formation. Global Perspectives on Tropical Cyclones, R. L. Elsberry, Ed., World Meteorological Organization, 63–105.
McTaggart-Cowan, R., T. J. Galarneau Jr., L. F. Bosart, R. W. Moore, and O. Martius, 2013: A global climatology of baroclinically influenced tropical cyclogenesis. Mon. Wea. Rev., 141, 1963–1989.
Mitas, C. M., and A. Clement, 2005: Has the Hadley cell been strengthening in recent decades? Geophys. Res. Lett., 32, L03809, doi:10.1029/2004GL021765.
Molinari, J., and D. J. Vollaro, 2013: What percentage of western North Pacific tropical cyclones form within the monsoon trough? Mon. Wea. Rev.,141, 499–505.
Rácz, Z., and R. K. Smith, 1999: The dynamics of heat lows. Quart. J. Roy. Meteor. Soc., 125, 225–252.
Ropelewski, C. F., and M. S. Halpert, 1987: Global and regional scale precipitation patterns associated with the El Niño/Southern Oscillation. Mon. Wea. Rev., 115, 1606–1626.
Sadler, J. C., 1975: The upper tropospheric circulation over the global tropics. University of Hawaii Department of Meteorology Rep. UHMET-75-05, 35 pp.
Stachnik, J. P., and C. Schumacher, 2011: A comparison of the Hadley circulation in modern reanalyses. J. Geophys. Res., 116, D22102, doi:10.1029/2011JD016677.
Waliser, D. E., and C. Gautier, 1993: A satellite-derived climatology of the ITCZ. J. Climate, 6, 2162–2174.
Wang, C.-C., and G. Magnusdottir, 2006: The ITCZ in the central and eastern Pacific on synoptic time scales. Mon. Wea. Rev., 134, 1405–1421.
Zhang, C., 2001: Double ITCZs. J. Geophys. Res., 106 (D11), 11 785–11 792.
