AN IMPROVED MODEL FOR THE MICROWAVE

BRIGHTNESS TEMPERATURE SEEN FROM SPACE

OVER CALM OCEAN
by

Your name here

A thesis submitted in partial fulfillment of the requirements for the degree of

MASTER OF ENGINNERING
in
ELECTRICAL ENGINEERING

UNIVERSITY OF PUERTO RICO

UNIVERSITY APEC
MAYAGÜEZ CAMPUS
2005

Approved by:

________________________________ __________________
Sandra L. Cruz-Pol, PhD Date
Member, Graduate Committee

________________________________ __________________
Sandra L. Cruz-Pol, PhD Date
Member, Graduate Committee

________________________________ __________________
Sandra L. Cruz-Pol, PhD Date
President, Graduate Committee

________________________________ __________________
Sandra L. Cruz-Pol, PhD Date
Representative of Graduate Studies

________________________________ __________________
Sandra L. Cruz-Pol, PhD Date
Chairperson of the Department
This thesis was auto-formatted by Dr. Sandra Cruz-Pol, please use to ease your life.

RESUMEN

This thesis was formatted so that you only type on top of it your material and it should redo
the Table of contents automatically by the command on the Menu: Insert>Reference>Index
and Tables. IMPORTANT: when pasting material to this file be sure to use
Edit>Paste_Special> Unformatted_Text, so that the format given by this file is not
changed. When pasting figures, be sure to use Paste_special>Picture(JPEG) to make the
size of your thesis file as much as 10 times smaller than using just Paste, which usually
utilized the BMP format for figure, making your file huge and not portable. The figures
should be inserted using Insert>References>Caption>(Figure), and typing the caption of
the figure. In some versions of Word® you can skip the Reference part. The tables should
be inserted using Insert>References>Caption>(Table). Word® should automatically
increase the figure number and add the chapter number to it. The equations are also set so
that you do Insert>References>Caption>(Equation) and the equation number increases
automatically. You can delete the word Equation later, if you prefer to display only the
equation number. Good luck! Hope this saves you a lot of work and time. SCP

This work presents models that predict extinction rates due to atmospheric gases for 35 GHz

and 95 GHz radars as a function of elevation angle. The minimum detectable radar

reflectivity (dBZemin) is computed for both wavelengths using radiosonde and microwave

radiometer measurements. In general, sensitivity decreases with elevation angle mostly

because water vapor and their corresponding highest extinction rates propagate through the

lower portion of the atmosphere.

ii
This thesis was auto-formatted by Dr. Sandra Cruz-Pol, please use to ease your life.

RESUMEN

Este trabajo presenta un modelo que predice la razón de extinción para señales de 33 y 95

GHz debido a los gases atmosféricos en función del ángulo de elevación. Se computo la

mínima reflectividad detectable por el radar (dBZemin) para ambas frecuencias usando medidas

de radiosonda y radiómetro de microondas. En general la sensitividad decrece con el ángulo

de elevación debido principalmente a que el vapor de agua y su correspondiente alta

extinción suceden en la porción baja de la atmósfera.

iii
This thesis was auto-formatted by Dr. Sandra Cruz-Pol, please use to ease your life.

To my family . . .

iv
This thesis was auto-formatted by Dr. Sandra Cruz-Pol, please use to ease your life.

ACKNOWLEDGEMENTS

During the development of my graduate studies in the University of Puerto Rico several

persons and institutions collaborated directly and indirectly with my research. Without their

support it would be impossible for me to finish my work. That is why I wish to dedicate this

section to recognize their support.

I want to start expressing a sincere acknowledgement to my advisor, Dr. Sandra Cruz-Pol

because she gave me the opportunity to research under her guidance and supervision. I

received motivation; encouragement and support form her during all my studies. With her, I

have learned writing papers for conferences and sharing my ideas to the public. I also want

to thank the example, motivation, inspiration and support I received from Dr. José Colom.

From these two persons, I am completely grateful. Special thanks I owe Dr. Stephen M.

Sekelsky for the opportunity of researching under his supervision, his support, guidance, and

transmitted knowledge for the completion of my work.

The Grant from NSF EIA 99-77071 provided the funding and the resources for the

development of this research. At last, but the most important I would like to thank my

family, for their unconditional support, inspiration and love.

v
This thesis was auto-formatted by Dr. Sandra Cruz-Pol, please use to ease your life.

Table of Contents
RESUMEN...............................................................................................................................................................II

RESUMEN..............................................................................................................................................................III

ACKNOWLEDGEMENTS.........................................................................................................................................V

TABLE OF CONTENTS....................................................................................................................................VI

TABLE LIST......................................................................................................................................................VII

FIGURE LIST..................................................................................................................................................VIII

1 INTRODUCTION......................................................................................................................................2
1.1 MOTIVATION..............................................................................................................................................2
1.2 GFJDHGJFHGJ.............................................................................................................................................2
1.3 GYYYYYYYYYY.........................................................................................................................................2
1.4 LITERATURE REVIEW.................................................................................................................................3
1.5 SUMMARY OF FOLLOWING CHAPTERS......................................................................................................5
2 THEORETICAL BACKGROUND..........................................................................................................6
2.1 RADIATIVE TRANSFER EQUATIONS...........................................................................................................6
2.1.1 Equations relating humidity profiles and microwave radiometer data to attenuation...................6
2.1.2 Water vapor profile and zenith attenuation statistics at 33 and 95 GHz........................................9
2.2 SCAN EQUATIONS...................................................................................................................................12
2.3 RADAR SYSTEM CHARACTERISTIC AND MCTEX EXPERIMENT LAYOUT............................................17
2.3.1 Maritime Continent Thunderstorm Experiment (MCTEX)...........................................................17
2.3.2 Radar Hardware of Cloud Profiling Radar System (CPRS).........................................................17
3 MICROWAVE ATMOSPHERIC ABSORPTION MODEL.............................................................19
3.1 ATMOSPHERIC ABSORPTION..................................................................................................................19
3.2 NEW MODEL RETRIEVED PARAMETERS..................................................................................................21
3.2.1 Bullet and Bullet Rosettes Toolbox for DDSCAT Program..........................................................22
3.3 BACKSCATTERING WITH DDSCAT.......................................................................................................23
4 CONCLUSIONS AND FUTURE WORK............................................................................................26

APPENDIX A. IDL CODES FOR DBZEMIN...................................................................................................29

APPENDIX B PROGRAMS FOR BULLET AND DWR................................................................................31

Appendix B1 IDL Program For Refraction Index...........................................................................................31

vi
This thesis was auto-formatted by Dr. Sandra Cruz-Pol, please use to ease your life.

vii
This thesis was auto-formatted by Dr. Sandra Cruz-Pol, please use to ease your life.

1 Table List

Tables Page

TABLE 2.1CPRS Parameters..................................................................................................10

TABLE 2.2 de Capacidad Instalada........................................................................................11
TABLE 2.2CPRS Operational Models....................................................................................11
TABLE 2.3 Mean values of the regions for CPRS data collected and dBZemin simulated.......16

viii
This thesis was auto-formatted by Dr. Sandra Cruz-Pol, please use to ease your life.

Figure List

Figures Page

Figure 2.1 dfgfdsgfdsfdsfjdshfj rocca........................................................................................7

Figure 2.2 dfgfdsgfdsfdsfjdshfj rocca........................................................................................7
Figure 2.3 sol.............................................................................................................................7
Figure 2.2 ghfjghfjghjdfghjdfghjfghjf.......................................................................................9
Figure 2.3 ghfjghfjghjdfghjdfghjfghjf.......................................................................................9
Figure 2.3 Profile of extinction rates (--33 GHz and —95 GHz)Profile of extinction rates (--
33 GHz and —95 GHz)...................................................................................................10
Figure 2.4 Minimum detectable signal for a single zenith pulse at different modes of radar
pulse width.(a) Mode 1:  = 200ns, (b) Mode 2:  = 500ns, (c) Mode 3:  = 1,000ns....11
Figure 2.5 Flowchart for the IDL routine used for calculating the dBZemin..........................13
Figure 2.6 Mnimum detectable dBZe in mode 1 (= 200 ns), (a) 33 GHz, (b) 95 GHz.......14
Figure 2.7 Minimum detectable dBZe in mode in mode 2 (= 500 ns), (a) 33 GHz, (b) 95
GHz..................................................................................................................................14
Figure 2.8 The plot on (a) depicts the radar reflectivity measured at 95GHz with CPRS and
plot on data at same time than CPRS data was collected at 95GHz................................15
Figure 2.9 Hill ratio comparison between various atmospheric models showing agreement of
the chosen water vapor absorption line shape with the radiometer data. (See text for
explanation of models' acronyms)...................................................................................16
Figure 3.1 Bullet and Bullet Rosettes with different angles of junction.................................20
Figure 3.2 Wind speed model relating 0 to wind speed for the MCW algorithm as calibrated
for Topex altimeter..........................................................................................................22
Figure 3.3 Methodology used to create a bullet formed by an array of N dipoles separated by,
(a) General process, (b) bullet 3D-view, and (c) Bullet rosette with 3 bullets...............23
Figure 3.4 Backscattering (10 logb) of different indexes of refraction, (a) Backscattering in
dB to 33GHz with 652 dipoles array, (b) Backscattering in dB to 95GHz ....................24
Figure 3.5 Variation of the number of raob profiles used depending on the limits in space and
time separation imposed on the data................................................................................25

ix
 2 INTRODUCTION
Knowledge of the state of the ocean plays a vital role in weather and ocean wave forecasting

models [Wilheit, 1979a] as well as in ocean-circulation models [Dobson et al., 1987]. One

approach to measuring the state of the ocean is by remote sensing of the ocean’s surface

emission. Microwave radiometers on satellites can completely cover the earth’s oceans.

Satellite radiometry offers numerous advantages over ship and buoy data. Some of these

advantages include the vast coverage of global seas, including locations where radiosonde or

buoys cannot be afforded, relatively low power consumption, no maintenance and continuous

operation under a wide range of weather conditions.

Measurements of the microwave brightness seen from the sea are used in the retrieval of

physical parameters such as wind speed, cloud liquid water and path delay. A suitable model

for these measurements includes contributions from atmospheric emission, mainly water

vapor and oxygen, and from ocean emission.

2.1 Motivation
2.2 Gfjdhgjfhgj

2.3 gyyyyyyyyyy

The need to improve the calibration of existing models for atmospheric and ocean emission is

motivated by several current and upcoming satellite remote sensing missions. In the case of

2
TMR, an improved atmospheric model would enhance the inversion algorithm used to

retrieve path delay information. Another case is the JASON satellite, a joint NASA/CNES

radiometer and altimeter scheduled to be launched in 2000 [JPL, 1998]. For JASON,

absolute calibration is performed by occasionally looking at calm water. This type of

calibration reduces the cost in hardware, complexity, size and power. However, the quality

of the calibration depends strongly on the accuracy of a model for the calm water emission.

In contrast, for the TMR an absolute calibration is performed using hot and cold references

carried by the satellite [Ruf et al., 1995].

In this document, a section is devoted to each of these models. In Part I, the development of

an improved microwave atmospheric absorption model is presented. Part II is dedicated to

ocean microwave emission. In both cases, a model is developed and iteratively adjusted to

fit a carefully calibrated set of measurements.

2.4 Literature Review

Seasat was the first satellite designed for remote sensing of the Earth's oceans. It was

launched in 1978 by the National Aeronautic and Space Administration (NASA). The

mission was designed to demonstrate the feasibility of global satellite monitoring of

oceanographic phenomena and to help determine the requirements for an operational ocean

remote sensing satellite system. It included the Scanning Multichannel Microwave

Radiometer (SMMR) which measured vertical and horizontal linearly polarized brightness

temperatures at 6.6, 10.7, 18, 21 and 37 GHz. The SMMR was used to retrieve surface wind

speed, ocean surface temperature, atmospheric water vapor content, rain rate, and ice

3
coverage. Unfortunately, the mission only lasted approximately 100 days due to a failure of

the vehicle's electric power system [Njoku et al.,1980].

, cloud water content, and ocean surface wind speeds [Hollinger et al., 1990].

In 1991 the European Space Agency launched The ERS-1 satellite. The primary mission of

ERS-1 was to perform remote sensing of the Earth's oceans, ice caps, and coastal regions by

providing global measurements of wind speed and direction, wave height, surface

temperatures, surface altitude, cloud cover, and atmospheric water vapor levels. The mission

included a nadir viewing radiometer operating at 23.8 and 36.5 GHz and co-aligned with the

altimeter to provide range corrections with 2 cm accuracy [Günther et al., 1993].

In 1998 the US Navy launched the GEOSAT Follow On (GFO), designed to provide real-

time ocean topography data. It includes a radar altimeter with 3.5 cm height measurement

precision. In addition, a dual frequency (22 and 37 GHz) water vapor radiometer is included

to provide path delay correction with an accuracy of 1.9 cm [Ruf et al., 1996].

Sekelsky et al. [1998, 1999] used simulations from the ice crystals backscattering at various

millimeter wavelengths using the DDA models on the version 5ª of DDSCAT, using a more

realistic density model where ice density is not constant, as in previous studies, but decreases

with the particles’ diameter. They calculated the dual-wavelength ratio (DWR). Their

findings also agree with previous studies where the shape and orientation are the principal

causes of error on the DWR estimates and other products.

Uncertainties in the improved model for atmospheric emission are significantly improved

over previous published models. The line-strength and width parameters' uncertainties are

4
reduced to 1% and 1.6%, respectively. The overall uncertainty in the new absorption model

is conservatively estimated to be 3% in the vicinity of 22GHz and approaching 8% at 32

GHz. The RMS difference between modeled and measured thermal emission by the

atmosphere, in terms of the brightness temperature, is reduced by 23%, from 1.36 K to 1.05

K, compared to one of the most currently used atmospheric models.

The modified ocean dielectric models exhibit significant improvements in the estimate of

TB. Of the two, the modified Ellison et al.[1977] model exhibits superior overall

performance, including the lowest bias at both frequencies, which is a very important

attribute indicative of the accuracy of the model. Its frequency dependence was decreased to

0.30K, which will allow for more reliable extrapolation to higher frequencies. In addition,

this modified model has the lowest dependence on sea surface temperature and the lowest

RMS difference for both 18GHz and 37GHz. Consequently, this is the model that we

recommend for future remote sensing applications involving microwave emissions from the

ocean emissivity of the ocean. The average error in the modified emissivity model, over the

range 18-40 GHz, is found to be 0.37%, which in terms of brightness temperatures, translates

into a model error of approximately 1K.

2.5 Summary of Following Chapters

We first develop the necessary background theory in Chapter 1. Chapter 2 deals with the

model theory, experiments and data analysis related to the atmospheric absorption model.

The third chapter presents the model theory, data, statement of the problem, and analysis for

the ocean emission model. Conclusions are presented in Chapter 4.

5
6
 3 THEORETICAL BACKGROUND

3.1 Radiative Transfer Equations

3.1.1 Equations relating humidity profiles and microwave radiometer data to

attenuation

The atmosphere receives most of its energy by means of solar electromagnetic radiation.

Some of this energy is absorbed by the atmosphere and some reaches the surface of the Earth

where it can also be absorbed or it can be reflected. Energy absorption implies a rise in

thermal energy and, therefore, temperature of the object. Any object with a temperature

above absolute zero emits electromagnetic radiation. Electromagnetic emission implies a

decrease in the object’s temperature. These processes, i.e. absorption and emission,

altogether help create a balance between the energy absorbed by the Earth and its atmosphere

and the energy emitted by them. The study of these energy transformation processes is called

radiative transfer.

The Planck function for spectral brightness describes the radiation spectrum of a blackbody

at thermal equilibrium. It is given by

2hf 3  1 
B f (T )  2  hf / kT  2.1
c e 1

Using the Rosenkranz’s model for gaseous attenuation due to oxygen, KO2(I), and a modified

Liebe’s model for gaseous attenuation due to water vapor, Kwv(l), for every layer (see Fig.

7
2.1) and for each radar frequency, 33 GHz and 95 GHz [Cruz-Pol, 1998; Keihm, et al. 2002]

total gaseous attenuation were calculated. The equation for Kwv(l) is [Cruz-Pol, 1998]. It is

given by shape and continuum terms.

Figure 2.1 dfgfdsgfdsfdsfjdshfj rocca

Figure 2.2 dfgfdsgfdsfdsfjdshfj rocca

Figure 2.3 sol

In this equations we delete the word Equation automatically inserted by Word® and we

formatted the text to the Right. You can leave the word Equation if you like. All the body

text is formatted as “justified” so that the margins are even..

TL  0.0109C L Pwv 3.5 e  2.1431   2.2

8
   1 1 
TS     2.3
f z   f z  f  2
 f z  f    
2 2


TC  C C 1.13  10 8 Pwv Pdry 3  3.57  10 7 Pwv 2 10.5  2.4

The absorption model for the water vapor resonance line is accomplished by the addition of

three parameters, given by CL = 1.064, CW = 1.066, and CC = 1.234. These are the parameters

for scaling the line strength, the line width and the continuum, respectively. Here f is the

radar frequency in GHz, fz is the water vapor resonant frequency, 22.235 GHz,  is the

inverse temperature, Pwv is the water vapor partial pressure, and Pdry is the difference between

total pressure, P, and the water vapor pressure, Pwv. Their respective equations are:

300
  2.5
t
sh
Pwv  2.6
0.7223
Pdry  P  Pwv 2.7

where sh is the specific humidity, t is the air temperature in Kelvin. The width parameter, ,

is defined as:


  CW 0.002784 Pdry 0.6  4.8 Pwv 1.1  2.8

The oxygen absorption model is defined as:

2
Pdry c 3 33
 f 
K O2 
  S  T    Ln  f
 fn 
 2.9
n odd 1

where c=0.5034 x 1012, S(T) is the line strength [Rosenkranz, 1993]

9
 S  T   S '  T0  2 e 0.0068952 n  n 1 1 2.10

3.1.2 Water vapor profile and zenith attenuation statistics at 33 and 95 GHz

Maritime Continent Island Thunderstorm Experiment was held during the Australian summer

monsoon. Thunderstorms develop in an environment with low shear and high moisture. The

data obtained by the radiosonde were corroborated with radiometer data. Collecting the

radiosonde measurements every day during the experiment, gaseous attenuation, specific

humidity and cumulative attenuation profiles were calculated for the complete experiment.

The average profile is shown in Figure 2.2.

Figure 2.4 ghfjghfjghjdfghjdfghjfghjf

Figure 2.5 ghfjghfjghjdfghjdfghjfghjf

Gaseous attenuation mean for 33 GHz is 0.11 dB/Km and 0.74 dB/Km for 95 GHz (see Fig.

2.3).

10
Figure 2.6 Profile of extinction rates (--33 GHz and —95 GHz)Profile of extinction rates
(--33 GHz and —95 GHz)

Equation (2.10) contains all the quantities needed to compute the response of a satellite-based

microwave radiometer to changes in atmospheric and surface variables.

The 33 GHz signal has more peak power than the 95 GHz signal (see Table 2.1) to

compensate for its smaller gain (wide bandwidth).

TABLE 2.1CPRS Parameters

W Ka band
band
Frequency (GHz) 95 33
Peak power (kW) 1.5 120
Average power (W) 15 120
Pulse width (ns) 500 200
Gain 105.8 104.83
Range gate spacing (m) 75 30
Pulse repetition freq. (kHz) 10 5
Noise figure (dB) 13 11
Bandwidth (MHz) 2 5
Beam width (deg) 0.18 0.50

11
 TABLE 2.2 de Capacidad Instalada

Thus, the 95 GHz signal has a comparable performance and has similar values of minimum

detectable signal to the 33 GHz signal, obtaining similar resolution and noise immunity for

both signals for a single pulse in zenith angle. This is shown in Figure 2.5. The other

modes’ parameters are shown in the Table 2.2

(a) (b) (c)

Figure 2.7 Minimum detectable signal for a single zenith pulse at different modes of
radar pulse width.(a) Mode 1:  = 200ns, (b) Mode 2:  = 500ns, (c) Mode 3:  = 1,000ns.

TABLE 2.3CPRS Operational Models

Mode 1 Mode 2 Mode 3

Pulse width (ns) 200 500 1,000
W Band. Pulse Repetition 10,000 10,000 10,000
Frequency (kHz)
Ka Band. Pulse Repetition 2,500 1,000 500
Frequency (kHz)
Bandwidth (MHz) 5 2 1

But when the radar scans and many pulses are sent, the radar performance does not behave in

the same way as when as sending a single pulse in zenith angle. So we need to analyze the

performance of scanning radar

12
3.2 Scan Equations

The one-way path loss, Ag, depends on the frequency being used. For frequencies where the

path loss degrades the signal strongly, higher power was used to minimize this effect.

After obtaining the atmospheric attenuation for every layer (see Fig. 2.1), we found the

cumulative gaseous attenuation. This one is calculated for a fixed angle and for every range

gate in which the radar operates. A matrix of radius times angles was used to save the

projected attenuation. Then the cumulative attenuation for specific angle and radius was

computed as:

Ta  TUP  s Ts e  ( 0, H ) sec  (1 s )(TDN  TC e  ( 0, ) sec )e  ( 0, H ) sec 2.11

Finally with the cumulative attenuation for every radius at a specific angle, the total path

loss, l, can be calculated. To implement all this procedure we used IDL program. IDL is a

language capable to process great amount of data, and a flow diagram in Figure 2.6 shows

the algorithm implemented in this work.

13
Figure 2.8 Flowchart for the IDL routine used for calculating the dBZemin

Graphs from calculations of the dBZemin when the radar operates in modes 1, 2, and 3, for

every radio and each angle at 33 and 95 GHz are plotted in Figures 2.7, 2.8, and 2.9. The

delta between two lines of the contour is 2 dB. The lightest bar colors represent larger

minimum reflectivity values that can be detected by the radar, i.e. less signal can be detected

in those areas.

14
 (a) (b)
Figure 2.9 Mnimum detectable dBZe in mode 1 (= 200 ns), (a) 33 GHz, (b) 95 GHz

(a) (b)
Figure 2.10 Minimum detectable dBZe in mode in mode 2 (= 500 ns), (a) 33 GHz, (b)
95 GHz

15
 (a)

Figure 2.11 The plot on (a) depicts the radar reflectivity measured at 95GHz with
CPRS and plot on data at same time than CPRS data was collected at 95GHz.

The radar begins to detect the cloud from a radius of 13 km and from an angle between 8 and

76 degrees. To the W band, the cloud looks much smaller than the one shown by the Ka

band. These data validate the simulation and confirm the effect of the attenuation of the W

band in angles smaller than 50 degrees (see Fig. 2.11a). Figures 2.10 and 2.11 show three

regions, these are the dBZemin that represent the CPRS data. We can see here that the radar

received a greater reflectivity than the minimum estimated reflectivity. We can see that this is

also true for the 95 GHz signal.

These results strongly suggest that VVW is the preferred choice for vapor absorption line

shape at 22 GHz. Note that the same finding was obtained by Hill [1986] when the ratio test

was applied to the original Becker and Autler [1946] laboratory data.

16
Figure 2.12 Hill ratio comparison between various atmospheric models showing
agreement of the chosen water vapor absorption line shape with the radiometer data.
(See text for explanation of models' acronyms).

The other regions behave in the same way. All the reflectivity mean values are within the

limits of the mean dBZemin simulated for both, the 33 GHz as for the 95 GHz. The other

mean values are listed in Table 2.3.

TABLE 2.4 Mean values of the regions for CPRS data collected and dBZemin simulated

Region 1 Region 2 Region 3

Mean dBZemin33GHz (dB) -27.9162 -28.8563 -29.1437
Mean Reflectivity at 33GHz (dB) 3.718391 7.22807 -2.63717

Mean dBZemin95GHz (dB) -12.9728 -17.7505 -23.7513

Mean Reflectivity at 95GHz (dB) -8.31193 -7.19231 -3.61205

17
3.3Radar System Characteristic and MCTEX Experiment
Layout

3.3.1 Maritime Continent Thunderstorm Experiment (MCTEX)

The MCTEX experiment was performed in the North Coast of Australia, and in the Bathurst

and Melville Islands. The principal objective of this experiment was to better understand the

physical processes, such as humidity balance over tropical islands on a maritime continent.

For this reason, the experiment was held between November 13th and December 10th, 1995;

season on which the transition phases occurs between the dry and wet seasons. The data of

this experiment were collected with different sensors. One set was collected by means of the

Cloud Profiling Radar System (CPRS). This one collected data on the Ka frequency band

(33.12 GHz) and W frequency band (94.92 GHz). Data from the W frequency band, 95

GHz, also was collected by the Airborne Cloud Radar. The NOAA radar collected data on

the S frequency band, at 2.8 GHz.

3.3.2 Radar Hardware of Cloud Profiling Radar System (CPRS)

The CPRS is a dual-frequency polarimetric Doppler radar system that works with two sub-

systems at 33 and 95 GHz. This was fully developed by the University of Massachusetts’

Microwave Remote Sensing Laboratory (MIRSL).

18
Table 2.1 shows the CPRS parameter. The CPRS has a programmable structure that allows

working in different modes of scanning. It has a high-speed VXI-bus-based data acquisition

and digital signal processing (DSP) system. A radome protects the system from atmospheric

effects. Both the 33 and 95 GHz sub-systems simultaneously transmit and receive by means

of a single aperture and not producing pointing errors between both frequencies. Table 2.1

shows other typical characteristics of the CPRS operation. The CPRS works in three

different operational modes, changing the pulse width and by consequence the pulse

repetition frequency and the bandwidth change. These values are shown in 2. 2. The CPRS

measures can obtain the reflectivity (Ze), mean fall velocity (u) linear depolarization ratio

(LDR), velocity spectral width (v), and the full Doppler spectrum (S(v)) [Firda, 1997;

Lohmeire, et al. 1997].

19
4 Microwave Atmospheric Absorption Model

An improved model for the absorption of the atmosphere near the 22 GHz water vapor line is

presented. The Van-Vleck-Weisskopf line shape is used with a simple parameterized version

of the model from Liebe for the water vapor absorption spectra and a scaling of the model

from Rosenkranz for the 20-32 GHz oxygen absorption. Radiometric brightness temperature

measurements from two sites of contrasting climatological properties  San Diego, CA and

West Palm Beach, FL  were used as ground truth for comparison with in situ radiosonde

derived brightness temperatures. The retrieval of the new model’s four parameters, related

to water vapor line strength, line width and continuum absorption, and far-wing oxygen

absorption, was performed using the Newton-Raphson inversion method.

4.1 Atmospheric Absorption

Various shapes of the bullet rosettes are observed (see Fig. 3.1). The angles among the

bullets within the rosette are random between 70º and 90º. Each bullet has a longitude

relation [Heymsfield, 1972], L (mm), versus wide, w (mm), (twice times the apothem) for

temperatures between –18º and –20 ºC given by

20
  TB1 TB1 TB1 TB1 
 C CW CC C X 
 L 
 TB 2 TB 2 TB 2 TB 2 
 C L CW CC C X 
J   TB 3 TB 3 TB 3 TB 3  3.12
 
 C L CW CC C X 
    
 T TBn TBn TBn 
 Bn 
 C L CW CC C X 

and the Gross Line shape is given by [Gross, 1955]

 1 for z / L'  0 (neutral conditions)


 u ( z / L' )   1  7 z / L' for z / L'  0(stable conditions) 3.13
(1  18z / L'  3 ).25 for z / L'  0(unstable conditions)
 u

Although DDA can describe any geometry, it is limited by a minimum distance d that should

exist between dipoles. This distance should be inversely proportional to any structural

longitude on the target and to the wavelength. Previous studies [Draine and Flatau, 1994]

sum up the two criteria in equation 3.6.

21
Figure 3.13 Bullet and Bullet Rosettes with different angles of junction

In this way the equations were determined for the bulk density, ρ, of the bullet, considering

the solid ice density as 0.9 g cm-3 and using the volume of ice in individual crystals

[Heymsfield, 1972]

As the Wiener’s theorem states [Oguchi, 1983], the complex index of refraction, m, depends

of the bulk density when dealing with dry ice particles:

Ln is proportional to the shape of the lines

 Y ( f  f )  Y ( f  f )
. Ln   n n n
 n n n 
Equation 3.14
 n2   f  f n  2  n2   f  f n  2 
 
The pressure-broadened line half-width is,


 n  0.001w Pdry  .8  11
. PH2 O  Equation 3.15
The O2 resonant lines are very close to each other and troposphere pressures are high enough

( > 100 mbars) to cause the lines to broaden and overlap. This is called collisional

broadening and is taken into account through the interference parameter.

4.2 New Model Retrieved Parameters

The final retrieved parameters, CL, CW, CC and CX, are shown in Table 2.1. As the table

indicates, the nominal parameters used in the L87R93 model are 3 to 7 percent lower.

Figures 2.7a-c depict plots of the brightness temperature for three climatological conditions.

Each graph has a plot corresponding to the L87R93 and new models. Also shown are the

radiometer measured brightnesses. The new estimated parameters show better agreement
22
with the WVR data. L87R93 model as the reference (therefore, by definition the L87R93

model is . In these figures we have included the L93 model which, as explained above, is

similar to L87R93 except that it has a higher water vapor line

Although DDA can describe any geometry, it is limited by a minimum distance d that should

exist between dipoles. This distance should be inversely proportional to any structural

longitude on the target and to the wavelength. Previous studies [Draine and Flatau, 1994]

sum up the two criteria in equation 3.6.

Figure 3.14 Wind speed model relating 0 to wind speed for the MCW algorithm as
calibrated for Topex altimeter.

4.2.1 Bullet and Bullet Rosettes Toolbox for DDSCAT Program

We developed two toolboxes for DDSCAT where we implemented the most common shapes

of the cirrus ice crystals, i.e. the bullet and bullet rosettes. Using a single DDSCAT

environment by means of the ddscat.par file [Draine and Flatau, 2000], we specified which

23
one of the geometries we wanted to use and parameters such as size, dielectric constant of the

material, and in general all the parameters related to the target to be analyzed.

(a)

(b) (c)

Figure 3.15 Methodology used to create a bullet formed by an array of N dipoles separated
by, (a) General process, (b) bullet 3D-view, and (c) Bullet rosette with 3 bullets.

4.3 Backscattering with DDSCAT

Once the bullet toolbox was created in DDSCAT, we proceeded to use it to simulate the

crystal’s backscattering at 33 and 95 GHz. Figure 3.4 shows the backscattering for one bullet
24
crystal of different sizes using several models for index of refraction and crystal density. The

figure shows the sensitivity of the backscattering to the index of refraction, showing the

necessity of considering the index of refraction for each size and density of the ice crystal,

and not assuming a constant density for all the bullets sizes.

(a) (b)

Figure 3.16 Backscattering (10 logb) of different indexes of refraction, (a) Backscattering in dB to
33GHz with 652 dipoles array, (b) Backscattering in dB to 95GHz .

It can also be seen that the backscattering obtained when varying the index of refraction

according to the particle size is not significantly different to the results obtained when using

constant indexes of refraction for different particle sizes.

Given that one of the objectives is to analyze the DWR, we designed an interface between

DDSCAT and IDL program. We developed a routine that iteratively collects data from IDL

such as the index of refraction, m, which is computed according to the particles size and the

index of refraction of the solid ice, ni, and saving m in DDSCAT to compute the

25
backscattering and again this value is saved in IDL to obtain the DWR. The DWR is defined

as [Sekelsky, et al. 1999]

  D  
     3.67     
 l 4 K I  h   b  D, l ,   D  2   

2 D 0 
e dD 
 
DWR  10 log 0 
 D 
3.16
     3 .67    

  4 K    2   D,  ,   D  2    e  

D0 
 h I l b h dD 
 0 
where l and h are the values of the smaller wavelength and greater respectively, KI is an

dimensionless quantity that depends on the index of refraction and on the density. For ice we

used 0.176 for both frequencies [Sekeslky, et al. 1999].

Figure 3.17 Variation of the number of raob profiles used depending on the limits in space and time
separation imposed on the data

26
 5 CONCLUSIONS AND FUTURE WORK

Recent work to determine the sea water dielectric coefficient was based on laboratory

measurements of sea water samples from different parts of the ocean. Although these

measurements should render good understanding of the emission from a calm ocean surface,

their accuracy in providing values of the ocean still needed to be examined. Our present

investigation of the specular sea emission seen from space provides field verification of the

sea water specular emissivity over broader regions of the oceans. In this work, we

investigate and adjust two ocean dielectric models using well calibrated radiometer data from

the TOPEX/Poseidon satellite mission, paying particular attention to reducing the overall

bias of the estimated brightness. In addition, we evaluate the performance of several models

for their dependence on salinity and sea temperature.

The modified models exhibit significant improvements in the estimate of TB. Of the two

modified models, ModE exhibits superior overall performance. It has the lowest bias at both

frequencies (0.16 and 0.14K, respectively), which is indicative of the accuracy of the model.

Its frequency dependence was decreased from -2.3 to 0.30K. In addition, ModE has the

lowest dependence on sea surface temperature and the lowest RMS difference of 2.58K and

3.52K for 18GHz and 37GHz, respectively. For these reasons, we recommend this model for

future remote sensing applications involving microwave emissions from the ocean.

27
REFERENCES

Altshuler, E. E. and R. A. Marr, “A comparison of experimental and theoretical values of

atmospheric absorption at the longer millimeter wavelengths,” IEEE Trans. Antennas
Propagat., vol. 36, no. 10, pp. 1471-1480, Oct. 1988.

Aydin, K. and C. Tang, “Millimeter wave radar scattering from model ice crystal
distributions,” IEEE Trans. Geosci. Remote Sens., vol. 35, pp. 140-146, 1997 a.

Cruz-Pol, S. L., C. S. Ruf and S. J. Keihm, “Improved 20-32 GHz Atmospheric Absorption
Model,” Radio Sci., vol. 33, no. 5, pp. 1319-1333, 1998.

Draine, B. T. and P. J. Flatau, “Discrete-dipole approximation for scattering calculations,” J.

Opt. Soc. Am. A, vol. 11, pp. 1491-1499, 1994.

Doviak, R. J.and D. S. Zrnic, Doppler Radar and Weather Observations, Second edition,
Academic Press, San Diego, 1993.

Evans, K. F. and J. Vivekanandan, “Multiparameter radar and microwave radiative transfer

modeling of nonspherical atmospheric ice particles,” IEEE Trans. Geosci. Remote Sensing.,
vol. 28, pp. 423-437, July 1990

Keihm, S. J., C. Ruf, V. Zlotnicki and B. Haines, “TMR Drift Analysis,” Jet Propulsion
Laboratory, Internal Report, October 6, 1997.

Klein, L. A., and C. T. Swift, “An Improved Model for the Dielectric constant of Sea Water
at Microwave Frequencies,” IEEE Trans. on Antennas Propagation, Vol. AP-25, No. 1, 1977.
Hogan, R. J. and A. J. Illingworth, “The potential of spaceborne dual-wavelength radar to
make global measurements of cirrus clouds,” J. Atmos. Oceanic Technol., vol. 16, 518-531.
1999

Keihm, S. J., Y. Bar-Server, and J. C. Liljegren, “WVR-GPS Comparison Measurement and

Calibration of the 20-32 GHz Tropospheric Water Vapor Absorption Model”, IEEE Trans.
Geosci. Remote Sensing. 2002, 40, No. 6, pp. 1199-1210

28
Lhermitte, R., “A 95 GHz Doppler radar of cloud observations,” J. Atmos. Ocean. Technol.,
vol. 4, pp. 36-48, 1987.

Li, L., S.M. Sekelsky, S.C. Reising, C.T. Swift, S.L. Durden, G.A. Sadowy, S.J. Dinardo,
F.K. Li A. Huffman, G.L. Stephens D.M. Babb, and H.W. Rosenberger, “Retrieval of
Atmospheric Attenuation Using Combined Ground-based and Airborne 95 GHz Cloud Radar
Measurements,” J. Atmos. Oceanic Technol., vol. 18, 1345-1353. 2001

Matrosov, S. Y. “Radar reflectivity in snowfall,” IEEE. Trans. Geosci. Remote. Sens., vol.
30, pp. 454-461, 1992.

Oguchi, T. “Electromagnetic wave propagation and scattering in rain and other

hydrometeors,” Proc. IEEE, vol. 71, pp. 1029-1078, 1983

Ray, P. S., “Broadband complex refractive indices of ice and water,” Appl. Opt., vol. 11, pp.
1836-1844, 1972

Rosenkranz, P. W., “Absorption of Microwaves by Atmospheric Gases”, In: Atmospheric

Remote Sensing by Microwave Radiometry, Chapter 2, Ed. By Jansen, Wiley, New York,
1993.

Sekelsky, S. M., “Multi-frequency radar Doppler Spectrum Measurements of Cirrus Clouds,”

Geoscience and Remote Sensing Symposium. IGARSS '01., vol. 2, 697 –699 2001.

Ulbrich, C. W., “Natural variations in the analytical form of the raindrop size distribution,” J.
Climate Appl. Meteor., vol. 22, pp. 1764-1775, 1983.

Wilheit, T.T., “The Effect of Wind on the Microwave Emission From the Ocean’s Surface at
37 GHz,” J. Geophys. Res., Vol. 84, No. C8, pp. 244-249, 1979.

29
6 APPENDIX A. IDL CODES FOR DBZEMIN

;********dBZemin Program********
;********MAIN PROGRAM*********

LoadCT, 5

sondefilename='c:/jorgemvg/prog-idl/DataAustralia/Radiosonde/sonde.951127.025800.cdf'

mwrfilename='c:/jorgemvg/prog-idl/DataAustralia/Radiometer/mwr.951127.000020.cdf'

;**Function to read microwave-radiometer data

get_mwr_cdfdata, mwrfilename, VAPcm, LIQcm, DEWflag, t_begin$, $
date$,unix_time,sec_into_UTCday

;**Function to read radiosonde data

get_sonde_cdfdata, sondefilename, tdry, sh, rh, dp, h, pres, $
wspd, deg, t_begin$, date$,unix_time,sec_into_UTCday

;**Function to read Radar data

Radar,z_mask_range_33,z_mask_range_95

;;********************************************************************************
h = h/1000. ;altitude [Km]
pres = pres/0.1 ;pressure [Kpascales]

; a extrapol le debe entrar h en (Km) y pres en (KPascales)

extrapol_general,z_mask_range_33,h,tdry,pres,sh,altura,temperatura,presion,humedad_especifica
altura = altura*1000.;altitude [m]
presion = presion*0.1 ;pressure [ mbars]
tdry = temperatura ;temperature [deg C]
sh = humedad_especifica ;specific humidity [gm^-3]
pres = presion
h = altura

;omit radiosonde data above 35 km to speed up processing

alt=30000.
hlimit=max(where(h LT alt))
tdry=tdry(0:hlimit) & sh=sh(0:hlimit)
h=h(0:hlimit) & pres=pres(0:hlimit)
;setup regular height grid for profiles

num_elem=500
del=alt/num_elem
h_prof = findgen(num_elem) * del ; 0-35km

30
tdry = INTERPOL(tdry, h, h_prof) ; regrid profiles
sh = INTERPOL(sh, h, h_prof) > 0.
pres = INTERPOL(pres, h, h_prof)
h = h_prof

; compare radiosonde and mwr data

L=0
FOR i=0,n_elements(h)-2 DO L=L+sh(i)*(h(i+1)-h(i))
L=0.001*L ; mm of water vapor in column from radiosonde profile
L = L*0.1 ; cm of vapor ... compare to Vapcm from microwave radiometer
; probably will not be exactly the same since different meas. locations

; if mwr data valid then use to correct radiosonde humidity profiles

indx = where(dewflag LT 1) ; filter out flagged data
sh = sh*mean((vapcm(indx)))/L ; scale radiosonde profile by mwr total
prSH=fltarr(2,n_elements(sh)/2)
FOR par=0,(n_elements(sh)/2)-1 DO BEGIN
prSH(0,par)=sh(par*2)
prSH(1,par)=h(par*2)/1000.
ENDFOR

for the gases atten. SLCP June 2001

PRO atten_humidity_liebe, sh,tdry,pres,fi, h, AGASEOUS,Agas_liebe, KGASEOUS

_ground(n,rg) = extinction rate at ground level [dBkm^-1]

height=h/1000. ; h esta en metros , height esta en kilometros
rangesamples = size(height)
rangesamples = rangesamples(1)
AH2O_liebe(i, j) = TOTAL(KH2O_liebe(i, 0:j)*ABS((height(1:j+1)-(height(0:j))) > 0.))
ENDFOR
ENDFOR
AH2O_liebe(*,rangesamples -1 ) = AH2O_liebe(*,rangesamples -2 )
PRO scanning_new2, sh,tdry,pres,fi, h, AGASEOUS,Agas_liebe, KGASEOUS,LF1,LF0,ATKF1,ATKF0

ATKF1(zeta,altura)=TOTAL(KGASEOUS_EQUIf1(zeta,0:altura)*ABS(((proyeccion_radio(zeta,1:altura+1)-
proyeccion_radio(zeta,0:altura))/sin(angles(zeta) * !pi/180)) > 0.))

ATKF0(zeta,altura)=total(KGASEOUS_EQUIf0(zeta,0:altura)*abs(((proyeccion_radio(zeta,1:altura+1)-
proyeccion_radio(zeta,0:altura))/sin(angles(zeta) * !pi/180)) > 0.))

dbz0=imgpolrec(dbz0, 0., 91., 0., 40., 0., 25., .03, 0., 25., .03)
ocu95=intarr(n_elements(ymax),20)
Position = [0.1, 0.9, 0.9, 0.95], Color=!P.Background
stop

END

31
32
APPENDIX B PROGRAMS FOR BULLET AND DWR

APPENDIX B1 IDL PROGRAM FOR REFRACTION INDEX

;****IDL PROGRAM***
;*** Refraction Index
n=200 ;
index=complexarr(2,n)
D=fltarr(n)
aeff=fltarr(n)
p=fltarr(n)
fi=fltarr(n)

for i=0, n-1 DO BEGIN

D(i)=(i+1)*1E-2 ;D[mm]
p(i)=0.78*D(i)^(-0.0038) ;Heymsfield density relationship bullet
pi=0.916 ; pi[g*cm^-3]
fi(i)=p(i)/pi
ni=[complex(1.785, 0.000235),complex(1.784, 0.00010)] ; 33GHz , 95GHz paper Ray 1972
f=fi(i)
for k=0, n_elements(ni)-1 DO BEGIN
n=ni(k)
index(k,i)=(2.+(n^2)+2.*f*(n^2-1))/(2.+(n^2)+f*(1-n^2))
end
aeff(i)=1e+3*D(i)/2 ;[um]
END
END