2166

IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 41, NO. 10, OCTOBER 2003

Global Mapping of Attenuation at Ku- and Ka-Band

V. Chandrasekar, Hiroki Fukatsu, and K. Mubarak
AbstractThe propagation of radio waves for earthspace slant path at C-band and higher frequencies are dominated by precipitation in the atmosphere. At a given frequency, attenuation depends on the length of the radio path, the size distribution, and the phase state of the hydrometeor profile. Using the observations from the Tropical Rainfall Measuring Mission (TRMM) spaceborne Ku-band (13.8 GHz) radar at low earth orbit of 350 km above earth, global attenuation maps are produced at the Ku-band frequency. A simple microphysical model for precipitation developed using hydrometeor size distributions and thermodynamic phase state is used to estimate attenuation and reflectivity observations at Ka-band (35 GHz) where numerous high-bandwidth satellite applications are being planned including the next-generation space-based radar for the Global Precipitation Mission (GPM). Differences in the microphysical structure in convective and stratiform precipitation are also incorporated in the model. The results show substantial attenuation variation in a 12-month period at both Ku- and Ka-bands over the various regions of the globe, including the contrast between land and ocean. The estimates of attenuation made at Ku- and Ka-band will be useful in the design and development of spaceborne systems. Index TermsAttenuation, precipitation, satellite communication, space-based radar, Tropical Rainfall Measuring Mission (TRMM).

Fig. 1. Schematic diagram of spaceborne sensor and precipitation.

I. INTRODUCTION T IS WELL KNOWN that precipitation can greatly affect the propagation of radio waves at frequencies above 3 GHz in various ways and, these effects must be taken into account in the design of radio frequency (RF) links as well as sensor systems such as radars [1]. Among these effects, the most serious is precipitation-induced attenuation. Even at a relatively low frequency such as 3 GHz (S-band), attenuation effects can be significant when the propagation path (or path length) passes through multiple precipitation cells of high intensity. Recently, higher frequency bands such as Ka-band are getting more attention for remote sensing applications and satellite communication because of the need for larger bandwidth and lower cost. However, as described above, the radio wave propagation is much more susceptible to precipitation attenuation at higher frequencies. Rain, clouds, and gaseous absorption by oxygen and water vapor can easily affect the signal and must be considered. In general, rain attenuation dominates this process and can produce attenuation of the order of 2030 dB (in the 2030-GHz range) [1], [2]. Estimates of attenuation and intrinsic reflectivity are some of the most important issues for spaceborne radar operations.

Data from the Tropical Rainfall Measuring Mission (TRMM) precipitation radar (PR) [3] are used to develop monthly maps of attenuation, in the earthspace propagation paths, over the tropical regions of the globe at Ku-band. The variability of the attenuation map is studied over a 12-month period in conjunction with the variability over land and ocean. One of the important contributions of the TRMM program is the global classification of precipitation into convective and stratiform types, that correspond to different vertical structure in the heating profiles and microphysical description. This paper develops models based on the convective/stratiform classification. A simple precipitation model is developed that consists of layers of hydrometeors as a function of altitude such as ice crystals, aggregates, graupel, and rain. Using the hydrometeor size distribution as well as the thermodynamic phase states as the descriptors, estimates of specific attenuation and reflectivity at Ka-band are obtained from Ku-band observations. The paper is organized as follows. Section II develops the theoretical characteristics of cumulative attenuation and radar reflectivity for a multitude of hydrometeor phase states, whereas Section III describes the TRMM PR dataset and the procedure to obtain the cumulative attenuation of the earthspace RF path. Section IV presents the results of the measured attenuation presented as monthly maps over the globe for a 12-month period, for the year 2000. Estimated attenuation maps for a 12-month period useful for both radar and communication applications are presented for Ka-band (35 GHz). Section V presents the estimation of total loss of signal in Ka-band as a conditional probability, under the condition that the radar signals are observed at Ku-band. The important results of the paper are summarized in Section VI. II. ATTENUATION AND REFLECTIVITY DUE TO PRECIPITATION A. Backscatter and Extinction Cross Sections The absorption and scattering of electromagnetic waves in precipitation are responsible for the attenuation. For frequen-

Manuscript received October 19, 2002; revised May 11, 2003. This work was supported by the National Aeronautics and Space Administration (NASA) Tropical Rainfall Measuring Mission (TRMM) Program. The authors are with the Colorado State University, Fort Collins, CO 80523-1373 USA (e-mail: chandra@engr.colostate.edu). Digital Object Identifier 10.1109/TGRS.2003.815973

0196-2892/03$17.00 2003 IEEE

CHANDRASEKAR et al.: GLOBAL MAPPING OF ATTENUATION AT Ku- AND Ka-BAND

2167

TABLE I PSD PARAMETERS AND TWO-FREQUENCY RELATIONS

(a)

(b)

Fig. 2. (a) Scatter plot of attenuation versus reflectivity for rain at Ku-band (13.8 GHz). (b) Scatter plot of attenuation versus reflectivity for rain at Ka-band (35 GHz).

(a) Fig. 3.

(b)

(a) Scatter plot of specific attenuation in rain for Ku- and Ka-band. (b) Scatter plot of reflectivity in rain for Ku- and Ka-band.

TABLE II PARAMETERS OF A = Z FOR RAIN, WET GRAUPEL, AND LOW-DENSITY AGGREGATES FOR KU- AND KA-BAND

PARAMETERS

OF

A(Ka)

TABLE III

= Ku) FOR RAIN, WET GRAUPEL, AND AGGREGATES

cA

of hydrometeors can be expressed in terms of the raindrop as [4] size distribution cies corresponding to Rayleigh scattering, the absorption cross section is proportional to the volume of hydrometeors and ). The specific atdominates the extinction cross section ( tenuation of a propagating wave through a medium composed dBkm where is the equivalent spherical diameter of raindrops. (1)

2168

IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 41, NO. 10, OCTOBER 2003

(a) Fig. 4.

(b)

(a) Observation scheme of TRMM (adapted from TRMM manual). (b) Schematic showing the variable clutter-free region around the globe.

For raindrops, the above equation can be parameterized up to 15 GHz as [4] (2) is dependent on wavelength. It is also where the coefficient a function of temperature. Fig. 1 shows a schematic diagram of a spaceborne sensor such as a radar or a communication satellite and the precipitation in earthspace path. The attenuation problem of communication satellite is simpler than that of a spaceborne radar. In the radar problem, the path-integrated attenuation must be known as a function of range. The propagation properties of electromagnetic waves in precipitation can be described in terms of a effective propagation constant where the deviation of from (the free-space propagation constant) is dependent on the composition of the is complex precipitation medium. It should be noted that written in terms of real and imaginary parts as (3) is responsible for the propagation phase shift where describes the attenuation. If the intrinsic radar whereas reflectivity due to precipitation at a range from the radar is , then the measured reflectivity is reduced due to attenuation and can be expressed as (4) and can be specified by the hydrometeor size Both is distribution and the thermodynamic phase states. In (4), in units per meter (m ), and is in meters. Converting to units typically used with radar, if path length is in kilometers, and
(b)

(a)

Fig. 5. (a) Storm over Melbourne, FL, August 13, 1998, 22:30. (b) Reflectivity versus altitude along one beam in Fig. 5(a).

CHANDRASEKAR et al.: GLOBAL MAPPING OF ATTENUATION AT Ku- AND Ka-BAND

2169

(a) Fig. 6. (a) Simple model for convective precipitation. (b) Simple model for stratiform precipitation.

(b)

(a)

(b)

Fig. 7. (a) Path-integrated attenuation along a beam in Fig. 5(a). (b) Measured reflectivity at Ka-band (35 GHz) versus altitude along the beam shown in Fig. 5(b).

specific attenuation is defined in decibels per kilometer, then it ; and (4) reduces to [4] can be shown that (5a) (5b) where denotes the discretized range bins. The intrinsic radar reflectivity is defined as [4] (6a) is the backscatter cross section (at the appropriate where ; polarization); is the wave length; and is the complex dielectric constant.

Attenuation is conventionally parameterized in terms of reflectivity of the form (6b) If (6b) is used to estimate attenuation from measured , then assuming continuity of precipitation reflectivity in range, an iterative solution to (5b) can be developed to build the intrinsic reflectivity profile, and this procedure is commonly known as Hitschfeld Bordan algorithm [5]. This method is fairly unstable, even with modest errors in the radar constants and parameterizations [6]. However, the attenuation correction technique can be bounded by comparing against a known target such as earths surface while observing from space. The attenuation correction procedure for TRMM radar is implemented using a hybrid of the HitschfeldBordan (HB) algorithm and the surface reference technique (SRT). The surface reference method observes

2170

IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 41, NO. 10, OCTOBER 2003

(a)

(b) Fig. 8. (a) Attenuation map at Ku-band for January 2000. (b) Attenuation map at Ka-band for January 2000.

the apparent decease in surface cross section caused by attenuation due to precipitation. This algorithm measures the vertical between the cloud top and the lowest reflectivity profile of height above the surface that is clutter free. The parameter is adjusted in the AZ relationship of (6b) in such a way that the path-integrated attenuation (PIA) estimate along the vertical profile will match the apparent reduction in . At low rain rates, due to attenuation may not be significant and the decrease in could fall below the natural fluctuations of the surface return. In this case, it is better to use only the HB algorithm [7]. Bolen and Chandrasekar [8] made an evaluation of the attenuation correction by comparing space-based and ground-based radar data in coincident precipitation and concluded that the procedure works fairly well and does not have any systematic bias. B. Microphysical Model of Precipitation The attenuation due to hydrometeors is determined by the extinction cross section, which is the sum of the scattering and absorption cross sections. In the Rayleigh scattering region, absorption is the dominant component of attenuation. However, at higher frequencies, both scattering and absorption need to be considered. In the following, simulations are used to study the attenuation through various hydrometeor particles such as rain, graupel, and wet aggregates. The hydrometeor size distribution plays an important role in determining the attenuation

characteristics of various hydrometeors in precipitation. In the Rayleigh scattering region, the attenuation is proportional to product of the volume and the imaginary part of the complex dielectric constant of the hydrometeor [4]. The dielectric constant changes with wavelength and temperature. Though these approximations are valid strictly for Rayleigh scattering, the general behavior with wavelength and temperature is maintained even at higher frequencies. The hydrometeor size distribution can be described as [9] (7) is the number of particles in the interval to where ; is the concentration; and is the probability density function describing the size distribution. Ulbrich [10] has shown that a gamma model for (7) can adequately describe many of the natural variabilities in the hydrometeor size distribution. In order to compare the size distribution of hydrometeors under widely varying precipitation rates, the concept of scaling drop size distribution has been used by several authors [11][13]. The can be expressed corresponding normalized (or scaled) as

mm

(8)

CHANDRASEKAR et al.: GLOBAL MAPPING OF ATTENUATION AT Ku- AND Ka-BAND

2171

(a)

(b) Fig. 9. (a) Attenuation map at Ku-band for March 2000. (b) Attenuation map at Ka-band for March 2000.

where is the normalized scaling constant [per millimeter per cubic meter (mm m )], and , , and are parameters of the gamma distribution related as (9) is the widely used volume weighted median diameter. and Median diameter is a physically meaningful parameter that . Thus, the indicates half the water content comes from three parameters of the hydrometeor size distribution namely , , and describe the natural variability of hydrometeor size distribution, and varying them over a wide range of naturally observed values [10] yields a physically realistic simulation of derived parameters such as radar reflectivity and attenuation. The variabilities of the parameters used in the simulation in this paper are shown in Table I. The shape parameter of the distribution is changed between 1 to 4 for rain, whereas it is kept as zero for all ice particles (implying exponential distribution). The range of parameters in Table I are chosen from a wide variety of published observations of hydrometeor size distributions [14]. Fig. 2(a) shows a scatter plot of attenuation versus reflectivity at Ku-band (13.8 GHz) for rain, whereas Fig. 2(b) shows a similar scatter plot at Ka-band (35 GHz). A cursory glance of Fig. 2(a) and (b) shows that the general variability of versus is of a power law form. The variability about the mean relation

in Fig. 2(a) shows that a fixed curve (or equation) will not be accurate at higher precipitation rates (or reflectivity), and that is why a finer definition is required, as done using the surface reference technique in TRMM PR. Fig. 3(a) shows a scatter plot of the specific attenuation through rain for Ku- and Ka-band. It should be noted here that these simulations are done at elevation angles of 90 . Details about the polarization-dependent properties of attenuation can be found in [4]. The results of Fig. 3(a) show that the specific attenuation in rain at Ka- and Ku-band frequencies are nearly linearly related. Therefore, the same relationship holds for cumulative attenuation also. The relation between the backscatter reflectivities at Ku- and Ka-band for a wide variety of raindrop size distributions are shown in Fig. 3(b). While the reflectivities are similar at lower reflectivity values, they are different at higher reflectivity levels due to non-Rayleigh scattering. A second-order polynomial fit is made to the reflectivities and is shown in Fig. 3(b). Similar computations are done for aggregates and graupel at 13.8 and 35 GHz using the model parameters listed in Table I, in order to develop specific attenuation relations to radar reflectivity as well as estimate attenuation and observed reflectivity at other frequencies. Table II shows the parameterization of the relation between specific attenuation and reflectivity of the for different types of hydrometeors, whereas form Table III shows the attenuation relation between Ku- and Ka-band frequencies for rain, wet graupel, and wet aggregates.

2172

IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 41, NO. 10, OCTOBER 2003

(a)

(b) Fig. 10. (a) Attenuation map at Ku-band for July 2000. (b) Attenuation map at Ka-band for July 2000.

The scaling relation for attenuation between Ku- and Ka-band for ice crystals is less accurate than the AZ relation; therefore the later is used in this paper. III. TRMM PRECIPITATION RADAR OBSERVATION AND APPLICATION TO ATTENUATION MEASUREMENTS In November 1997, the TRMM observatory was launched as a joint project between the National Aeronautics and Space Administration (NASA) and the National Space Development Agency of Japan (NASDA). One of the main objectives of TRMM is to measure rainfall over the tropics. TRMM hosts a variety of instruments, but we limit our discussion to the precipitation radar. The precipitation radar is a spaceborne weather radar operating in the Ku-band (13.8 GHz). The satellite was on a 350-km circular orbit (with a 35 inclination angle), but was boosted in August 2001 to just over 400 km. The data discussed in this paper is at the 350-km orbit. The precipitation radar scans electronically from left to right (looking in the direction of the flight) every 600 ms with a swath width of 215 km. Fig. 4(a) adapted from TRMM handbook shows a schematic of the TRMM PR operation [3]. For a given beam, the PR begins recording samples at a fixed distance from the satellite and records a certain number of samples along the ray. The starting distance and the number of samples are different for each ray. Assuming the satellite altitude is 350 km, the sampling begins about 23 km above mean sea level (MSL)

and extends down to the surface. The PR data used in this research has a range resolution of 250 m. Numerous details such as surface detection and geolocation are involved in the extraction and analysis of TRMM PR data, which are skipped here for brevity. The earths surface produces a strong echo compared to those of the precipitation scatterers specially close to ground. This problem is more pronounced at off-nadir angles in comparison to nadir angles. In order to ensure that surface clutter does not contaminate the data used in this paper, a surface clutter-free region (clutter-free certain according to TRMM PR documentation [3]) is determined and only data in clutter-free regions are considered for analysis. Fig. 4(b) shows the sketch of variable clutter-free altitude as a function of position around the globe. The difference between measured reflectivity and attenuation corrected reflectivity at a specified range from the radar yields the cumulative attenuation on the earthspace path. The range profile of attenuation in TRMM is obtained as a combination of the HB algorithm and surface reference technique as explained in Section II. A vertical profile of reflectivity is shown in Fig. 5(a), for data taken over Melbourne, FL on August 13, 1998. Range bin 80 is the mean sea level for this dataset. Fig. 5(b) shows the measured and the corrected reflectivity through one of the rays. In this dataset, the clutter-free height is estimated to be about 1.5-km altitude. Therefore, the measurements below this altitude are discarded; however, the data immediately above are linearly extrapolated down to the surface as shown in Fig. 5(b). The corresponding

CHANDRASEKAR et al.: GLOBAL MAPPING OF ATTENUATION AT Ku- AND Ka-BAND

2173

(a)

(b) Fig. 11. (a) Map of missed observation estimates at Ka-band (January 2000). (b) Map of missed observation estimates at Ka-band (July 2000).

estimate of cumulative attenuation is shown later in Fig. 7(a) (along with the estimate at Ka-band). The results of Ku-band estimates in Fig. 7(a) show that cumulative attenuation is about 5 dB for the specific case presented. In order to estimate the attenuation as well as the observed reflectivity pattern at Ka-band, the following microphysical model was developed. While this model may not provide an accurate description of the spacetime microphysical structure of precipitation, it is sufficient for studying the bulk attenuation properties. The main steps in the model are as follows. 1) Precipitation is classified into convective and stratiform precipitation [15]. 2) Region below the melting layer is rain. 3) Region at the melting layer is characterized based on convective or stratiform precipitation. a) Ice phase of stratiform precipitation is described by a 1-km layer of melting aggregates at the melting layer below a layer of dry snow [3], [15]. b) Ice phase of the convective layer is described by 1.5-km thick layer of melting graupel (at the melting layer), below a layer of ice crystals. Fig. 6(a) and (b) shows the schematic of the microphysical model for convective and stratiform precipitation. Though the above model is simplistic from a microphysical perspective, being constrained by the observed melting layer (as opposed to a fixed altitude), the convective/stratiform structure and

measured reflectivities makes the model fairly accurate for the estimation of attenuation and reflectivities. The parameters listed in Tables II and III can be used to describe the attenuation versus frequency relations as well as A-Z relations of the various portions of the precipitation regions in Fig. 6. Fig. 7(a) and (b) shows the estimated path-integrated attenuation as well as the observed reflectivity at Ka-band for the data shown in Fig. 5. It can be seen that the estimates of attenuation at 35 GHz are in the range of values reported in the literature [2]. In addition, the observed reflectivity at 35 GHz is severely attenuated due to the enhanced attenuation through rain and melting layer at 35 GHz. The procedure developed in this section is extended to observations around the globe using TRMM data, and the results are presented as monthly maps in the following sections. IV. MONTHLY MAPS OF ATTENUATION AT KU- AND KA-BAND TRMM PR observations are used to develop monthly maps of attenuation at Ku- and Ka-bands. The maps at Ku-band are direct measurements, whereas those at Ka-band are estimates based on the model developed in this paper. Fig. 8(a) shows the attenuation map at the surface for Ku-band for January 2000, whereas Fig. 8(b) shows the corresponding estimates at Ka-band. The color scales are adjusted to the full scale of each 0.5 ) area and therefore map. Each pixel in the plot is (0.5 of fairly high resolution. The overlay shows the standard

2174

IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 41, NO. 10, OCTOBER 2003

(a)

(b)

(c) Fig. 12. (a) Typhoon Jelawat near Japan, observed by TRMM on August 7, 2000 (altitude is 2 km). The grid is in latitude and longitude. (b) Measured reflectivity at the center of the typhoon Jelawat. (c) Map of missed observation estimates at Ka-band for typhoon Jelawat.

map of land/ocean boundaries. The plots are restricted to the latitudes of 35 coinciding with the coverage of TRMM PR. Some features are obvious from Fig. 8(a), namely the Southern Hemisphere experiences large attenuation in January compared to Northern Hemisphere, because of the seasonal nature of precipitation. In addition, the attenuation over land surface is on the average much larger than that of the ocean, except in the intertropical convergence zone. The results of Fig. 8 show

0.5 that the attenuation at the surface averaged over 0.5 area can be of the order of 30 dB at Ka-band. It should be 0.5 area, the Ka-band attenuation noted that within the 0.5 could peak higher than the average value. Fig. 9(a) and (b) shows similar maps for March, 2000, which is spring time in the Northern Hemisphere and fall in the Southern Hemisphere. The results of Fig. 9(a) and (b) shows an increase in attenuation in the lower latitudes over North America, while the Amazon

CHANDRASEKAR et al.: GLOBAL MAPPING OF ATTENUATION AT Ku- AND Ka-BAND

2175

and Central Africa remain active. Fig. 10(a) and (b) shows the attenuation maps of Ka- and Ku-band for the month of July (or summer time in the Northern Hemisphere). Those show the expected contrast with the attenuation maps of January 2000. While monthly maps of attenuation are available for all 12 months, all are not presented for brevity. The next important application of the analysis developed here is the feasibility of using 35-GHz radar for global precipitation measurement purposes, in a dual-frequency mode of operation. V. ESTIMATES OF RADAR REFLECTIVITY MAPS AT KA-BAND Among the various technologies considered for spaceborne observation of precipitation, dual-frequency observations at Kuand Ka-band provide great promise of being able to yield better rainfall estimates as well as derive vertical structure of the drop size distribution parameters [6]. Several airborne experiments have validated the potential of the dual-frequency technique. It is obvious from the attenuation maps shown in the previous section that the cumulative attenuation along a path can exceed many tens of dB at 35 GHz. The spaceborne weather radar design for TRMM has a noise floor such that minimum detectable reflectivity factor in precipitation is about 17 dBZ. Similar considerations can also be assumed for Ka-band. In order to utilize the dual-frequency radar observations at 14 and 35 GHz for precipitation estimation (as well as DSD retrieval) it is important the 35-GHz radar returns stay above the noise floor. For quantitative applications such as DSD retrieval, a reflectivity level of 3 dB above the noise floor (20 dBZ) can be assumed as this threshold. The results of Section II showed that the reflectivity at Ka-band could be different from Ku-band especially at inis tense precipitation. In addition, the observed reflectivity reduced due to attenuation as described by (4). Using a combination of parameterization for Ka-band reflectivity and attenuation, maps of observed reflectivity at Ka-band can be created. Cross validation of TRMM observed reflectivities with ground based radar indicate that only about 2% to 3% under estimate of the precipitation intensity due to the lower limit of 17 dBZ [16]. However, that is definitely not going to be the case for Ka-band observation. In order to provide a quantitative measure of missed precipitation observation monthly maps of the percentage of missed precipitation at Ka-band, under the condition that it is observed at Ku-band are constructed. Fig. 11 shows maps of the percentage of missed observations at Ka-band given that it is observed at Ku-band. Fig. 11(a) shows the maps of the percentage of missed observations for January, while Fig. 11(b) shows the corresponding maps for July. The results are shown for observation at 2-km altitude. The results of Fig. 11 quantitatively demonstrate the percentage of missed observations at Ka-band. These results for lower altitudes will be worse as evishown in Fig. 7(b). Observation denced by the profile of of large-scale tropical systems over the ocean is one of the important applications of spaceborne precipitation measurement systems. Fig. 12 shows the observation of Typhoon Jelawat near Japan as observed on August 7, 2000. Fig. 12(a) shows large areal map of reflectivity along the TRMM track at Ka-band, whereas Fig. 12(b) shows the enlarged area of reflectivity centered on the typhoon measured at Ku-band, all measurements at

2-km altitude. Fig. 12(c) shows the percentage of missed observations in Ka-band for a typhoon. Both the results of Figs. 11 and 12(c) indicate that using 2-km altitude as reference, only a small portion of the precipitation observations are missed as presented in monthly maps over the globe as well as observations over a large-scale tropical system such as a typhoon. VI. SUMMARY AND CONCLUSION The distribution of precipitation and its vertical structure dominates the electromagnetic propagation properties of earthspace radio frequency paths at frequencies higher that S-band. Cumulative path attenuation due to precipitation is the most significant effect. Using measurements from the TRMM PR, monthly maps of attenuation are developed over the globe. The variability of these attenuation maps is studied over a 12-month seasonal cycle for the year 2000. The attenuation maps showed the expected seasonal variability of attenuation between Northern and Southern Hemispheres. In addition, the observation shows the contrast between land and ocean. A simple microphysical model was developed to estimate the reflectivity and attenuation due to precipitation at Ka-band based on the observations at Ku-band. This model incorporated the variability in the microphysical structure between convective and stratiform precipitation that have been extensively studied in the TRMM program. The estimation of attenuation at Ka-band is useful for design of space systems at Ka-band. In addition, the estimation of observed reflectivity at Ka-band provides a basis to estimate the fraction when Ka-band measurements will be completely lost due to attenuation indicating that dual-frequency techniques cannot be used. Analysis over the 12-month seasonal cycle and a typhoon using TRMM observations indicate that, using 2-km altitude as reference, the Ka-band observations are lost only for a small fraction of the observations. ACKNOWLEDGMENT The lead author acknowledges the suport from the Goddard Visiting Fellow program. REFERENCES
[1] Proc. 23rd NASA Propagation Experimenters Meeting (NAPEX XXIII) and the Advanced Communications Technology Satellite (ACTS) Propagation Studies Workshop, Falls Church, VA, June 24, 1999, JPL Pub. 99-16. [2] J. D. Beaver and V. N. Bringi, The application of S-band polarimetric radar measurements to Ka band attenuation prediction, Proc. IEEE, vol. 85, pp. 893909, June 1997. [3] NASDA and NASA, TRMM Science Data and Information System, National Space Development Agency of Japan (NASDA) and National Aeronautics and Space Administration (NASA), NASA Goddard Space Flight Center, Vol. 3, Release 5.03, 2000. [4] V. N. Bringi and V. Chandrasekar, Polarimetric Doppler Weather Radar: Principles and Applications: Cambridge University Press, 2001. [5] W. Hitschfeld and J. Bordan, Errors inherent in the radar measurement of rainfall at attenuating wavelengths, J. Meteorol., vol. 11, pp. 5867, 1954. [6] R. Meneghini and T. Kozu, Spaceborne Weather Radar. Norwood, MA: Artech House, 1990. [7] T. Iguchi, T. Kozu, R. Meneghini, J. Awaka, and K. Okamoto, Rain profiling algorithm for the TRMM precipitation radar, J. Appl. Meteorol., vol. 39, pp. 20382052, 2000.

2176

IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 41, NO. 10, OCTOBER 2003

[8] S. Bolen and V. Chandrasekar, Quantitative cross-validation of space radar signals from ground-based polarimetric radar observations, Radio Sci., 2002, to be published. [9] V. Chandrasekar and V. N. Bringi, Simulation of radar reflectivity and surface measurements of rainfall, J. Atmos. Oceanic Technol., vol. 4, pp. 464478, 1987. [10] C. W. Ulbrich, Natural variations in the analytical form of the drop size distribution, J. Climate Appl. Meteorol., vol. 22, pp. 17641775, 1983. [11] R. S. Sekhon and R. C. Srivastava, Doppler radar observations of drop size distributions in a thunderstorm, J. Atmos. Sci., vol. 28, pp. 983994, 1971. [12] P. T. Willis, Functional fits to some observed drop size distributions and parameterization of rain, J. Atmos. Sci., vol. 41, pp. 16481661, 1984. [13] J. Testud, S. Oury, R. Black, P. Amayenc, and X. Dou, The concept of Normalized distribution to describe raindrop spectra: A tool for cloud physics and cloud remote sensing, J. Appl. Meteorol., vol. 40, pp. 11181140, 2001. [14] H. R. Pruppacher and J. D. Klett, Microphysics of Clouds and Precipitation, 2nd ed. Dordrecht, The Netherlands: Kluwer, 1997. [15] J. Awaka, T. Iguchi, and K. Okamoto, Early results on rain type classification by the Tropical Rainfall Measuring Mission (TRMM) precipitation radar, in Proc. 8th URSI Commission F Open Symp., Aveiro, Portugal, 1998, pp. 143146.

[16] S. M. Bolen and V. Chandrasekar, Quantitative cross validation of space-based and ground-based radar observations, J. Appl. Meteorol., vol. 39, pp. 20712079, 2000.

V. Chandrasekar, photograph and biography not available at the time of publication.

Hiroki Fukatsu, photograph and biography not available at the time of publication.

K. Mubarak, photograph and biography not available at the time of publication.