An airborne microwave radiometer and measurements of cloud liquid water

Acknowledgements This work was jointly supported by the Key Pro-Chinese Science Bulletin2003 Vol. 48 Suppl I 82 -87jects of the Ministry of Science and Technology of China (Grant Nos.2001BA610A-06 and 2001BA901A41), the National Natural ScienceAn airborne microwaveFoundation of China (Grant Nos. 40333033 and 40175001) and the CASradiometer and measurementsReferences1. Shepherd, J. M., Pierce, H, Negri, A. J,, Rainfall modification byof cloud liquid watermajor urban areas: observations from spaceborne rain radar on theTRMM salelie, Journal of Applied Meteorology, 2002, 41: 689 -_LEI Hengchi', JIN Dezhen?, WEI Chong'& SHEN Zhilai'701.2. Rozoff, C. M.,. Cotton, W. R., Adegoke, J.O.. Simulation of St.1. Institute of Atmospheric Physics, Chinese Academy of Sciences,Louis, Missouri, land use impacts on thunderstorms, Journal of2. Jilin Weather Modification Office, Changchun 130062, ChinaBeijing 100029, China;Applied Meteorology, 2002, 42: 716- -738.Correspondence should be addressed to Lei Hengchi (e-mail: leihc@Sun, L, Xu, L. Z., Study on genesis conditions of convectivemail.iap.ac.cn)weather in Beijing area in summer, ACTA Scientiarum NaturaliumAbstract A single -channel (9.5 mm) airborne microwaveUniversitatis Sunyatseni (in Chinese), 1996, 35: 189 - -193.radiometer with one antenna is developed. The retrieval4. Chen, L. D. Chen, s. P., The investigation on the severe convective methods and primary observation results of cloud liquidactivities over Beijing area, Scientia Meteorologica Sinica (in Chi- water and super-cooled cloud liquid water are discussed. Thenese), 1994, 14(3): 216--224.aircraft experiments show that the cloud liquid water andWang, X. F, Ding, Y. H, Study on method of short -range forecastsuper-cooled liquid water can be sensitively monitored atsome level of accuracy by the radiometer. The results ofof severe convective weather in Beijing area, Chinese Joumal ofcloud liquid water content are reasonable and correspondAtmospheric Science (in Chinese), 1994, 18 (2): 173- -183.well with the surface radar echo intensity. The design of the6. Sun, M. S.,. Wang, X. M., Luo, Y. et al, A prospect forecasting airborne radiometer and its retrieval methods are feasible,method study of severe convective weather in Beijing area, Quar-giving it application value.terly, Journal of Applied Meteorology (in Chinese), 1996, 7: 336 - Keywords: airborne microwave radiometer, microwave re-343.mote- sensing, retrieval methods, calibration methods, cloud liquid7. Fu, D. H, Guo, X. L, Xiao, W. A. et al, Numerical study on theformation a severe storm accompanied with gale and heavy rain in DOI: 10.1360/03wd0462Beijing, Journal of Nanjing Institute of Meteorology (in Chinese),2003, 26(2): 190 - 200.The cloud liquid water content is an important p-rameter in atmospheric physics, and weather modific-8. Guo, x. L, Huang, M. Y, Haicloud model with hail-bin microphysics, Atmospheric Research,formation and growth ina 3.Dation studies focus on the super-cooled liquid water con-tent. At present, only a few methods can be used to meas-2002, 63:59 - 99.9. Guo x. L. Huang, M. Y., Hong, Y. C. et al., A study ofure cloud liquid water and super-cooled liquid water, butthese methods cannot satisfy the needs of research andthee-dimensional hail category hailstorm, part I: model descriptionapplication. Among the measuring methods, the directand the mechanism of hail recirculation growth, Chinese Journal ofdetection instruments have been developed in the past,Atmospheric Science (in Chinese), 2001, 25(5): 707 -720.such as the hot-wire device, the forward- scattering spec-10. Guo, X. L, Huang, M.Y., Hong, Y.C. et al, A study of three- trometer probe (FSSP) manufactured by PMS (Particledimensional hail category hailstorm, part II: Characteristics ofMeasurement System Inc.)， the cloud droplet colli-hail-caegory size distribution, Chinese Joumnal of Atmospheric sion-copy device, and the freezing ice device that meas-Science (in Chinese), 2001, 25(6): 856 . - -864.ures the super-cooled liquid water manufactured by11. Dudhia, J.. Gll, D.. Guo, Y. R. etal,. SUNCAR Mesoseale Mod- Rosemount Inc. Now direct measurement has high accu-eling System Tutorial Class Notes and User' s Guide: MM5 Mod-racy in local position, but the sample volume is limitedeling System Ve rsion 3, June 2000.and the sample representation is not ideal. The sample12. Hjelmfelt, M. R., Orville, H. D., Roberts, R. D. et al, Observa-representation is markedly improved with the passive e-tional and numerical study of a microburst line producing storm,mote-sensing methods introduced into meteorology sincethe 1960s. The application of ground base microwave 1a-Journal of the Atmospheric Sciences, 1989. 46(2): 731 - 2743.13. Guo, X.L, Nino, H., Kimura, R., Numerical modeling on a haz-diometers t中国煤化工in mso-miroscale and wedEcame a new toolardous microburst-producing hailstorm. Towards Digital Earth-because of thYHC N M H Gn detection accu-Proceedings of the International Symposium on Digital Earth, Bei-racy, automatic and continuous duty cycle, and ability tojing: Chinese Science Press, 1999. 383 - -398.detect super-cooled liquid water content from mixed-(Received August 26, 2003; accepted December 31, 2003) phase cloud, etc. But ground-based microwave radiome-82Chinese Science Bulletin Vol. 48 Supp..I December 2003ters cannot meet the requirements of meso -micro scale was chosen to sense the cloud liquid. To construct the sta-studies because it is not easy to move the instrumentstistical cloud sample and perform simulations through thequickly, which limits the observation scope. The inevita-radiation transmission model, the history sounding data ofble trend of installing remote-sensing instruments on air-Changchun weather station of Jilin Province in July wereplane for studing on meso -micro scale weather wasused as the background air samples; a general method waspointed out in the 1980s-. This was a better applicationintroduced to construct the cloud sample'5 ! with a conceptsince airborne upward microwave radiometers have notmodel of the vertical liquid water profile in Jilin Prov-only the merits of ground-based microwave radiometersincelb. The result showed that the total liquid water con-but they can also directly enter the cloud and observe ittent Lot can be expressed as a quadratic function of thewith flexibility. At present, work on the design and appli-receiver brightness temperature Tbgrd at the ground. Othercation of airborne radiometers is still rare. Only a fewthan the existing results of ground-based microwave ra-works have reported the use of airborne upward micro-diometer studies, we further obtained the new result thatwave radiometers to detect cloud liquid water content inthe U.S.A. and Russial2-3]. In prior studies, the radiometerthe vertical integrated super-cooled liquid can be ex-was installed in the aircraft with two antennas mounted onpressed as a quadratic function of the receiver brightnesstop of the fuselage. The instrument switches automatcallytemperature Tboc at the zero level. From these we assumedbetween two fixed antennas which are elevated at equalthat the retrieval formulation of airplane-detected verticalangles above the longitudinal axis of the aircraft; oneintegrated super-cooled liquid at different levels can bepoints forward, the other af. The vertical cross section of expressed as a quadratic function of the receiver bright-the distribution of cloud liquid water can be obtained by ness temperature whose coffcients vary with altitude.using the method which is similar in principle to variousThe assumption was confirmed by later retrieval methodimaging techniques, such as computer-aided tomographystudies. It showed that the theory of the measurement ofas used in the medical field4. The technique still has vertical integrated total cloud liquid water or integratedproblems in accuracy and is currently being examined total super-cooled liquid water content of stratus cloudtheoretically; it needs time before it can be applied insystems using a single band radiometer is reasonable.practice.Quantitative estimates also showed the differences of sig-Precipitation enhancement operations via aircraftnal dynamic variation between airborne radiometers andseeding. have been crried out in more .than 10 provinces ground-based radiometers. The airtborne radiometer wein North China to overcome the drought problems in ag-designed should exactly leceive a brightness signal fromriculture and industry. These peculiar conditions not onlyas high as 200 K to as low as 10 K in an environmenthave already put forward the practical demand to developwhose temperature varies greatly from 30.( at the surfaceairborne microwave radiometers but also have offered awide platform for their application. According to the de-minutes. We had not encountered the situation of so greatmand of weather modification and aircraft conditions,a dynamic range, especially the very weak signal in thebased on the experience of using ground-based radiome-low temperature range, when we designed ground-basedters for many years, as well as the progress of microwaveradiometers before. We clearly knew the technical diffi-technology domestically in recent years, the developmentof airborne stragh-upward single antenna radiometers culties and rcognized that the new scheme should replacemay quickly lead to their practical application. For thisthe already-existing ground-based radiometer designreason, the Institute of Atmospheric Physics, Chinesescheme. The prior research results gave us the basis of theAcademy of Sciences (CAS), in cooperation with the Jilininstrument index.Weather Modification Office, commissioned the Institute(.. i) Instrument technique scheme， index， andof Geography, CAS, to construct a single band (9.5 mm),configuration. We designed the radiometer with a newsingle antena airtborne upward microwave radiometer. scheme called“digital amplification fluctuation compen-Then the retrieval method was developed and aircraft ob- sation" that was put forward by the Institute of Geography,servation was carried out during the precipitation enCAS. The radiometer design is shown in the block dia-hancement season from April to July in 200i and 2002.gram of Fig. 1. By periodically judging the variation ofThe problems and their resolution during the development system amplification with a radio frequency switch toof the instruments, the calibration and retrieval methods, examine the system output of a benchmark, we can mod-and some trial results of the measurement of cloud liquid ify the radiomeey 4^W d馆itital monitor unit.water and super-cooled liquid water content are brieflyThe receiver中国煤化工pling as a way todescribed in this paper.fit the needIY片C N M H G brightness tem-perature and to soIve ue procen U1 10W radiation bright-1 Instrument and calibrationness temperature calibration. The coupled noise tempera-(_ i) Prior research. The frequency of 31.65 GHzture is about 350 K. The stability is better than 0.1 K usingChg数fjence Blletin Vol.48 Supp.. I December 200333the side loop of a solid noise source to compose the refer- points of (Te1, Tb1) and (Te2, Tb2) as a calibration curveence benchmark source.in rectangle coordinates we call it a two-point calibration.The instrument operates at 31.65 GHz and the medi-The result of measuring the instrument resolution inate frequency bandwidth is 400 MHz. The size of thelaboratory testing indicated that the output signal meanpyramid horn antenna is 58.2.>79.0 mm2, the bandwidth square root deviation varied from 0.12 K to0.18 K, whichof 3dB is 7.2 ; the integration time is Is, the sensitivity is can approximately represent the instrument sensitivity.0.2 K, the cllction data rate is once every 2 seconds, and The field clear -sky fight observation indicated that thethe power is 200 W.sensitivity varied from0.12 K to 0.65 K.The pyramid horm is installed on the top of the air- 2 Retrieval methodplane, vertically oriented towards the sky and protected by(_ i) Radiation transmission equation and retrievalan antenna cover. The receiver and data acquisition systemmethod. The remote sensing equation of an airborneare placed in the airplane. The data can be displayed andupward radiometer in conditions under the Rayleigh-Jeansstored on the computer.approximation and in a parallel plane atmosphere isAnten.T(y)= T(o).cJ, K()djSolid-s.DTE(1)Isolatorsource"dh,where y is the flight height, To(y) is the brightness tem-Dig. control unitperature of cloud atmospheric downward radiation rDirec.couplerceived by the radiometer, T6(- x) is the cosmic backgroundradiation brightness temperature which is a constant 2.7 K,上↑P. volt.T(h) is the profile of atmospheric temperature, K:(h) is theDirec. couplerTR、RF switchreg.-net.mass total absorption coefficient per unit volume at heighth, which is the sum of water vapor absorption coffi-cient Kav, oxygen absorption coefficient Kao and cloudwater absorption coefficient Kac. These absorption coeffi-MIXcients are not only functions of wavelength but they alsoAnalog switchrelate to an atmospheric parameter. The oxygen and watervapor absorption coefficients are functions of atmosphericpressure, temperature and humidity, and they can be cal-Lo|If amplifierV. amp.culated from the sounding profile datal8. The cloud water0Habsorption coefficient is a function of temperature andcloud liquid water content M(h)8. Because of the lack ofFig. 1. Block diagram of aitbormne microwave radiometer.information on cloud liquid water content in soundingdata, a cloud model was constructed to simulate the verti-(ii) Practical calibration method. The instrument cal distribution of cloud water based on the characteristicsvas calibrated by using a two-point calibration method of a study object that will be discussed in the next section.which joins the absolute calibration and meteorology cali-Combining a piece of the sounding profile and the simu-bration, because the antenna cannot be pitched and due to lated cloud liquid distribution generates a piece of theother limitation conditions. To calibrate the high tempera- cloud atmosphere sample. At any observation height y, theture, an antenna cover made from microwave absorption integrated cloud liquid water content L(v) can be writtenmaterial was used to cover the antenna, and the instrumentoutput code Te1 and environment temperature were re-L())=」。M (h)dh.(2)corded. The material has the radiation characteristics of ablackbody because the absolute absorption of microwavesA statistical retrieval method is used to select datameans that the brightness temperature Tb1 is equal to thefrom the abundant historical sounding data that can repre-physical temperature at radiation balance. To calibrate the sent the climate properties of Jjilin Province and have thelow temperature, we recorded the instrument output code seasonal characteristics tn he ahle tn ralrulate a great dealTe2 when the airplane landed or flew when there is noof Tb() and中国煤化工in the function ofcloud in the instrument zenith direction. Then we calcu- dependent v. CNMH Gt variable Tb()lated the clear-sky downward radiation brightness tem- through statisaral ncgiosuu as uic lcucval formula.perature Tb2 at the surface or corresponding flying level(-1) Cloud model and samples. The geometricusing a real-time sounding. Using the line between theframe of cloud liquid vertical distribution is determined by84Chinese Science Bulletin Vol. 48 Supp..I December 2003a relative humidity threshold comparison. The model method of section 2(- i) by selecting sounding samplescloud base height is determined by introducing meteoro- from April to July that have 100 pieces each month fromlogical eye-observed minimum cloud base height. In di-the Changchun weather station. To ensure the objective-agnosing the level above the cloud base, we consider theness of the testing, these samples are different than thosecloud to exist when the relative humidity at a given level used to calculate the rtrieval cofficients in section 2(- i)is greater or equal to that at the cloud base. There are twoThe integrated cloud liquid content value computed fromimprovements seen in comparing domestic work with thateq. (2) is taken as the true value of the simulation. Thedone abroad: first is the introduction of observed cloud observation value of the simulation is obtained from eq. .base, the second is using the real time cloud base relative(1). Then we obtain the retrieval value of integrated cloudhumidity to replace the threshold relative humidity value.liquid water by retrieving the simulation observation usingWith these two improvements, the retrieval accuracy in-the retrieval coefficients. The retrieval precision can becreases apparently. Cloud liquid water content is chosenestimated by comparing the deviation of the retrievalfrom a vertical distribution model of typical stratus. This value and the simulation true value.means that the peak height hp of the cloud liquid water(2) Test result. The statistical result that uses acontent vertical distribution is placed at a quarter of themonthly unit from April to July gives us that the retrievalheight of the cloud thickness and the peak cloud watererror tends to decrease as the height increases, and thecontent M max(g/m ) is proportional to the relative humidityRMS (root mean square) relative deviation is 15% to 25%at the cloud base. Cloud water content decreases linearlyat the surface and 5% to 10% at a height of 6 km.with height from the peak to the cloud base and to the3 Measurement of cloud liquid water contentcloud top, and it becomes zero at the cloud base and cloudtop. We assume that no liquid super- cooled water exists(- i) Outline of observation.The aircraft. installedwhen the temperature is lower than -202Cairborne radiometer made 10 flights totalling 20 hours,To calculate the retrieval formula cofficients, we including two clear-sky calibration flights during thechoose four months of historical sounding data with eachweather modification period from April to July in 2001month having 100 pieces from April to July as the basicand 2002. The observation was carried out following thedata to generate the cloud atmospheric samples.operation flight course in most situations, but only a few(ii) Retrieval formula and coefficients.pre-designed observations were made. The seeding andforming the regression of cloud sample calculation resultsobservation cloud was static stratus or stratus of embed-from eq. (1) and eq. (2) in Jilin shows that total liquid wa-ded convection cells. The variation of cloud integratedter content L(h) above the flight height of the middle andliquid water content and super- cooled liquid water contentlow layers of the troposphere has a quadratic relation withspans two orders of magnitude from 10' to 103 g/m’ inbrightness temperature Tb(h). The regression equation isdifferent types of cloud and different locations in the samecloud. In all observations, horizontal flight below the zerowritten aslevel was only taken on 8 July, 2001 while the others wereL(h)= a(h)+a2 (h) T (h) +a3(h) T (h),over the zero level. As we can see, all the aircraft observa-where a1(h), a2(h) and a3(h) are the quadratic regressiontions include the warm layer and super-cooled layer due tocoefficients that we wanted. In the equation, we use h astake-off and landing. It should be pointed out that thereheight to replace y in eqs. (1) and (2).was a short period of clear sky by real time eye measure-The retrieval coefficients change with height and ment on 8 July, 2001. The data processing showed that theseason. The result of the retrieval cofficients as functions minimum integrated cloud liquid was 0 g/m3 at the corre-of height every month tells us that a1(h), a2(h), and a3(h) sponding time. The agreement between the clear-sky pe-can all be expressed as polynomial functions of height. If riod and minimum zero value shows that the instrumentwe choose the first four terms, the fiting accuracy is and retrieval method can reflect the true value of cloudgreater than 97 percent. The fiting expression isliquid water content.(- i) Sample analysis. We give a sample analysis ofa;(h)=2 bj;:h', (j=1,2,3), .(4) observing cloud liquid water and super cooled liquid wa-ter on 22 April, 2002. The airplane took off from Chang-where bj are the coefficients of the fitted polynomial. Ifchun at 08:25, passing Shuanliao, Siping, Liaoyuan, Pan-we know the brightness temperature measured by the ra-shi and Shuangyang, and landed at Changchun at 1 1:00.diometer, we will obtain the cloud liquid water content inCloud types中国煤化right line The Asthe air via eqs. (3) and (4).top height waF the radar echo.(_ i) Numerical simulation of retrieval precisionFigureYHC N M H Gll variables. The(1) Test method and test sample. We use a numericalabscissa axis is time, the solid line is integrated liquid Wa-simulation method to test retrieval precision. The inde- ter content L (g/m3), the dashed line is the flight heightpendent test sample of cloud is constructed using the curve measured by GPS， the fine solid line is theChit数jence Bulletin Vol. 48 Supp.I December 200335zero-level height. The marks of CC, SL, SP, LY, PS, and per-cooled liquid water content in the unit of g/m', whichSY on the abscissa represent the times the airplane passed are 0.14, 0.22, and 0.05 g/m', respectively. The values areChangchun, Shuangliao, Siping, Liaoyuan, Panshi, and reasonable.Shuangyang respectively. Most of the observation flightoccurred above the zero level which was located at about1460 m height, except during take-off and landing. Wedivided the period into three time series, and in the figureeach is marked as a-b, c-d, and e-f, where the super-cooled22 April, 200209:33:33 \liquid water content had three levels., 10700 -. 4000200 km100 kmKcc600- 3000500日400-5 3000Cr2000是120-一25200-t 1000100 .BLγps sy c|08:0009:0010:0011:00Fig. 3. The PPI radar echo at 09:33, 22 April, 2002.Fig. 2. Time series of measurements on 22 April, 2002.From the above analysis, it is shown that the de-Data analysis shows that the cloud liquid contentsigned airborne radiometer can objectively reflect thecorresponds to radar echo intensity at the surface. Four magnitude and variation of liquid and super-cooled liquidPPI echoes obtained during the flight are marked as R1,water content in different cloud systems, and the observa-R2, R3, and R4 as the radar scanning times on the abscissation results have practical value.in Fig. 2. Fig. 3 gives the echo at the time of R3 (09:33),(i) Vertical profile of liquid water content.where the airplane position markedas |■gives the radar vertical profile of liquid water Content can be gotten byobservation time. From Fig. 2, the cloud liquid water con-using the radiometer observed data when aircraft declinestent in section ab (09:02- 09:34) is as low as about 250or ascends because of the high detection sensitivity andg/m2 on average, which is the flight line between Chang-time resolution of the radiometer. Using the observed se-chun, Shuanliao, and Siping. Radar echo intensity is ries data of L when aircraft declines or ascends, we calcu-weaker than 20 dbz. Cloud liquid water content in section late the ratio of the L difference dL between two suitablec-d (09:38 -10:08) is between 300 to 700 g/m2, whichis flight heights and the height difference dh to obtain theapparently greater than the other two time series. At the vertical profile of liquid water content, w,same time, the timescale of 2 to 4 minutes for which thew= -dL/ dh,space scale is 8 to 16 km in fluctuation may be the ine-where the minus sign represents the opposite trend be-quality structure of convection when the airplane passedween integrated cloud liquid variation and height varia-Siping and Liaoyuan to Panshi. From Fig. 3, the strong tion.echo intensity above 30 dbz really existed between SipingFigure 4 gives the vertical profile of liquid waterand Liaoyuan, and the maximum echo intensity reachedcontent of the ascend flight on 8 July, 2001; the height25dbz between Liaoyuan and Panshi. In Fig. 2, cloud liq-interval dh is 200 m. The operation flight line is covereduid water content in section ef (10:14- 10:50) reaches by stratus, Sc at the lower-layer and As at the upper-layer.the minimum between 200 to 100 g/m2. In that period, the The profile shows that the cloud base is about 800 m; theairplane flight from Shuangyang to Changchun entered a liquid water content peak value is 0.42 g/m', 700 m aboveweak echo area below 5dbz.The analysis again gives posi-the cloud base; liquid water content decreases as heighttive correlation between cloud integrated super cooled increases above the peak point; the liquid water content is .liquid water content and radar PPI echo intensity.only 0.05 g/m3 at a height between 1700 to 2000 m (900The observed cloud thickness above the flight height to 1200 m ab中国煤化工uid water contentat the three time series is about 1500, 2080, and 3300 mslightly increiYHCNMHult corresponds tothat can be calculated according to radar observed clouda general straltop height. Dividing the average integrated super-cooled shows the potential of the airborne radiometer to measureliquid content by cloud thickness gives the average 91- the cloud profile. If we improve the flight route, the air-86.Chinese Science Bulletin Vol. 48 Supp..I December 2003borme radiometer will have a bright future as an instru- acceptable precision of weather modification in the area.ment to measure the cloud liquid water profile.The observation data analysis tells us that the radi-ometer can objectively reflect cloud liquid and su4.0per- cooled liquid water content and variation. A compari-8 July, 2001son analysis gives the positive correlation between the09:34-09:423.0-upper layer cloud liquid water content and radar echo in-tensity at the surface. The observation results have a prac-星20-tical application value that proves the new instrument anddesign scheme meet the detection demands, and that the0-instrument calibration method and retrieval method arreasonable.As a new instrument, the emphasis of this paper on0.0.00.10.203).40.5instrument validation and the observation analysis is pre-W/g.m"paratory. Further study will focus on how to demonstratedifferent cloud systems, cloud liquid and super-cooledFig. 4. A case of rtrieved vertical profile of liquid water content.liquid water content distribution at different locations.4 ConclusionUncertainty exists in the calibration method and retrievalThis paper introduces the design of an upward air-method, and these need to be improved in the future.borne radiometer. The prior research through numericalAcknowledgements This work was supported by the National Naturalsimulation shows the feasibility of single band verticalScience Fundation of China (Grant No. 40275002) and partly by theupward detection and establishes the design difficulties of[ Jnion I aboratorv for Weather Modification Between Jilin Province andCMA.requiring high sensitivity and big span due to great signaldynamic variations, and especially the need for exactReferencesmeasurement of a weak signal. The pre-existing scheme of1. Robert, J.. Serafin, Gerard Szcjwach, Byron B. Phillips, Applica-ground-based radiometers cannot meet the demands, sotions of airborne remote sensing in atmospheric sciences research, J.we developed a radiometer with a new scheme called aG. R., 1986, 91(C2): 2510- 2516.“digital amplification fluctuation compensation radiome-2. Warner, J. Drake, J. F., Field tests of an airborme remote sensingter”. The specifications of the radiometer are given in thetechnique for measuring the distribution of liquid water in convec-paper. The paper also gives the result of sensitivity intive cloud, J. Atmos. Oceanic Tech., 1988, 5: 833- -843.laboratory experiments and field flights that shows tha3. Koldaev, A. V., Yu, V. Melnichuk, Mironov, A. F.. Remote sensingthe index reaches the design requirements. A practicalinvestigation of cloud liquid water space distribution, Proceedingscalibration method suitable for field flight is also dis-cussed in the paper.Vol. 1, Montreal: EIsevier Science Publishers, 1992, 573- -575.To get the retrieval method of airborne upward a4. Drake, J. F., Warner, J, A theoretical study of the accuracy of 10-diometer remote sensing of cloud liquid water, this papermographic retrieval of cloud liquid with an airborne radiometer, J.discusses the microwave radiation transmission model of aAtmos. Oceanic Tech., 1988, 5: 844- -857.parallel plane atmosphere that changes with instrument5. Decker, M. T, Westater, E. R.. Guiraud, F. O., Experimentalheight in real time and the cloud liquid water vertical dis-evaluation of ground-based microwave radiometer sensing of t-tribution model through a cloud sample constructed by amospheric temperature and water vapor profiles, J. Appl. Meteor.,sounding profile. Numerical simulation results show that1978, 17: 1788 -1795.cloud vertical path integrated liquid water content can be6. Wang, x. L, Jin, D. z, Treatise on Cloud and Precipitation Physicsexpressed as a quadratic function of atmospheric down-(in Chinese), Beijing: Beijig Science and Technology Press, 1987,ward radiation brightness temperature at each level of the39. -53middle and lower layers of the troposphere; this function7. Huang, R. H., Zou, S. x., Remote sensing of precipitable wateris the retrieval formula of the radiometer observed value.vapor and liquid water of cloudy atmosphere by dual-wavelengthThe three coefficients of the retrieval formula vary withmicrowave中国煤化工(in Chinese), 1987,location, month, and height. Numerical simulation of an11(4): 397independent sample gives that the retrieval error decreases8. Zhou,X.J.MYHC N M H GMicrowave Radiaionas height increases, and the RMS relative error is 15% toand Remote Sensing (in Chinese), Beiing: Science Press, 1982.25% at the surface and 5% to 10% at 6 km height. Thismeans that the retrieval formula and coefficients meet the(Received August 22, 2003; accepted December 31, 2003)Chit数jence Bulletin Vol. 48 Supp.I December 200337