Abstract

The tropical cyclone rainfall climatological study performed for the North Pacific was extended to the North Atlantic. Similar to the North Pacific tropical cyclone study, mean monthly rainfall within 444 km of the center of the North Atlantic tropical cyclones (i.e., that reached storm stage and greater) was estimated from passive microwave satellite observations during an 11-yr period. These satellite-observed rainfall estimates were used to assess the impact of tropical cyclone rainfall in altering the geographical, seasonal, and interannual distribution of the North Atlantic total rainfall during June–November when tropical cyclones were most abundant. The main results from this study indicate 1) that tropical cyclones contribute, respectively, 4%, 3%, and 4% to the western, eastern, and entire North Atlantic; 2) similar to that observed in the North Pacific, the maximum in North Atlantic tropical cyclone rainfall is approximately 5°–10° poleward (depending on longitude) of the maximum nontropical cyclone rainfall; 3) tropical cyclones contribute regionally a maximum of 30% of the total rainfall northeast of Puerto Rico, within a region near 15°N, 55°W, and off the west coast of Africa; 4) there is no lag between the months with maximum tropical cyclone rainfall and nontropical cyclone rainfall in the western North Atlantic, whereas in the eastern North Atlantic, maximum tropical cyclone rainfall precedes maximum nontropical cyclone rainfall; 5) like the North Pacific, North Atlantic tropical cyclones of hurricane intensity generate the greatest amount of rainfall in the higher latitudes; and 6) warm El Niño–Southern Oscillation events inhibit tropical cyclone rainfall.

Introduction

One of the main driving forces for the motion of the earth's atmosphere is provided by tropical heat sources generated by the combination of clouds and precipitation. Tropical cyclones are an important source of rainfall for agriculture and other water applications over the regions of the subtropics and Tropics. However, questions remain concerning the amount that tropical cyclones contribute to the total (i.e., combined tropical and nontropical cyclone) rainfall, how these tropical cyclone rainfall patterns are distributed geographically and seasonally, and how climate variations such as ENSO events affect tropical cyclone rainfall.

Using passive microwave satellite observations, these questions have been addressed by Rodgers et al. (2000) for the North Pacific for 1987–98 during the tropical cyclone season (i.e., June–November) with the following results. Tropical cyclones contribute approximately 7% of the rainfall in the entire domain of the North Pacific during a tropical cyclone season, with maximum regional contributions of approximately 30% northeast of the Philippines and 40% off the lower Baja California coast. It has also been shown that tropical cyclone rainfall in the western North Pacific is affected significantly by ENSO variations in the opposite sense of what might be expected. For example, during the El Niño events that occurred during these years, high sea surface temperatures (SSTs) and evaporation are produced that create more total precipitation than normal in the eastern and central North Pacific. Moreover, the El Niño years produce lower-than-normal SSTs and nontropical cyclone precipitation in the western North Pacific. However, tropical cyclone rainfall amounts in the western North Pacific are greater than normal despite the lower SSTs because of the El Niño–induced lower-tropospheric circulation patterns that are favorable for tropical cyclone genesis and intensification.

The next question that should be addressed is the manner in which tropical cyclones influence the distribution of total North Atlantic rainfall. In this study, monthly North Atlantic tropical cyclone rainfall and the rainfall generated by all other North Atlantic systems are determined using data from the Special Sensor Microwave Imager (SSM/I) instrument on the Defense Meteorological Satellite Program (DMSP) satellites. The SSM/I observations are collected over the North Atlantic domain (0°–35°N by 0°–100°W), during the months of June–November when tropical cyclones are most abundant, for 1987–89 and 1991–98 (there were few SSM/I data available during 1990). These monthly rainfall observations are then used to examine geographical, seasonal, and interannual variations in North Atlantic tropical cyclone rainfall.

Data analyses

Special Sensor Microwave Imager radiometer

The DMSP satellites circle the globe 14.1 times per day along a near sun-synchronous orbit at an altitude of 833 km and a 98.8° inclination. The SSM/I sensors on DMSP F-8, F-10, F-11, F-13, and F-14 satellites measure reflected and emitted dual-polarized microwave radiation at frequencies of 19.4, 37.0, and 85.5 GHz and vertically polarized microwave radiation at 22.2 GHz. The SSM/I instruments scan conically at a 45° angle from nadir and have an observational swath width of approximately 1400 km at the earth's surface. The instantaneous footprint of the observation varies with channel, with the finest resolution being 15 km at 85 GHz. The F-8 SSM/I was fully operational from the first week of July in 1987 to February of 1990, and the F-10 SSM/I was partially operational from December of 1990 to January of 1992 and was fully operational thereafter. The F-11 SSM/I had been fully operational since January of 1992, and the F-13 and F-14 SSM/Is were, respectively, launched in May of 1995 and May of 1996 and have been fully operational ever since. The ascending (descending) F-8, F-10, F-11, F-13, and F-14 orbits, respectively, cross the equator near the eastern North Pacific at approximately 0020 (1150), 0405 (1630), 0134 (1308), 2335 (1208), and 0147 (1429) UTC. Further information concerning the SSM/I sensor measurements and orbital mechanics can be found in Hollinger (1991).

SSM/I brightness temperature–rain-rate algorithm

Rain rates are obtained from the SSM/I-derived brightness temperatures with an algorithm developed by Adler et al. (1994). The algorithm is called the Version-2 Goddard Scattering Algorithm (GSCAT-2). The algorithm uses a combination of the channels at 37.0, 22.2, and 19.9 GHz to define potential raining regions over the ocean and the 85.5-GHz channel to derive rain rates greater than 1 mm h−1 within the raining areas. Both the rain criteria and the rain-rate–brightness temperature relationship are based on cloud-model calculations (Adler et al. 1991). The algorithm is chosen for its computational simplicity and its enhanced instantaneous field of view in observing rain rates from the 85.5-GHz channel (i.e., 15 km × 15 km). Real mean rain rates derived from SSM/I for the inner core (i.e., within 111 km of the tropical cyclone center) of the 1987–89 western North Atlantic tropical cyclones using the GSCAT-2 algorithm are found [see Table 2 in Rodgers et al. (1994)] to compare favorably with most of the early estimates that are obtained from rain gauges, water vapor budget studies, and earlier satellite-based microwave observations.

Sampling technique

The analysis area for this study covers both land and water regions of the North Atlantic, divided into two regions. The western North Atlantic region is located in 0°–35°N and 50°–100°W, and the eastern North Atlantic area of analyses is located in 0°–35°N and 0°–50°W (see Fig. 1). In the western region, the eastern North Pacific tropical cyclone rainfall is eliminated from the sample by deleting the tropical cyclones in that region, and nontropical cyclone rainfall is kept. The two North Atlantic domains are arbitrarily chosen to partition the ocean basins into equal geographical areas. The domain was also chosen to eliminate regions of the North Atlantic in which tropical cyclones are rarely observed [near the equator because of limited Coriolis forcing (i.e., latitudes less than 5°N) or at high latitudes because of strong vertical wind shear that is greater than 10 m s−1 and/or SSTs that are less than 26°C (i.e., latitudes greater than 35°N)].

Fig. 1.

Plan view of the North Atlantic showing the designated western and eastern regions (within boxes) that were sampled for total, nontropical cyclone, and tropical cyclone mean monthly rainfall

Fig. 1.

Plan view of the North Atlantic showing the designated western and eastern regions (within boxes) that were sampled for total, nontropical cyclone, and tropical cyclone mean monthly rainfall

To assemble the mean monthly tropical cyclone rainfall data, SSM/I-derived tropical cyclone rain rates are used. Only the rain rates that are observed within 444-km radius of the center of circulation of North Atlantic tropical cyclones by SSM/I are sampled. This tropical-cyclone rain-rate sampling area is chosen to encompass the majority of the rainfall that is contributed by tropical cyclones, which includes the eyewall and the inner and outer rainbands. The center of the sampled tropical cyclones for the time of the SSM/I passes is interpolated from the best-track data. The spatial resolution of the SSM/I (15 km) and the algorithm used in this study (Adler et al. 1994) will not reproduce the largest, fine-scale rainfall rates found in tropical cyclones but will produce reasonable area-averaged (satellite footprint size) rainfall rates and rainfall coverages.

Within the 444-km circular domain, the SSM/I sensors were able to monitor the rain rates at least once for 132 western and 111 eastern North Atlantic tropical cyclones. The sample does not include tropical cyclones that never reached storm stage. From these observations, the SSM/I tropical cyclone rain rate samples include 38 depressions, 57 storms (18–32 m s−1), and 37 hurricanes (33 m s−1 and higher) in the western North Atlantic and 33 depressions, 51 storms, and 27 hurricanes in the eastern North Atlantic. Moreover, for tropical cyclones that occur over land, the SSM/I-derived rain rates are only sampled for dissipating tropical cyclones that are followed by the best-track reports (Jarvinen et al. 1984). Therefore, for these reasons, the study will slightly underestimate the total tropical cyclone rainfall. The geographic distribution of the satellite observations used in this study is shown in Fig. 2 in terms of the total number of satellite footprints identified as being associated with tropical cyclones per unit area (0.5° latitude by 0.5° longitude). The pattern of tropical cyclone frequency clearly shows the main concentrations of storms and the main storm tracks across the Atlantic from Africa, the recurvature at the longitude of Puerto Rico, and the development and recurvature in the Carribbean and Gulf of Mexico.

Fig. 2.

Geographic distribution of SSM/I satellite footprints associated with tropical cyclone rainfall used in this study. The units are total satellite pixels per unit area, with the area being 0.5° lat by 0.5° long

Fig. 2.

Geographic distribution of SSM/I satellite footprints associated with tropical cyclone rainfall used in this study. The units are total satellite pixels per unit area, with the area being 0.5° lat by 0.5° long

To estimate the monthly tropical cyclone rainfall, the SSM/I-derived rain rates that are sampled within 444 km of the center of each tropical cyclone are accumulated and averaged for a 2.5° latitudinal–longitudinal grid. That number is converted to mean tropical cyclone rain rate by dividing by the total sample of SSM/I data points over the grid in the month. This approach includes the overpasses that have zero cyclone rain (i.e., either rain that is not associated with a tropical cyclone or no rain at all). This mean rain rate is converted to monthly rainfall by multiplying by the number of hours in the month. The procedure is identical to that used in Rodgers et al. (2000) and is defined in Eq. 1 (these values will be referred to as “tropical cyclone rainfall” in the remaining portion of the text):

 
formula

where RAcyclone = rainfall amount per month (mm month−1) contributed by tropical cyclones within a 2.5° latitudinal–longitudinal grid, RRcyclone = SSM/I observed rain rate (mm h−1) contributed by tropical cyclones during a given month within a 2.5° latitudinal–longitudinal grid, PixelsSSM/I = number of SSM/I pixels, and Hoursmonth = number of hours in a given month.

The total rainfall from all systems is then calculated considering all SSM/I observations and will be referred to as “total rainfall” in the remaining portion of the text. Last, the nontropical cyclone rainfall is calculated from the differences between the total rainfall and that estimated from tropical cyclones and will be referred to as “nontropical cyclone rainfall” in the remaining portion of the text. The total rainfall estimates on monthly timescales presented using this particular SSM/I-based technique compare favorably with other satellite-based analyses, including the community Global Precipitation Climatology Project results (Huffman et al. 1997).

To make the SSM/I observations more homogeneous from year to year and from month to month, the following adjustments in the SSM/I-derived mean rainfall datasets are made. First, the accumulated monthly rainfall amounts for the interannual rainfall analyses are generated from a single SSM/I satellite from 1992 to 1998, when multiple SSM/Is were flown. This procedure is done to avoid some part of the interannual differences being due to the difference in sampling (one satellite or two), which will include a difference in the time of day of the observations. For the other rainfall analyses for which this factor is not important, all available SSM/I data are used. Second, given that the F-8 SSM/I was not operational until after June of 1987 and during the tropical cyclone season of 1990, the mean interannual rainfall set was limited to months of July–November. In addition, the mean seasonal rainfall analyses for the months of June–November only contained 1988–89 and 1991–98 (e.g., total of 60 months).

Sea surface temperatures

To examine the variation of SSTs during the El Niño and La Niña years during this 10-yr period, global mean monthly SSTs for a given year were used. The month of September was chosen because it is the month with the maximum tropical cyclone occurrences. These SSTs were provided by the National Centers for Environmental Prediction and averaged for a 2.5° latitudinal–longitudinal horizontal grid (Reynolds and Smith 1994).

Results

Geographical distribution of nontropical cyclone and tropical cyclone rainfall

The upper, middle, and lower panels of Fig. 3 show, respectively, the geographical distribution of mean monthly nontropical cyclone rainfall, the mean tropical cyclone rainfall, and the percentage of rainfall contributed by tropical cyclones (i.e., the ratio of tropical cyclone rainfall to total rainfall) over the North Atlantic during the 65-month period. The upper panel of Fig. 3 indicates that the region with the greatest nontropical cyclone rainfall amount (i.e., greater than 300 mm month−1) is associated with the ascending branch of the Hadley circulation that helps to maintain the ITCZ regions and the active baroclinic zone off the east coast of United States. A rainfall maximum off the east coast of the United States is associated with the mean position of the baroclinic zone. Regions of light rainfall amounts (less than 100 mm month−1) likely are associated with the descending branch of the Hadley circulation over vast regions of the subtropics, northern Africa, southern Spain, and the east and southeastern United States. It is noted that the dry region over the eastern region of the North Atlantic extends southwestward toward the region between Puerto Rico and the coasts of Venezuela and Columbia. The nontropical cyclone mean monthly rainfall amounts for the 65-month period are found to be, respectively, 209, 123, and 166 mm month−1 for the western, eastern, and the entire North Atlantic. The North Atlantic mean monthly rainfall amounts are somewhat less than that found in the North Pacific for the same time period.

Fig. 3.

A plan view showing the SSM/I-derived mean monthly North Atlantic rainfall amounts (mm month−1) contributed by nontropical cyclone systems and (middle) tropical cyclones and the (bottom) fractional amount of tropical cyclone rainfall (percentage) for Jul–Nov 1987, Jun–Nov 1988–89, and 1991–98 (65-month period). Brown and blue backgrounds designate, respectively, nonraining land and ocean areas. The color bar code delineates the percentage or rainfall amounts for each figure

Fig. 3.

A plan view showing the SSM/I-derived mean monthly North Atlantic rainfall amounts (mm month−1) contributed by nontropical cyclone systems and (middle) tropical cyclones and the (bottom) fractional amount of tropical cyclone rainfall (percentage) for Jul–Nov 1987, Jun–Nov 1988–89, and 1991–98 (65-month period). Brown and blue backgrounds designate, respectively, nonraining land and ocean areas. The color bar code delineates the percentage or rainfall amounts for each figure

The middle panel of Fig. 3 suggests that the maximum tropical cyclone rainfall is concentrated in the subtropical latitudes from the middle North Atlantic west toward the Gulf of Mexico. No tropical cyclone rainfall is found off the west coast of Spain and Africa and equatorward of 5°N latitude. The regional area with the greatest tropical cyclone rainfall (i.e., greater than 30 mm month−1) occurs east of Puerto Rico and north of the Lesser Antilles. This is the region in which many tropical cyclones during the period recurved and momentarily intensified. The mean monthly rainfall contributed by tropical cyclones during the period is respectively 9, 3, and 6 mm month−1 for the western, eastern, and entire North Atlantic domain. In general, the tropical cyclone rainfall map agrees very well with the mean climate data for tropical cyclone occurrence (Neumann et al. 1999).

By comparing the middle and upper panels of Fig. 3, it can be seen that, like the North Pacific (Rodgers et al. 2000), the maximum North Atlantic tropical cyclone rainfall is slightly poleward of the maximum nontropical cyclone rainfall found in the ITCZ. This result is especially obvious in the percentage fields seen in the lower panel of Fig. 3. It can be seen that the regions in which tropical cyclones contribute the greatest rainfall are in the dry regions of the subtropics north and west of the ITCZ. For example, the area northeast of Puerto Rico, for which the nontropical cyclone rainfall for this period is less than 150 mm month−1, receives more than 30% additional rainfall from tropical cyclones. Thus, it is likely that without the presence of the tropical cyclones, which Puerto Rico endures, annual rainfall there would be significantly less. Figure 3 also indicates that tropical cyclones contributed as much as 10% to the total rainfall over land areas of the southeastern United States, Yucatan Peninsula, and Central America. However, it should be reemphasized that tropical cyclone rainfall over land and along coastal regions may be underestimated in this study for the reasons given in section 2. Also, shallow orographic rainfall that is confined primarily below the freezing level may be missed or underestimated by the GSCAT technique, which is dependent on scattering by ice to estimate precipitation. The mean percentages of rainfall contributed by tropical cyclones during this period for the western, eastern, and entire North Atlantic are, respectively, 4%, 3%, and 4%.

Zonally averaged nontropical cyclone and tropical cyclone profiles for both the western and eastern North Atlantic are seen in Fig. 4. In the western North Atlantic (Fig. 4a), there is a broad zonal region of high tropical cyclone rainfall amounts (>10 mm month−1) between 15° and 32°N, with a maximum of 15 mm month−1 near 27°N. Tropical cyclones at 27°N contribute nearly 10% to the total rainfall, and the maximum is approximately 20° north of the maximum nontropical cyclone rainfall (Fig. 4b) found in the ITCZ. In the eastern North Atlantic (Fig. 4c), the zone of maximum tropical cyclone rainfall is nearly one-half of that found in the western North Atlantic (i.e., 7 mm month−1), occupies a smaller zonal region, contributes less than 4% to the total rainfall, and is located near 11°N. This axis of maximum rainfall is only 5° poleward of the zone of maximum nontropical cyclone rainfall (Fig. 4d). This North Atlantic zonal analysis indicates that during the tropical cyclone months the western North Atlantic tropical cyclones are more numerous and intense and move to higher latitudes as they begin to interact with the westerlies.

Fig. 4.

The SSM/I-derived 2.5°-lat nontropical cyclone and tropical cyclone zonally averaged mean monthly rainfall amounts (mm month−1) for, respectively, the (a), (b) western and (c), (d) eastern North Atlantic for the 65-month period

Fig. 4.

The SSM/I-derived 2.5°-lat nontropical cyclone and tropical cyclone zonally averaged mean monthly rainfall amounts (mm month−1) for, respectively, the (a), (b) western and (c), (d) eastern North Atlantic for the 65-month period

Seasonal variation of tropical cyclone rainfall

The seasonal variation of the North Atlantic tropical cyclone mean rainfall for the months of June–November is seen in Fig. 5. The figure shows that eastern North Atlantic tropical cyclone rainfall increases from June to September and then decreases after September. The eastern North Atlantic tropical cyclones that contribute rainfall during August and September usually develop from African easterly wave disturbances. They can be the most intense tropical cyclones of the season as they propagate westward towards North and Central America or recurve northwestward into the westerlies.

Fig. 5.

A plan view showing the SSM/I-derived mean monthly North Atlantic tropical cyclone rainfall amounts (mm month−1) contributed by tropical cyclones for 1988, 1989, and 1991–98 for Jun–Nov. The brown and blue backgrounds designate nonraining land and ocean areas, respectively. The color bar code delineates the rainfall amounts

Fig. 5.

A plan view showing the SSM/I-derived mean monthly North Atlantic tropical cyclone rainfall amounts (mm month−1) contributed by tropical cyclones for 1988, 1989, and 1991–98 for Jun–Nov. The brown and blue backgrounds designate nonraining land and ocean areas, respectively. The color bar code delineates the rainfall amounts

In the western North Atlantic, tropical cyclone rainfall can be found throughout the season, with the greatest amounts located in the subtropical regions east-northeast of Puerto Rico during July–September. During June, October, and November, the regions of heavy tropical cyclone rainfall can be found in the Gulf of Mexico or the western Caribbean Sea. These tropical cyclones, however, do not always originate from the North African easterly wave disturbances but are sometimes generated in the cyclogenesis regions of the Gulf of Mexico and Caribbean Sea.

Figure 6 shows a histogram of the western (Fig. 6a) and eastern (Fig. 6b) North Atlantic nontropical cyclone (black bars) and tropical cyclone (dashed bars) monthly mean rainfall amounts. The mean monthly rainfall and the percentage (the percentage is the number above the bar graph) of rainfall contributed by tropical cyclones peaks in August–October in the western North Atlantic; the tropical cyclone rainfall peaks in September in the eastern North Atlantic. In the western North Atlantic, the months that contain the largest tropical cyclone rainfall amounts (i.e., August–October) coincide with the months that have the greatest nontropical cyclone rainfall. These tropical cyclones are usually generated by the North African easterly wave disturbances but can also be generated by other atmospheric conditions. However, in the eastern North Atlantic, the month with the greatest tropical cyclone rainfall amounts (i.e., September) precedes the month with the greatest nontropical cyclone rainfall. These results suggest that the eastern North Atlantic tropical cyclones generated by the North African easterly wave disturbance were more active and wetter in September, whereas the combination of the ITCZ and baroclinic systems created greater nontropical cyclone rainfall during October.

Fig. 6.

A histogram showing the SSM/I-derived averaged mean monthly rainfall amounts (mm month−1) contributed by nontropical cyclone systems (dark shade) and tropical cyclones (hatched shade) for 1988, 1989, and 1991–98 during Jun–Nov for the (a) western and (b) eastern North Atlantic. Percentage above the bar graph represents the percentage of rainfall contributed by tropical cyclones during the month

Fig. 6.

A histogram showing the SSM/I-derived averaged mean monthly rainfall amounts (mm month−1) contributed by nontropical cyclone systems (dark shade) and tropical cyclones (hatched shade) for 1988, 1989, and 1991–98 during Jun–Nov for the (a) western and (b) eastern North Atlantic. Percentage above the bar graph represents the percentage of rainfall contributed by tropical cyclones during the month

Figure 7a, which shows the zonally averaged mean monthly tropical cyclone rainfall in the western North Atlantic during the early and late summer and autumn months, reemphasizes the seasonal change. It is clearly seen that larger tropical cyclone rainfall (>20 mm month−1) between 16° and 32°N occurs during the late summer. In the eastern North Atlantic (Fig. 7b), there is a late summer tropical cyclone rainfall season. During this time, larger mean tropical cyclone rainfall (>15 mm month−1) occurs between 10° and 17°N latitude. Cyclone rainfall of less than 7 mm month−1 occurs at all latitudes during the other months. In both the western and eastern North Atlantic regions, the tropical cyclones that create the greatest zonally averaged mean rainfall during the late summer are those that are developed from the North African easterly wave disturbances (Pasch et al. 1998). Note also from the figure that there is no latitudinal shift during the season, unlike what was observed in the western North Pacific (Rodgers et al. 2000).

Fig. 7.

The SSM/I-derived 2.5°-lat zonally averaged mean monthly rainfall amounts (mm month−1) contributed by tropical cyclones for 1988, 1989, and 1991–98 during the early summer (Jun–Jul), late summer (Aug–Sep), and autumn (Oct–Nov) seasons for the (a) western and (b) eastern North Atlantic

Fig. 7.

The SSM/I-derived 2.5°-lat zonally averaged mean monthly rainfall amounts (mm month−1) contributed by tropical cyclones for 1988, 1989, and 1991–98 during the early summer (Jun–Jul), late summer (Aug–Sep), and autumn (Oct–Nov) seasons for the (a) western and (b) eastern North Atlantic

Rainfall and tropical cyclone intensity

The geographical distribution of tropical cyclone rainfall contributed by depressions, storms, and hurricanes is seen in Fig. 8. Rainfall from depressions is generally less than 10 mm month−1 and is concentrated in areas in the central and eastern North Atlantic and in the western Caribbean and Gulf of Mexico. Rainfall from tropical storms (middle panel) is more widespread and extends significantly into the subtropics. The hurricane rainfall is concentrated in the Western Atlantic, and the peak northeast of Puerto Rico in Fig. 2 is due primarily to hurricanes (bottom panel in Fig. 8). The preferred path of maximum hurricane rainfall traces eastward north of the Greater Antilles and then splits, with one path recurving northeastward into the westerlies and another moving toward the Florida coast. The “bulls-eye” hurricane rain maximum off the coast of Honduras is associated with Hurricane Mitch. It is also seen that in the western North Atlantic (30°N, 60°W by 40°N, 50°W) the greater amounts of rainfall are generated at the higher latitudes by tropical cyclones of hurricane intensity.

Fig. 8.

A plan view showing the SSM/I-derived mean monthly North Atlantic tropical cyclone rainfall amounts (mm month−1) contributed by tropical cyclones at depression, storm, and hurricane stages during Jun–Nov of 1988, 1989, and 1991–98. The brown and blue backgrounds designate nonraining land and ocean areas, respectively. The color bar delineates the rainfall amounts

Fig. 8.

A plan view showing the SSM/I-derived mean monthly North Atlantic tropical cyclone rainfall amounts (mm month−1) contributed by tropical cyclones at depression, storm, and hurricane stages during Jun–Nov of 1988, 1989, and 1991–98. The brown and blue backgrounds designate nonraining land and ocean areas, respectively. The color bar delineates the rainfall amounts

The statistical information seen in Fig. 9 indicates that tropical storms actually have the greatest contribution to tropical cyclone rainfall, especially in the eastern Atlantic. This result is partially due to the larger number of storm-stage events (actually satellite overpasses in this study) in the eastern Atlantic but also to an apparent slightly higher mean rainfall rate for storm stage than for hurricanes. The depression stage contributes significantly less than storms and hurricanes in the western Atlantic but contributes as significantly as hurricanes in the eastern portion of the ocean.

Fig. 9.

A histogram showing the SSM/I-derived mean monthly rainfall amounts (mm month−1), percentage of contribution, and number of observations for the western (dark shade) and eastern (light hatched) North Atlantic tropical cyclones at depression, storm, and hurricane stages during the 65-month period

Fig. 9.

A histogram showing the SSM/I-derived mean monthly rainfall amounts (mm month−1), percentage of contribution, and number of observations for the western (dark shade) and eastern (light hatched) North Atlantic tropical cyclones at depression, storm, and hurricane stages during the 65-month period

The zonally averaged tropical cyclone rainfall as a function of tropical cyclone intensity for the western (Fig. 10a) and the eastern (Fig. 10b) Atlantic suggests that as tropical cyclones become more intense there is a narrowing of the zonal peak and a poleward shift. A similar shift has been noted in the Pacific Ocean (Rodgers et al. 2000). In the western North Atlantic, at 10°–15°N, hurricanes are contributing only as much as depressions, but at 20°–25°N, hurricanes are contributing more than either depressions or storms. In the eastern Atlantic, hurricane rainfall peaks at a higher latitude than for either depressions or storms.

Fig. 10.

The SSM/I-derived 2.5°-lat zonally averaged mean monthly rainfall amounts (mm month−1) contributed by the (a) western and (b) eastern North Atlantic tropical cyclones at depression, storm, and hurricane stages during the 65-month period

Fig. 10.

The SSM/I-derived 2.5°-lat zonally averaged mean monthly rainfall amounts (mm month−1) contributed by the (a) western and (b) eastern North Atlantic tropical cyclones at depression, storm, and hurricane stages during the 65-month period

Interannual variation of tropical cyclone rainfall

Within the North Atlantic basin at timescales sampled in this study, the atmospheric forcing mechanisms that have the greatest control on the interannual variation of tropical cyclone frequency and intensity are those related to ENSO events (Gray 1984a; Goldenberg and Shapiro 1996). According to Gray (1984a), the enhanced central and eastern North Pacific tropical SSTs and ocean evaporation during the El Niño years help to shift the convective region of the ITCZ more eastward, thereby exposing the North Atlantic to stronger westerly outflow and vertical wind shear. This increase of vertical wind shear over the North Atlantic, in turn, causes an enhancement in upper-tropospheric ventilation and less convective development (Reuter and Yau 1986; DeMaria 1996). During La Niña and neutral years, on the other hand, there is little eastward shift in the convective region of the ITCZ because of the cooler eastern and central North Pacific SSTs and less vertical wind shear over the North Atlantic. These results were substantiated in the studies that examined the long-term history of strike probability and damage caused by hurricane landfall in the western North Atlantic (Bove et al. 1998; Pielke and Landsea 1999). These studies suggested that there is a greater hurricane strike probability and damage from hurricanes occurring during La Niña and neutral years than during El Niño years.

However, as mentioned earlier, the interannual variation of hurricanes is only partially explained by ENSO events. There are other factors that suppress or enhance the interannual variability of the frequency and intensity of North Atlantic tropical cyclones. Other interannually varying factors such as the variation of SSTs and surface pressures that these tropical systems encounter and the phase of the quasi-biennial oscillation (Gray 1984a,b) can influence the frequency and intensity of tropical cyclones. Short-term climate changes such as the variation of rainfall in the western Sahel region of Northern Africa (Landsea and Gray 1992) as well as longer-term climate changes caused by the linkage between the ocean and atmosphere that is associated with the North Pacific oscillation (Gershanov and Barnett 1998) and the North Atlantic oscillation (Hurrell 1995; Gray et al. 1997) may have a longer-term effect on the frequency and intensity of tropical cyclones. Therefore, it is likely that ENSO events alone do not necessarily produce an anomaly in the frequency and intensity of North Atlantic tropical cyclones, particularly if the ENSO event is weak.

Tropical cyclone rainfall anomalies for each individual year are seen in Fig. 11. The rainfall anomalies are constructed by subtracting the mean July–November climatological tropical cyclone rainfall from the mean July–November tropical cyclone rainfall for each of the 11 yr. In general the magnitude of the anomalies in the Atlantic basin are smaller than those discussed by Rodgers et al. (2000) for the North Pacific. This difference between oceans is due to the less frequent and somewhat weaker North Atlantic cyclones (i.e., compare Table 1 of Rodgers et al. 2000 with Table 1 of this paper).

Fig. 11.

A plan view showing the SSM/I-derived mean monthly North Atlantic tropical cyclone rainfall anomalies (mm month−1) for 1987–89 and 1991–98 during Jul–Nov period. White background designates nonraining land and oceans areas. The light and dark shades delineate, respectively, the positive and negative rainfall anomalies

Fig. 11.

A plan view showing the SSM/I-derived mean monthly North Atlantic tropical cyclone rainfall anomalies (mm month−1) for 1987–89 and 1991–98 during Jul–Nov period. White background designates nonraining land and oceans areas. The light and dark shades delineate, respectively, the positive and negative rainfall anomalies

Table 1.

The total number of tropical cyclones and those of depression, storm, and hurricane intensity observed from all available SSM/Is in the western and central North Atlantic during Jul–Nov of the three El Niño and La Niña years

The total number of tropical cyclones and those of depression, storm, and hurricane intensity observed from all available SSM/Is in the western and central North Atlantic during Jul–Nov of the three El Niño and La Niña years
The total number of tropical cyclones and those of depression, storm, and hurricane intensity observed from all available SSM/Is in the western and central North Atlantic during Jul–Nov of the three El Niño and La Niña years

The time history of the Niño-3.4 (i.e., 5°N–5°S and 120°–170°W) SSTs seen in Fig. 12 indicates that during the July–November months of this 11-yr period there were three El Niño–La Niña events that clearly showed warm SST anomalies followed by cool SST anomalies (1987/88, 1994/95, and 1997/98) (Barnston and Ropelewski 1992). These 2-yr couplets are delineated in Fig. 13 as an E for El Niño year and an L for La Niña year. The three couplets show clear evidence of an ENSO signal in the pattern (Fig. 11) and statistics (Fig. 13) of the tropical cyclone rainfall that would be expected in the North Atlantic. It is clear from Figs. 11 and 13 that the years with limited amount of tropical cyclonic rainfall during El Niño years are followed by a year with above-normal tropical cyclone rainfall during La Niña years. The La Niña effect in 1995 continues into 1996, with negative SST anomalies in the Pacific (Fig. 12) and large amounts and percentages of tropical cyclone rainfall in the Atlantic. An examination of SST patterns in the Atlantic indicates that the interannual variation of Atlantic SST is inconsistent with the precipitation changes and that the dominant influence is the Pacific-based ENSO circulation changes.

Fig. 12.

Time history of the SST anomalies (°C) in the Niño-3.4 region from 1985 to 1999. Light (dark) shades represent positive (negative) SST anomalies

Fig. 12.

Time history of the SST anomalies (°C) in the Niño-3.4 region from 1985 to 1999. Light (dark) shades represent positive (negative) SST anomalies

Fig. 13.

A histogram showing the SSM/I-derived nontropical cyclone (dark shade) and tropical cyclone (hatched shade) mean monthly rainfall amounts (mm month−1) for the (a) western and (b) eastern North Atlantic during the Jul–Nov period for 1987–89 and 1991–98. Percentage values indicate the fractional amount of rainfall contributed by tropical cyclones. Notations E and L represent El Niño and La Niña years as defined by the Niño-3.4 SST anomalies

Fig. 13.

A histogram showing the SSM/I-derived nontropical cyclone (dark shade) and tropical cyclone (hatched shade) mean monthly rainfall amounts (mm month−1) for the (a) western and (b) eastern North Atlantic during the Jul–Nov period for 1987–89 and 1991–98. Percentage values indicate the fractional amount of rainfall contributed by tropical cyclones. Notations E and L represent El Niño and La Niña years as defined by the Niño-3.4 SST anomalies

The North Atlantic tropical cyclone frequency and intensity data for the three El Niño–La Niña couplet years are seen in Table 1. The table reveals that there are also differences in the number and intensity of the North Atlantic tropical cyclones during the warm and cool ENSO events. The information in the table clearly demonstrates that the total number of western and eastern North Atlantic tropical cyclones is considerably greater during La Niña years as compared with during El Niño years. It is also observed that there are more intense western and eastern North Atlantic tropical cyclones during La Niña years. These statistical findings are not surprising and support the tropical cyclone rainfall information previously mentioned.

To examine further the relationship between tropical cyclone rainfall and the Niño-3.4 SST anomalies for the tropical cyclone season during the 11-yr period, a linear regression is performed. The linear-regression line and the correlation coefficient between the Niño-3.4 SST anomalies and western and eastern North Atlantic tropical cyclone rainfall are seen in Fig. 14. The figure clearly shows a negative correlation coefficient between the parameters of −0.52 and −0.67, respectively, in the western and eastern North Atlantic, whereas the correlation coefficient between Niño-3.4 SST anomalies and nontropical cyclone rainfall for both the western and eastern North Atlantic is nearly zero (figure not shown).

Fig. 14.

A scatterplot of the tropical eastern North Pacific SST anomalies in the Niño-3.4 region and the mean monthly rainfall amounts (mm month−1) contributed by tropical cyclones during Jul–Nov period for 1987–89 and 1991–98 for the (a) western and (b) eastern North Atlantic. The best-fit line (broad), the equation of the line, and the linear correlation coefficient are shown for each North Atlantic ocean basin

Fig. 14.

A scatterplot of the tropical eastern North Pacific SST anomalies in the Niño-3.4 region and the mean monthly rainfall amounts (mm month−1) contributed by tropical cyclones during Jul–Nov period for 1987–89 and 1991–98 for the (a) western and (b) eastern North Atlantic. The best-fit line (broad), the equation of the line, and the linear correlation coefficient are shown for each North Atlantic ocean basin

The difference between the El Niño–La Niña pattern for the nontropical cyclone and tropical cyclone rainfall across the western and eastern North Atlantic is seen in Fig. 15. The figures are constructed by taking the differences between El Niño years and La Niña years of the three couplets. The upper panel of the figure indicates that there is a decrease in the mean rainfall contributed by North Atlantic noncyclone systems along the ITCZ (i.e., from 30°W to Central America). On the other hand, there is a small increase in rainfall (<80 mm month−1) within the subtropical regions of the North Atlantic east of North America and between 30° and 50°W during El Niño years. These results agree in a general sense with those of Ropelewski and Halpert (1987, 1989), although their analysis was limited to rain gauges over land.

Fig. 15.

A plan view of the difference between the mean monthly rainfall amounts (mm month−1) contributed by the North Atlantic (top) nontropical cyclone systems and (bottom) tropical cyclones during El Niño years (1987, 1994, and 1997) minus the La Niña years (1988, 1995, and 1998). Only Jul–Nov rainfall is used for this calculation. Brown and gray backgrounds designate, respectively, land and ocean areas with little rainfall differences. The color bar delineates the magnitude of the rainfall differences

Fig. 15.

A plan view of the difference between the mean monthly rainfall amounts (mm month−1) contributed by the North Atlantic (top) nontropical cyclone systems and (bottom) tropical cyclones during El Niño years (1987, 1994, and 1997) minus the La Niña years (1988, 1995, and 1998). Only Jul–Nov rainfall is used for this calculation. Brown and gray backgrounds designate, respectively, land and ocean areas with little rainfall differences. The color bar delineates the magnitude of the rainfall differences

The lower panel of Fig. 15 (scale is smaller than upper panel) shows that tropical cyclone rainfall is less (by a magnitude of up to 35 mm month−1) over the majority of the North Atlantic during El Niño years than during La Niña years. The only exception is within a limited area of the North Atlantic subtropics north of the Lesser Antilles in which mean tropical cyclone rainfall was greater (<15 mm month−1) during El Niño years. It is interesting but not surprising to note that the enhanced vertical wind shear that is more prevalent during El Niño years influences not only the convection in tropical cyclones but convection within the ITCZ of the North Atlantic. On a percentage basis, the difference is much larger for the tropical cyclone rainfall than for the noncyclone rainfall.

The zonally averaged nontropical cyclone and tropical cyclone mean monthly rainfall during El Niño–La Niña years is shown in Fig. 16 for the western and eastern North Atlantic. The figure indicates that there is relatively little difference in the zonally averaged nontropical cyclone rainfall between El Niño and La Niña periods for both the western (Fig. 16b) and eastern (Fig. 16d) North Atlantic, although differences related to the patterns in Fig. 15 can be seen. Tropical cyclone rainfall is significantly less at essentially all latitudes during El Niño years as compared with La Niña periods, for both the western (Fig. 16a) and eastern (Fig. 16c) North Atlantic.

Fig. 16.

The SSM/I-derived 2.5°-lat zonally averaged mean monthly rainfall amounts (mm month−1) for the tropical cyclones and nontropical cyclones, respectively, for the (a), (b) western and (c), (d) eastern North Atlantic during El Niño (solid lines) and La Niña years (dashed lines). Only Jul–Nov rainfall is used for this calculation

Fig. 16.

The SSM/I-derived 2.5°-lat zonally averaged mean monthly rainfall amounts (mm month−1) for the tropical cyclones and nontropical cyclones, respectively, for the (a), (b) western and (c), (d) eastern North Atlantic during El Niño (solid lines) and La Niña years (dashed lines). Only Jul–Nov rainfall is used for this calculation

Conclusions and summary

Similar to the North Pacific tropical cyclone study (Rodgers et al. 2000), rainfall estimates made from satellite passive microwave (SSM/I) observations are used to estimate the North Atlantic monthly rainfall amounts contributed by tropical cyclones that became of storm intensity and greater. These rainfall estimates during the approximate years of 1987–98 were used to assess the impact of tropical cyclone rainfall on the geographical, seasonal, and intraannual distribution of total rainfall.

The main results of this study suggest the following.

  1. Tropical cyclones contribute, respectively, 4%, 3%, and 4% to the total western, eastern, and the entire North Atlantic rainfall during the tropical cyclone season. The greatest contributions of rainfall from tropical cyclones are nearly 30% and are found northeast of Puerto Rico, within the middle subtropical North Atlantic, and west of Africa. Much like the North Pacific, it appears that tropical cyclones augment the annual rainfall, particularly near the Lesser and Greater Antilles Islands, Mexico, Central America, and the southeastern United States, and can be a significant source of water for agriculture and other purposes.

  2. The maximum zonally averaged tropical cyclone rainfall is located poleward of the maximum zonally averaged nontropical cyclone rainfall within the North Atlantic. This fact is particularly true in the western North Atlantic, where tropical cyclones recurve northward under the influence of the westerlies.

  3. The greatest amount of tropical cyclone rainfall is contributed by systems of storm intensity in both the western and eastern North Atlantic. Cyclones of storm stage also have an apparently slightly higher mean rain rate than that of hurricane stage.

  4. There is no monthly lag between maximum tropical cyclone rainfall and maximum nontropical cyclone rainfall in the western North Atlantic; in the eastern North Atlantic, maximum tropical cyclone rainfall precedes maximum nontropical cyclone rainfall. This lag may be attributed to the fact that tropical cyclone rainfall is more dependent on the frequency of African easterly wave disturbances that reach their maximum in late August and early September, whereas nontropical cyclone rainfall is mostly dependent on the presence and strength of both the ITCZ and baroclinic systems that reach their combined maximum in October.

  5. Unlike the North Pacific tropical cyclone climatological rainfall study, the warm ENSO events have the opposite effect on the tropical cyclone rainfall over the North Atlantic by inhibiting tropical cyclone rainfall by subjecting these systems to greater upper-tropospheric ventilation and vertical wind shear. At the same time, warm or cool ENSO years had little influence on the nontropical cyclone rainfall. Also unlike the North Pacific tropical cyclone rainfall climatological study, the tropical cyclone rainfall differences during the couplet El Niño–La Niña years of 1987/88, 1994/95, and 1997/98 were much less in the North Atlantic basin than that found in the North Pacific, because of weaker and less frequent numbers of tropical cyclones.

  6. The satellite-estimated tropical cyclone rainfall observations during the couplet El Niño–La Niña years of 1987/88, 1994/95, and 1997/98 were consistent with the early North Atlantic tropical cyclone studies in that the amount of tropical cyclone rainfall was influenced more by atmospheric forcing than by sea surface energy flux processes.

Acknowledgments

Funding support was provided by NASA Office of Earth Science. The authors thank Dr. Ramesh Kakar of NASA Headquarters for his support of this work.

REFERENCES

REFERENCES
Adler
,
R. F.
,
H-Y. M.
Yea
,
N.
Prasad
,
W-K.
Tao
, and
J.
Simpson
.
1991
.
Microwave simulations of a tropical rainfall system with a three-dimensional cloud model.
J. Appl. Meteor
30
:
924
953
.
Adler
,
R. F.
,
G. J.
Huffman
, and
P. R.
Keehn
.
1994
.
Global tropical rain estimates from microwave-adjusted geosynchronous IR data.
Remote Sens. Rev
11
:
125
152
.
Barnston
,
A. G.
and
C. F.
Ropelewski
.
1992
.
Prediction of ENSO episodes using canonical correlation analysis.
J. Climate
5
:
1316
1345
.
Bove
,
M. C.
,
J. B.
Elsner
,
C. W.
Landsea
,
X.
Niu
, and
J. J.
O'Brien
.
1998
.
Effect of El Niño on U.S. landfalling hurricanes, revisited.
Bull. Amer. Meteor. Soc
79
:
2477
2482
.
DeMaria
,
M.
1996
.
The effect of vertical shear on tropical cyclone intensity change.
J. Atmos. Sci
53
:
2076
2087
.
Gershanov
,
A.
and
T. P.
Barnett
.
1998
.
Inderdecadal modulation of ENSO teleconnections.
Bull. Amer. Meteor. Soc
79
:
2715
2725
.
Goldenberg
,
S. B.
and
L. J.
Shapiro
.
1996
.
Physical mechanisms for the association of El Niño and West African rainfall with Atlantic major hurricane activity.
J. Climate
9
:
1169
1187
.
Gray
,
W. M.
1984a
.
Atlantic seasonal hurricane frequency. Part 1: El Niño and 30 mb quasi-biennial oscillation influence.
Mon. Wea. Rev
112
:
1649
1668
.
Gray
,
W. M.
1984b
.
Atlantic seasonal hurricane frequency. Part II: Forecasting its variability.
Mon. Wea. Rev
112
:
1669
1683
.
Gray
,
W. M.
,
J. D.
Sheaffer
, and
C. W.
Landsea
.
1997
.
Climate trends associated with multidecadal variability of Atlantic hurricane activity.
Hurricanes: Climate and Socioeconomic Impacts, H. F. Diaz and R. S. Pulwarty, Eds., Springer-Verlag, 15–53
.
Hollinger
,
J. T.
1991
.
DMSP Special Sensor Microwave/Imager calibration/validation final report.
Vol. 11. Naval Research Laboratory Tech. Rep., 12–25
.
Huffman
,
G. J.
and Coauthors,
.
1997
.
The Global Precipitation Climatology Project (GPCP) combined precipitation dataset.
Bull. Amer. Meteor. Soc
78
:
5
20
.
Hurell
,
J. W.
1995
.
Decadal trends in the North Atlantic oscillation: Regional temperature and precipitation.
Science
269
:
676
679
.
Jarvinen
,
B. R.
,
C. J.
Neumann
, and
M. A. S.
Davis
.
1984
.
A tropical cyclone data tape for the North Atlantic Basin, 1886–1983: Contents, limitations, and uses.
NOAA Tech. Memo. NWS NHC 22, 21 pp
.
Landsea
,
C. W.
and
W. M.
Gray
.
1992
.
The strong association between western Sahel monsoon rainfall and intense Atlantic hurricanes.
J. Climate
5
:
435
453
.
Neumann
,
C. J.
,
B. R.
Jarvinen
,
C. J.
McAdie
, and
G. R.
Hammer
.
1999
.
Tropical cyclones of the North Atlantic Ocean, 1871–1998.
National Climatic Data Center/Tropical Prediction Center/National Hurricane Center Historical Climatology Series 6-2, 206 pp
.
Pasch
,
R. J.
,
L. A.
Avila
, and
J-G.
Jing
.
1998
.
Atlantic tropical systems of 1994 and 1995: A comparison of a quiet season to a near-record-breaking one.
Mon. Wea. Rev
126
:
1106
1123
.
Pielke
,
R. A. Jr
and
C. N.
Landsea
.
1999
.
La Niña, El Niño, and Atlantic hurricane damages in the United States.
Bull. Amer. Meteor. Soc
80
:
2027
2034
.
Reuter
,
G. W.
and
M. K.
Yau
.
1986
.
Numerical modeling of cloud development in a sheared environment.
Beitr. Phys. Atmos
60
:
65
80
.
Reynolds
,
R. W.
and
T. S.
Smith
.
1994
.
Improved global sea surface temperature analyses using optimum interpolation.
J. Climate
7
:
929
948
.
Rodgers
,
E. B.
,
S. W.
Chang
, and
H. F.
Pierce
.
1994
.
A satellite-observational and numerical study of precipitation characteristics in western North Atlantic tropical cyclones.
J. Appl. Meteor
33
:
129
139
.
Rodgers
,
E. B.
,
R. F.
Adler
, and
H. F.
Pierce
.
2000
.
Contribution of tropical cyclones to the North Pacific climatological rainfall as observed from satellites.
J. Appl. Meteor
39
:
1658
1678
.
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
.
Ropelewski
,
C. F.
and
M. S.
Halpert
.
1989
.
Precipitation pattern associated with the high index phase of the Southern Oscillation.
J. Climate
2
:
268
284
.

Footnotes

Corresponding author address: Dr. Robert F. Adler, Mesoscale Atmosphere Processes Branch (Code 912), Laboratory for Atmospheres, NASA Goddard Space Flight Center, Greenbelt, MD 20771. adler@agnes.gsfc.nasa.gov