Quantifying the relationship of large-scale environmental conditions such as relative humidity with hurricane intensity and intensity change is important for statistical hurricane intensity forecasts. Our composite analysis of 9 years of Atmospheric Infrared Sounder (AIRS) humidity data spanning 198 Atlantic tropical cyclones (TCs) shows that environmental relative humidity (ERH) above the boundary layer generally decreases with time as TCs evolve. Near the surface, ERH stays approximately constant. ERH generally increases with increasing TC intensity and intensification rate. Rapidly intensifying TCs are associated with free tropospheric ERH more than 10% (relative to the averaged ERH for all TCs) larger than that for weakening TCs. Substantial azimuthal asymmetry in ERH is also found, especially for the TCs attaining the highest intensities and largest intensification rates at distances greater than 400 km away from the TC center. In the front-right quadrant relative to TC motion, rapid intensification is associated with a sharp gradient of ERH in the upper troposphere, with a decrease from the near to the far environment between 400 hPa and 300 hPa. The ERH gradient weakens with the decrease of intensification rate. This radial ERH gradient might be a useful predictor for the statistical forecast of TC intensification.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 While our understanding of tropical cyclones (TCs) has improved tremendously in the past several decades, forecasts of TC genesis, spin-up and subsequent (especially sudden) intensity changes still present significant challenges. Official intensity forecasts from the National Hurricane Center (NHC) for Atlantic and Eastern North Pacific TCs have not shown much improvement in the last 20 years [DeMaria et al., 2007]. This is because TCs are sensitive to many factors, within the storm and in its surrounding environment. For example, TC structure and intensity are sensitive to vertical wind shear in the environment [DeMaria, 1996; Frank and Ritchie, 2001; Zehr, 2003], which may be poorly forecasted by operational and research models. Also, relatively subtle variations in sea-surface temperature (SST) or ocean heat content can cause a TC intensity to shift several categories on the Saffir-Simpson scale within a short period of time [Sun et al., 2007].
 The available moisture in the TC's environment represents another poorly understood influence on intensity, thereby presenting a limit to predictability. While high mid-tropospheric relative humidity (RH) appears to be necessary for rapid intensification and the attainment of maximum intensity [e.g.,Kaplan and DeMaria, 2003; Emanuel et al., 2004; Hendricks et al., 2010; Kaplan et al., 2010], dry air intrusions have a negative influence on TC intensification as dry air ingestion promotes the formation of cold downdrafts, which transport low θeair into the sub-cloud layer and storm inflow [e.g.,Emanuel, 1989]. In an idealized modeling study, Braun et al.  showed that low humidity air reaching the inner core induces asymmetric convective activity which weakens TCs [e.g., Nolan and Grasso, 2003; Nolan et al., 2007]. They further showed that the time for TCs to reach maturity varies with the proximity of dry air to the center of circulation. When dry air is located 270 km away or further from the center of the vortex, its impact on TC intensity is insignificant.
 Some studies [e.g., Barnes et al., 1983; Wang, 2009] have shown that substantial and extensive moisture may also promote a net negative influence on TC strength by facilitating the formation of TC rainbands. The idealized modeling study of Hill and Lackmann , which varied the environmental RH (ERH) in the region ≥100 km beyond the TC core, suggests that larger ERH results in the establishment of wider TCs with more prominent outer rainbands. However, TC development, as measured by time series of maximum 10 m wind speeds, was nearly insensitive to ERH despite the variation in rainband activity.
Kaplan and DeMaria examined the mid-tropospheric (850–700 hPa) ERH relation with rapidly intensifying (RI) TCs in the North Atlantic basin using the NHC HURDAT file [Jarvinen et al., 1984] and the Statistical Hurricane Intensity Prediction Scheme (SHIPS) [DeMaria and Kaplan, 1999] database. Hendricks et al. conducted composite analyses using the Navy Operational Global Atmospheric Prediction System (NOGAPS) global analysis. Both studies found that RI events over the Atlantic basin are associated with larger RH in the middle troposphere than non-RI events.
 Analyses using satellite observations have been rather limited. Shu and Wu  examined the influence of the Saharan air layer (SAL) on TC intensity with three years of RH data from the Atmospheric Infrared Sounder (AIRS) instrument. They defined the SAL intrusion in the AIRS RH data as the nearest location of dry (RH ≤ 30%) air between 600 and 700 hPa. Their analysis incorporating 37 TCs during 2005–2007 suggested that the dry SAL air had a favorable influence on TC intensity when present in the northwest quadrant of TCs but a negative impact when the dry air approached to within 360 km, mostly in the southwest and southeast quadrants.
 In this study, we examine all TCs over the North Atlantic from 2002 to 2010. The RH analyses are stratified with respect to the radial distance from the TC center, altitude, maximum intensity attained by the TCs, and intensification rate. The primary goals of this study are to quantify the relationships between ERH and TC intensity and intensification rate, and improve our understanding of the impact of environmental moisture on TC development. In particular, the results of this study may help improve statistical models, which still show high skill in TC intensity forecasts when compared to advanced mesoscale numerical models [Kaplan et al., 2010].
2. Data and Method
 The AIRS onboard the Aqua satellite since 2002 has provided near-daily global coverage of the tropospheric water vapor profile from space [Divakarla et al., 2006; Susskind et al., 2003]. The AIRS RH retrievals sample a broad ∼1300 km swath at approximately 01:30 and 13:30 local time with a horizontal resolution of ∼45 km near nadir. We use the Level 2 RH retrieval (version 5). The relative uncertainty of the RH retrieval is estimated to be 9% at 250 hPa and below, with no systematic bias [Gettelman et al., 2006]. The six-hourly best track data for North Atlantic TCs are obtained from the Automated Tropical Cyclone Forecasting System (ATCF) at the NHC. This study summarizes the statistical behavior of 198 North Atlantic TCs, 74 of which achieved at least Category 1 intensity on the Saffir-Simpson scale, with a total of 2914 samplings observed by AIRS during the period of 2002 and 2010.
 Composites of ERH with respect to radial distance from the TC center, altitude, and quadrant with respect to TC motion are constructed. The TC center position at the local AIRS observational time is linearly interpolated from the best track data. Three zones of radial distances from the TC center are defined: the near environment (200–400 km), intermediate environment (400–600 km) and far environment(600–800 km). Using the best track data, four quadrants are established relative to TC motion in this study, numbered clockwise from the TC's front-right (Q1) to front-left sides (Q4). As TCs in the North Atlantic preferentially move westward, Q1 (Q4) roughly corresponds to the northwest (southwest) quadrant in geographic coordinate. There is a long tradition of using motion-based coordinates in composite construction [e.g.,George and Gray, 1976], and motion itself contributes to storm asymmetry along with vertical wind shear and friction [e.g., Corbosiero and Molinari, 2003; Chen et al., 2006; DeMaria, 1996; Shapiro, 1983]. Future work will consider other coordinate systems, such as those based on vertical shear.
3.1. Composite ERH as a Function of Time
 Each TC is examined for the ±72 h period around its time of maximum intensity (Tmax). A composite temporal evolution of ERH is obtained by averaging ERH for each TC at the same time relative to Tmax. As shown in Figure 1, although individual measurements are quite scattered, the composite average ERH near surface (indicated by 1000–925 hPa layer) is about 82% for all four quadrants, with small variations throughout the 6-day period. ERH decreases with time at all altitudes above the boundary layer through the middle troposphere, and for all radial distances outward from the TC center. Furthermore, the magnitude of ERH declines from the near to the far environment. The average 500–600 hPa ERH in Q1 within the near environment is 56% at 72 h prior to peak intensity, dropping to 52% at Tmax and further diminishing to 37% by Tmax+ 72 h. Over that same 6-day period, the far environment ERH at the same level declines from 42% to 36%. This ERH trend is possibly a result of TC-induced subsidence bringing down dry air from above that desiccates the lower and middle troposphere. Land influences could also play a role as TCs translate west- and northwest-ward. The physical factors contributing to the temporal drying effect warrant further investigation.
3.2. Composite ERH as a Function of TC Intensity
 The maximum wind speed (Vmax) from the best track data is used as an index for TC intensity and the ERH is then stratified with respect to TC intensity. Observed ERHs are normalized by the mean RH profile for the 198 TCs (see Figure S1 in the auxiliary material) and plotted as a function of Vmax for different radial distances from the TC center in Figure 2.
 In the near environment (Figure 2a), the TCs attaining the largest intensities (Category 5) possess a pronounced tendency towards having larger middle and upper tropospheric RH. This is seen in all quadrants and at all altitudes above the boundary layer. However, the changes of ERH with TC intensity are not linear. The correlation between the TC intensity and RH between 850 and 700 hPa (RH850) is 0.03 at Q1, but not statistically significant. The differences among TC categories are not always statistically significant. At radial distances exceeding 400 km from the TC center (Figures 2b and 2c), the composite ERH displays significant azimuthal asymmetry above the boundary layer. Relative to TC motion, the front quadrants (Q1 and Q4) have smaller mid-tropospheric RH while the rear quadrants (Q2 and Q3) have larger, especially in the far environment of Category 5 cases, where RH between 400 and 300 hPa (RH400) is 23% in Q1 and 51% in Q3.
3.3. Composite ERH as a Function of TC Intensification Rate
 The ERH is further stratified with respect to the TC intensification rate. The intensification rate at a particular time is defined as the Vmax difference between that time and 6 hours later. Five intensity change bins are defined: rapidly intensifying (RI), intensifying (I), neutral (N), weakening (W) and rapidly weakening (RW). RI (RW) corresponds to the top (bottom) 5% and the other three equally sample the rest of intensification rates for the 198 TCs. Following Hendricks et al. , RW cases are not included in our discussions. The ranges of intensification rate and the sample sizes for each category are given in Table 1.
Table 1. Averaged RH for Weakening (W:−4.75 ≤ ∆Vmax < −0.75 m s−1 per 6 hrs), Neutral (N: −0.75 ≤ ∆Vmax < 2.25 m s−1 per 6 hrs), Intensifying (I: 2.25 ≤ ∆Vmax < 4.75 m s−1 per 6 hrs), Rapidly Intensifying (RI: ∆Vmax ≥ 4.75 m s−1 per 6 hrs) Cases, and the Differences Between RI and the Other Groupsa
RI – W
RI – N
RI – I
Intensification rate (ΔVmax) is defined as the 6-hour Vmax change. The sample size for each category is: 455 (W), 1592 (N), 500 (I) and 191 (RI). The second column is the mean value averaged for all 198 TC cases. The last column (Corr) is the correlation between the intensification rate and RH. The bold face denotes statistical significance at the 95% confidence level. The unique feature of RH400 in Q1 is italicized.
RH850 (850–700 hPa RH, %)
RH400 (400–300 hPa RH, %)
 Azimuthal asymmetry above the boundary layer in the intermediate and far environments is evident, in particular during RI (Figure 3). Similar to the composites with respect to intensity (Figure 2), Q2 and Q3 are generally more moist, while Q1 and Q4 are drier. However, correlations between ERH and intensification rate are quite low for all quadrants and environmental sectors (Table 1), which appears to be consistent with Kaplan et al. , who found other environmental characteristics to be more skillful predictors of Atlantic basin RI. And yet, this is because simple linear relationships can obscure the potentially important variations discussed below.
 First, Table 1 and Figure 3 demonstrate that ERH tends to be positively associated with intensification rate, especially above the boundary layer in the near and intermediate environments. Storms undergoing RI possess larger than average ERH, while weakening TCs are below the mean for the 198 TCs. As an example, for Q1 within the intermediate environment, RH between 850 and 700 hPa (RH850) is 58% while weakening, increasing to 62% at the neutral stage, both being below the overall average (64.3%; see Table 1). RH850 is about average (64%) during the intensifying stage and further increases to above average (66%) for RI.
 This means pairwise ERH differences between intensification categories can be sizable and significant. The 7.5% separating RI and W at 850 hPa is more than 10% above the averages for the two categories, as well as the overall mean, and exceeds the AIRS measurement uncertainty at this level. Like the other values listed with bold type in Table 1, this is significantly different from zero at the 95% level.
 At the 400 hPa level, the ERH differences in the near environment are even greater, and a radial variation in upper tropospheric humidity emerges in Q1, particularly during RI (Figure 3 and Table 1). RH400 in the near environment shifts from below (for W and N) to above average (for I and RI), representing a RH change of about 9% (Table 1). Yet, for the far environment, the lowest ERH at this level is found at the RI stage for this quadrant, a decrease of 5% with respect to weakening TCs. It is striking that ERH actually decreases with intensification rate.
 Thus, as highlighted in Figure 4, the horizontal moisture gradientbetween the near and far environments in Q1 is largest during the RI stage, which is significant at the 99% level. This gradient is considerably smaller (but still statistically significant at the 90% level) during the intensifying stage, and of opposite sign for weakening cases. The RI stage's combination of larger and smaller ERH in the near and far environments, respectively, may reflect the influence of the storm-induced circulation or is possibly a controlling factor for TC intensification. This unique feature has not been documented before, and might yield a skillful predictor for statistical hurricane forecast models, potentially not less important than RH850 itself.
4. Conclusion and Discussion
 In this study, the ERH observed by AIRS is investigated in association with 198 TCs over the North Atlantic between 2002 and 2010. Composites of ERH with respect to radial distance from the TC center, altitude, and quadrant with respect to TC motion are constructed. The cases are also stratified with respect to time, TC intensity and intensification rate. The principal findings from this composite study of observational data are:
 1. ERH in the free troposphere decreases with time as TCs evolve while ERH in the boundary layer stays approximately constant within ±72 hours from the time that TCs reach maximum intensity. The ERH decrease in the free troposphere is possibly contributed by TC-induced subsidence and/or land influence.
 2. Higher intensity TCs tend to have larger ERH than lower intensity TCs although the trend is not linear and not always statistically significant.
 3. ERH above the boundary layer in the near environment generally increase with TC intensification rate. Rapidly intensifying TCs are associated with larger ERH than weakening and neutral TCs. However, the difference between rapidly intensifying and intensifying cases are not always statistically significant.
 4. The azimuthal asymmetry of ERH becomes evident at radial distances >400 km. The rear quadrants tend to have larger ERH and the front quadrants appear to have lower ERH.
 5. In the front-right quadrant (Q1), a sharp decrease in upper tropospheric (above 400 hPa) RH from the near to the far environment occurs during rapid intensification. This radial RH gradient is weaker for TCs with lower intensification rates. For weakening TCs, Q1 has slightly larger upper tropospheric ERH in the far environment than in the near environment. This radial RH gradient may reflect the influences of the storm-induced circulation or is possibly a controlling factor for TC intensification. This radial RH gradient might be a useful predictor for the forecast of TC intensification.
 The AIRS-centric investigation provides new insights regarding the environmental moisture within which TCs grow, decay, and propagate. Our findings show a systematic difference (on the order of several percent) between the storms of different intensity or intensification rate. This systematic difference represents a signal that cannot be simply dismissed via limitations inherent in the measurements. The relationship of ERH with TC intensity and intensification rate, especially its azimuthal and radial variations, may lead to improvements in TC intensity forecasts from statistical models.
 There are remaining questions that warrant further investigation, particularly in regards to whether the observed relationships represent the impact of ERH on TC development, or a more complex set of nonlinear interactions between a TC and its environment. For example, is the dry air in the front-right quadrant in the intermediate and far environments providing a favorable (by suppressing rainband convection) or detrimental influence on TC intensification? Or, is it simply a result of the TC circulation (from subsidence drying)? Additional numerical model experiments could help clarify the role of environmental moisture in TC evolution. This observational analysis will be valuable for the validation of numerical and statistical models.
 The authors thank Sundararaman Gopalakrishnan and Tomislava Vukicevic for valuable discussions, and Mark DeMaria for suggestions. The helpful comments from two anonymous reviewers are appreciated. The work is conducted at the Jet Propulsion Laboratory, California Institute of Technology, under contract with NASA. The authors thank the funding support from the NASA HSRP program.
 The Editor thanks the two anonymous reviewers for their assistance in evaluating this paper.