Consistency of upper-tropospheric water vapor measurements from a variety of state-of-the-art instruments was assessed using collocated Geostationary Operational Environmental Satellite-8 (GOES-8) 6.7-μm brightness temperatures as a common benchmark during the Atmospheric Radiation Measurement Program (ARM) First International Satellite Cloud Climatology Project (ISCCP) Regional Experiment (FIRE) Water Vapor Experiment (AFWEX). To avoid uncertainties associated with the inversion of satellite-measured radiances into water vapor quantity, profiles of temperature and humidity observed from in situ, ground-based, and airborne instruments are inserted into a radiative transfer model to simulate the brightness temperature that the GOES-8 would have observed under those conditions (i.e., profile-to-radiance approach). Comparisons showed that Vaisala RS80-H radiosondes and Meteolabor Snow White chilled-mirror dewpoint hygrometers are systemically drier in the upper troposphere by ∼30%–40% relative to the GOES-8 measured upper-tropospheric humidity (UTH). By contrast, two ground-based Raman lidars (Cloud and Radiation Test Bed Raman lidar and scanning Raman lidar) and one airborne differential absorption lidar agree to within 10% of the GOES-8 measured UTH. These results indicate that upper-tropospheric water vapor can be monitored by these lidars and well-calibrated, stable geostationary satellites with an uncertainty of less than 10%, and that correction procedures are required to rectify the inherent deficiencies of humidity measurements in the upper troposphere from these radiosondes.
It is well known that better understanding of characteristics of upper-tropospheric water vapor (UTWV) is an essential part in the prediction of future climate (Held and Soden 2000). Since the trapping of longwave radiation is proportional to the logarithm of water vapor concentration, small spatiotemporal changes in water vapor in the upper troposphere can have a significant radiative impact (e.g., Sohn and Schmetz 2004). Thus, various in situ and remote sensing instruments have been developed to observe upper-tropospheric water vapor more accurately. However, inconsistencies between measurements have been reported, due to inherently disparate characteristics of each instrument and/or a deficiency in some instruments, indicating that there is no perfect observation of upper-tropospheric water vapor (e.g., Ferrare et al. 1995; Soden and Lanzante 1996; Revercomb et al. 2003; Turner et al. 2003). Considering that inconsistent measurements may results in spurious climate signal (e.g., Ross and Gaffen 1998), it is of great importance to characterize the current ability of observing upper-tropospheric water vapor from various instruments, and to develop methods of reducing inconsistencies among these instruments based on a common, stable benchmark.
In 1995, the U.S. Department of Energy started the Atmospheric Radiation Measurement Program (ARM) to characterize and improve atmospheric water vapor measurements (Revercomb et al. 2003). In doing so, a series of water vapor intensive observation periods (IOPs) have been designed at the Southern Great Plains (SGP) Cloud and Radiation Test Bed (CART) site. Among these IOPs, the ARM First International Satellite Cloud Climatology Project (ISCCP) Regional Experiment (FIRE) Water Vapor Experiment (AFWEX) was conducted during 27 November–15 December 2000, focusing on developing techniques to reduce uncertainties in measurements of upper-tropospheric water vapor (Tobin et al. 2002; Revercomb et al. 2003). Measurements from a variety of in situ, ground-based, and airborne instruments were taken in conjunction with radiance measurements at 6.7 μm from the Geostationary Operational Environmental Satellite (GOES) during AFWEX.
Clear-sky 6.7-μm (i.e., water vapor channel) brightness temperatures are closely associated with weighted layer mean relative humidity between 200 and 500 hPa (i.e., upper troposphere) (e.g., Soden and Bretherton 1993; Sohn et al. 2000; Chung et al. 2007). Figure 1 shows the characteristics of GOES-8 water vapor channel in terms of the normalized weighting function. Temperature and humidity profiles of a standard midlatitude summer atmosphere and a standard midlatitude winter atmosphere were used to compute weighting functions for two satellite viewing angles (0° and 48.49°). It is noted that the water vapor content for the summer atmosphere is roughly 2 or 3 times that for the winter atmosphere. Although the vertical location of the peak depends on the humidity change (also temperature change in part) and satellite viewing angle, Fig. 1 indicates that the weighting function of the GOES-8 water vapor channel generally peaks between 200 and 500 hPa. Furthermore, considering that geostationary satellites have good spatiotemporal coverage, it is reasonable to use collocated water vapor channel measurements from geostationary satellite as a common, stable benchmark for intercomparing upper-tropospheric water vapor measurements from diverse instruments. Adopting this approach, Soden et al. (2004) compared radiosonde (Vaisala model RS80-H) and Raman lidar observations of upper-tropospheric water vapor with collocated GOES radiances at 6.7 μm during the previous four ARM IOPs, and found that Vaisala radiosonde observations exhibit a systematic dry bias relative to GOES and Raman lidar measurements.
In this study, extending the work of Soden et al. (2004), we intend to assess the consistency of upper-tropospheric water vapor measured from a variety of radiosondes and lidars based on collocated GOES-8 6.7-μm brightness temperatures during AFWEX, and to suggest a reliable method for acquiring an accurate and consistent upper-tropospheric water vapor dataset that can be used as an important climate monitoring tool.
2. Data and methodology
The Vaisala RS80-H radiosondes (hereinafter Vaisala) have been launched as the primary radiosonde to measure vertical profiles of temperature and humidity during AFWEX. In addition to Vaisala radiosonde, two additional radiosondes were used [i.e., the Meteolabor Snow White chilled-mirror dewpoint hygrometer (hereinafter Chilled-Mirror) and a Sippican Mark II Microsonde carbon hygristor radiosonde (similar to VIZ radiosonde, hereinafter VIZ-CH)]. The Chilled-Mirror and VIZ-CH sensors were launched together on the same balloon. The characteristics and accuracy of these radiosonde humidity sensors can be found in Ferrare et al. (2004), Miloshevich et al. (2004), and Soden et al. (2004).
Two ground-based Raman lidars [the CART Raman lidar (CARL) and the National Aeronautics and Space Administration (NASA) Goddard Space Flight Center scanning Raman lidar (SRL)] equipped with Nd:YAG laser have also been operated during AFWEX and have produced profiles of water vapor mass mixing ratio from the measurements of Raman scattering from water vapor and nitrogen (Goldsmith et al. 1998; Whiteman and Melfi 1999). It is noted that Raman lidar measurements of upper-tropospheric water vapor are only available during night (Ferrare et al. 2004). In addition to ground-based lidars, NASA’s Lidar Atmospheric Sensing Experiment (LASE) flown on the NASA DC-8 has been operated to measure absolutely calibrated water vapor profiles above and below the DC-8 in the nadir and zenith modes simultaneously (Tobin et al. 2002; Revercomb et al. 2003; Ferrare et al. 2004). LASE uses the differential absorption lidar (DIAL) technique to observe high-resolution profiles of water vapor and aerosol backscatter (Ismail et al. 2000). Six DC-8 flights were made over the ARM SGP site between 30 November and 10 December 2000, collecting approximately 26 h of profile data (Ismail et al. 2002; Ferrare et al. 2004).
To assess the relative consistency of the ARM measurements of upper-tropospheric water vapor, GOES-8 brightness temperatures in the 6.7-μm water vapor channel are used as a common benchmark. Measurements of the GOES pixel (spatial resolution of ∼4 km) closest to the SGP CART site in northern Oklahoma were selected every 30 min from 27 November to 15 December 2000. The noise equivalent delta temperature (NEΔT) for the GOES-8 6.7-μm channel is estimated to be ∼0.2 K (0.3 K) at 290 K (230 K) (Menzel et al. 1998; Ellrod et al. 1998). In addition to 6.7-μm brightness temperatures (T6.7), we use the 11-μm brightness temperatures (T11) to estimate the cloud-top temperature, and to discard pixels believed to be contaminated by high- and/or midlevel clouds. Cloud-contaminated pixels are determined using the method of Soden (1998), that is, pixels with brightness temperature differences (ΔTb = T11 − T6.7) less than 25 K are determined to be cloud contaminated. It is noted that results are not sensitive to the chosen temperature threshold values. For example, Soden (1998) found that changing the threshold temperature from 25 to 30 K results in root-mean-square (rms) differences in seasonal-mean upper-tropospheric humidity of <1%.
b. Comparison procedures
Relative consistency of upper-tropospheric water vapor measurements from radiosondes and lidars is evaluated by directly comparing the radiosonde and lidar measurements with the GOES-8 water vapor channel observations using a forward modeling approach. In the case of radiosondes, moisture (relative humidity or dew/frostpoint temperature) and coincident temperature profiles are inserted into a radiative transfer model [the High Resolution Infrared Radiation Sounder (HIRS) Fast Forward Program; Joiner et al. 1998] to calculate the radiance observed by GOES-8 under those conditions. A satellite zenith angle of 48.49° is used to account for the viewing angle for the SGP central facility (36.609°N, 97.485°W) from GOES-8. By contrast, only water vapor mixing ratio profiles are produced from lidar observations. Therefore, Vaisala-observed temperature profiles are interpolated to lidar observation times from the nearest radiosonde launches.
To facilitate the interpretation of this comparison, both the observed and simulated radiances are consistently transformed to a vertically weighted average of the upper-tropospheric relative humidity (UTH; Soden and Bretherton 1993), that is,
where θ and p0 denote the satellite zenith angle and a normalized reference pressure (i.e., the pressure of a 240-K isotherm divided by 300 hPa), respectively. Regression coefficients a and b are tuned to replicate detailed radiative transfer calculations.
Since the normalized reference pressure is approximately constant for a short time period at a fixed location, the variation of p0 is much smaller than humidity variation. Therefore, differentiating Eq. (1) with an assumption of constant p0 results in the following approximate form (Soden 1998):
3. Comparison based on GOES-8 measurements
Figure 2 shows the time series of GOES-8 6.7-μm brightness temperature (solid line) from 27 November to 15 December 2000 over the SGP central facility. The UTH scale corresponding to the brightness temperature is derived from Eq. (1) with a constant value of p0 = 1.3 and is denoted on the right-hand side of the graph. Note that the UTH increases exponentially with a decrease in 6.7-μm brightness temperature. Shaded regions indicate the presence of high- and/or midlevel clouds that are estimated from the cloud-screening method.
Observed GOES-8 6.7-μm brightness temperatures range from 215 to 250 K during AFWEX with a large temporal variability. The UTH values indicating supersaturation are occasionally shown, due to the presence of thin cirrus clouds that were not detected from the infrared-based cloud-screening algorithm (e.g., Chung et al. 2008; Kim et al. 2009). It is also noted that observed brightness temperature exhibits a distinct diurnal variation (e.g., Chung et al. 2007), implying that more frequent in situ and ground-based observations are needed to resolve high-frequency variation of upper-tropospheric water vapor. Furthermore, considering that water vapor is the primary absorber of longwave radiation in clear-sky conditions, frequent measurements of upper-tropospheric water vapor with high accuracy are crucial to determine outgoing longwave radiation, and to validate radiative transfer model calculations (Tobin et al. 2002).
Simulated brightness temperatures from the profiles of humidity and temperature from Vaisala (stars), Chilled-Mirror (triangles), and VIZ-CH (circles) radiosondes are compared to observed GOES-8 measurements in Fig. 2. Despite the uncertainty of radiosonde humidity sensors in very cold and dry conditions (Miloshevich et al. 2004), the temporal variation of simulated brightness temperatures is generally consistent with that of GOES-8 observations. However, the radiosonde-simulated brightness temperatures show a distinct warm bias (and thus dry bias) compared to GOES-8 observations except for the VIZ-CH humidity sensor. This upper-tropospheric dry bias is consistent with the results of previous studies (e.g., Ferrare et al. 1995, 2004; Soden and Lanzante 1996; Turner et al. 2003; Miloshevich et al. 2004; Soden et al. 2004).
The left panel of Fig. 4 shows a scatterplot of radiosonde-simulated versus GOES-8 observed brightness temperatures. The GOES-8 pixels contaminated by clouds based on the cloud classification scheme described in section 2 are excluded from the comparison. As shown in Fig. 2, Vaisala and Chilled-Mirror sensors exhibit a significant dry bias relative to collocated GOES-8 observations. Relatively larger differences from GOES-8 under cold brightness temperatures may be attributable to thin cirrus that is not detected from the cloud-screening method. On the other hand, this bias is not distinct in the case of the VIZ-CH humidity sensor.
Statistical results for comparison between the radiosonde-simulated and GOES-8 observed brightness temperatures are given in Table 1. The bias and rms of Vaisala humidity sensor relative to GOES-8 are 3.5 and 4.0 K, respectively. This bias corresponds to ∼30%–40% relative bias in UTH according to Eq. (2). These values are similar to the results of Soden et al. (2004), confirming that Vaisala radiosondes have inherent problems in measuring water vapor in the cold and dry conditions. Although the magnitude is much smaller than this study and Soden et al. (2004), a similar dry bias of Vaisala humidity sensor was noted in Ferrare et al. (2004): they showed that the water vapor profiles from the Vaisala humidity sensor are about 10% drier in the upper troposphere by comparing with LASE measurements during AFWEX. The dry bias of the Vaisala humidity sensor is attributable to sensor calibration error at low temperatures, contamination by the packing material, and a time-lag error due to slow sensor response to changes in the ambient relative humidity at low temperatures (Miloshevich et al. 2003, 2004, and references therein). A couple of studies have shown that applying moisture correction schemes to radiosonde measurements reduces the inherent dry bias in the upper troposphere (Tobin et al. 2002; Ferrare et al. 2004; Miloshevich et al. 2004).
Although the number of collocations is small, the bias and rms of the Chilled-Mirror sensor are similar to those of Vaisala; that is, Chilled-Mirror radiosondes are also ∼30%–40% drier in the upper troposphere relative to the satellite measurements. A slightly smaller dry bias of 10%–15% was noted in the comparison with LASE during AFWEX (Ferrare et al. 2004), confirming the problem of the Chilled-Mirror sensor under very cold, dry conditions. Unlike the Vaisala and Chilled-Mirror sensors, the VIZ-CH humidity sensor exhibits a slight moist bias of 0.1 K relative to GOES-8 observations. However, the correlation is low, suggesting that the smaller bias is a fortuitous result of compensating errors and not the result of a more accurate measurement. For example, a larger hysteresis effect may be offsetting the sensor’s dry bias (Soden et al. 1994). From comparison with other instruments, Ferrare et al. (2004) showed that the VIZ-CH humidity sensor exhibits poor performance at low relative humidity.
The time series of brightness temperatures simulated from CARL (stars), SRL (triangles), and LASE (circles) are shown in Fig. 3. Although the number of SRL and LASE observations is smaller than that of CARL since the SRL and LASE measurements were taken only for a couple of days, all three lidars show generally good agreement with GOES-8 observations. The collocated GOES–lidar pairs in the absence of high- and/or midlevel clouds are denoted in Fig. 4b. In contrast to the radiosonde cases, simulated brightness temperatures using humidity profiles measured from lidars do not show a distinct dry bias with respect to the collocated GOES-8 observations. Considering that temperature profiles from Vaisala sensors are combined with lidar moisture profiles when performing the radiative transfer calculations, it is clear that differences in brightness temperatures arise from differences in the moisture profiles and not the temperature profile.
Table 1 shows statistical results of the comparison with collocated GOES-8 observations (see the lower three rows). Bias ranging from 0.5 to 0.9 K indicates that a relative error in UTH is within approximately 10%. Such a difference between the GOES-8 and lidar measurements implies that both GOES-8 and lidar observations meet the required accuracy (<10%) for the upper-tropospheric water vapor measurements (Tobin et al. 2002, 2006). The rms difference of each lidar is much less than that of the radiosonde humidity sensors. In addition, correlation coefficients are generally larger than those for three radiosondes. Therefore, these results indicate that lidar observations of upper-tropospheric water vapor are more consistent with GOES-8 measurements. From this comparison, it suggests that radiosonde humidity measurements are drier in the upper troposphere by ∼20%–30% relative to all three lidars.
Using LASE as a reference during AFWEX, Ferrare et al. (2004) assessed the accuracy of CARL and SRL at night, and found that both CARL and SRL exhibit more moist upper troposphere relative to LASE (see Fig. 9 of Ferrare et al. 2004). By contrast, Table 1 indicates that simulated brightness temperatures from CARL and SRL show a slight warm (and thus dry) bias compared to that of LASE. The disparity between Ferrare et al. (2004) and this study may be attributable to differences in sample size of collocation, spatial coverage, and temporal sampling (refer to Table 1 of Ferrare et al. 2004).
However, it is likely that differing approaches to the vertical averaging also contribute. Specifically, Ferrare et al. (2004) compare vertical averages of upper-tropospheric mixing ratio, whereas the 6.7-μm brightness temperatures measure a vertically averaged upper-tropospheric relative humidity. Since the vertical average of a mixing ratio will be weighted toward lower levels where the radiosonde sensors are considered more reliable, one would expect them to exhibit a smaller bias than the vertically averaged relative humidity. Disparity in the radiosonde dry bias between this study and Ferrare et al. (2004) also seems to be attributable to the same reason.
To examine the possibility that such a difference can be produced from the same humidity profile, we conducted sensitivity tests showing how the magnitude of dry bias of radiosonde humidity sensors is affected by a chosen variable for comparison. For the standard midlatitude summer atmosphere and standard midlatitude winter atmosphere shown in Fig. 1, water vapor mixing ratio is linearly decreased from 0% at 600 hPa to 20% (case 1), to 40% (case 2), and to 60% (case 3) at 200 hPa from the original profiles as shown in the left panel of Fig. 5 with an assumption that humidity measurements are more reliable toward lower levels in the upper troposphere. It is noted that the original humidity profiles are used for the pressure levels greater than 600 hPa.
Radiative transfer simulations were carried out using the three modified humidity profiles with the original temperature profile. A satellite zenith angle of 48.49°, corresponding to the viewing angle for the SGP central facility from GOES-8, was used for the brightness temperature simulations. Then, simulated brightness temperature for the three cases was subtracted from the value computed from the profiles shown in Fig. 1 to obtain the brightness temperature difference (ΔTb). In addition, water vapor between 200 and 600 hPa was integrated to produce upper-tropospheric water vapor, that is,
where q and g denote specific humidity and gravitational acceleration, respectively. Using the UTWV value of the original profile, the fractional UTWV difference was computed for the three cases and compared with brightness temperature difference. The right panel of Fig. 5 shows that the dry bias increases from 4.3% (3.7%) to 12.9% (11.0%) for the standard midlatitude summer (winter) atmosphere in terms of fractional UTWV difference. By contrast, the brightness temperature difference increases from 0.9 (0.6) to 3.1 K (1.9 K) for standard midlatitude summer (winter) atmosphere. Equation (2) implies that these values of brightness temperature difference approximately correspond to fractional UTH differences of 9% (6%)–31% (19%) in the case of the standard midlatitude summer (winter) atmosphere. Considering that the water vapor amount exponentially decreases with height (e.g., see the left panels of Fig. 1), and that the accuracy of radiosonde humidity measurements is not much higher in the cold and dry conditions, it is possible that a larger dry bias can be explained in terms of fractional UTH difference compared to fractional UTWV difference. Thus, these sensitivity tests validate that the magnitude of dry bias can be varied depending on which quantity between relative humidity and mixing ratio is vertically integrated for comparison.
4. An assessment of the absolute accuracy of GOES-8 measurements
While the spatiotemporal resolution of GOES-8 makes it an effective relative benchmark, it is also important to evaluate the absolute accuracy of the GOES-8 water vapor channel measurements (e.g., Sohn et al. 2000). For this purpose, we compared the GOES-8 radiances with measurements from the National Polar-orbiting Operational Environmental Satellite System (NPOESS) Aircraft Sounder Test Bed-Interferometer (NAST-I), an airborne high-spectral resolution infrared radiometer flown as part of AFWEX. NAST-I measures the influence of water vapor on the infrared emission. The GOES-8 radiances were collocated to NAST-I observation for four flights (29 November, 4, 6, and 9 December). The high-resolution NAST-I spectra were convolved with the response function for the GOES-8 6.7-μm channel, and a limb correction based on Eq. (1) was applied to account for differences in viewing angle between GOES-8 and NAST-I.
Figure 6 shows the brightness temperature difference between NAST-I and GOES-8 as a function of NAST-I field-of-view (FOV) angle ranging from −45° to 45°. The bias for each flight averaged over all view angles is 0.41 K (29 November), −0.1 K (4 December), 0.27 K (6 December), and −0.45 K (9 December). There is slight evidence of a dependence of the bias on NAST-I view angle. Indeed the detection of such a subtle bias is somewhat remarkable given the uncertainties in both measurements. These results suggest that GOES-8 radiance measurements agree with NAST-I within ∼0.5 K. While there is a small tendency for the GOES-8 brightness temperatures to be slightly smaller (in other words, wetter) compared to NAST-I, this comparison suggests that the calibration errors on GOES-8 are too small to explain a significant portion of the bias between GOES-8 and the radiosondes/lidars, thereby implying that the brightness temperatures of the 6.7-μm water vapor channel from well-calibrated geostationary satellites can be used as a reliable benchmark for evaluating the consistency of upper-tropospheric water vapor measurements from a variety of humidity sensors.
Meanwhile, it should be noted that recent satellite sensors on board polar-orbiting satellites such as Microwave Limb Sounder (MLS), Atmospheric Infrared Sounder (AIRS), Infrared Atmospheric Sounding Interferometer (IASI), and NPOESS are observing upper-tropospheric water vapor with an enhanced spectral, vertical resolution. Since these sensors are expected to continue providing more detailed observations of upper-tropospheric water vapor, it is of great importance to reconcile these measurements with the historic observations in the perspective of the long-term climate monitoring.
5. Summary and conclusions
In this study, radiance observations from the GOES-8 6.7-μm channel were used as a common benchmark to assess the consistency between upper-tropospheric water vapor measurements from different instruments. Using a profile-to-radiance approach, upper-tropospheric water vapor profiles from radiosondes (i.e., the Vaisala RS80-H radiosonde, the Meteolabor Snow White chilled-mirror dewpoint hygrometer, and the Sippican Mark II Microsonde carbon hygristor radiosonde) and lidars (i.e., CARL, SRL, and LASE) were compared with collocated GOES-8 measurements during AFWEX at the SGP CART central facility site.
Comparison showed that radiosonde profiles (Vaisala RS80-H radiosondes and Meteolabor Snow White chilled-mirror dewpoint hygrometers) are ∼30%–40% drier in the upper troposphere relative to the GOES-8 UTH. By contrast, three lidars (CARL, SRL, and LASE) agree to within 10% in terms of UTH. These results indicate that correction procedures for observations from radiosonde humidity sensors are required to obtain upper-tropospheric water vapor consistent with other instruments. In addition, good agreement between GOES-8 and collocated NAST-I measurements validates the absolute calibration of the geostationary satellite observations.
Meanwhile, it should be noted that UTH cannot be estimated from the IR observations in the presence of high- and/or midlevel clouds because of the cloud contamination. Considering that those clouds only develop under favorable atmospheric conditions (i.e., atmosphere with higher moisture content), excluding the hypothetical clear-sky humidity in cloudy regions renders the IR-based clear-sky UTH drier in comparison with the all-sky value (e.g., Sohn et al. 2006; Sohn and Bennartz 2008). By contrast, since microwave radiation is less sensitive to the presence of those clouds than the IR observations, UTH can be estimated in both clear and cloudy areas. Thus, limitation due to the inherent sampling bias of IR observations in the presence of those clouds should be considered in climate studies.
In conclusion, this study showed that geostationary satellite measurements can provide an effective tool for intercomparing upper-tropospheric water vapor from different instruments, and that upper-tropospheric water vapor can be monitored by lidars and satellites with an uncertainty of less than 10%. We expect that the upper-tropospheric water vapor dataset constructed from synergistic use of geostationary satellite and lidars can play a crucial role in climate monitoring and climate model evaluation.
Considering that instrumental changes in upper-tropospheric water vapor observation can make a serious impact on long-term climate monitoring, it is of great importance to provide a globally consistent and accurate dataset. From this perspective, it should be emphasized that the current ongoing efforts of the Global Space-based Inter-Calibration System’s (GSICS; http://gsics.wmo.int/index_en.html) activities play a crucial role in reconciling a variety of satellite observations, and thus in the long-term climatological studies.
We thank Dave Tobin for providing the NAST-I radiances, Rich Ferrare and David Whiteman for providing the lidar moisture profiles, and the DOE/ARM Program for supporting and distributing data from the AFWEX program. The authors thank three anonymous reviewers for their constructive and valuable comments, which led to an improved version of the manuscript. The authors also thank Prof. Byung-Ju Sohn of Seoul National University for valuable discussion. This research was supported by a grant from the NOAA Climate Program Office.
Corresponding author address: Dr. Brian J. Soden, Rosenstiel School of Marine and Atmospheric Science, University of Miami, 4600 Rickenbacker Causeway, Miami, FL 33149. Email: email@example.com