Poceedings of the 4th WSEAS Intenational Confeence on REMOTE SENSING (REMOTE'8) Channel Fequency Optimization of a Futue AMSU-B-Like Millimete-Wave Radiomete fo Integated Wate Vapo Retieval in pola egions HAIBO ZHAO, JUNGANG MIAO Electomagnetics Laboatoy Beihang Univesity Xueyuan Road 37, Haidian Distict, Beijing, 1191 CHINA, PEOPLE S REPUBLIC OF Abstact: - Wate vapo infomation is vey impotant to undestanding global climate change, especially in the climatologically sensitive pola egions. In 1998, Miao poposed a method to etieve Integated Wate Vapo (IWV) content using SSM/T2 channels ove the ice coveed pola egions. It has been poved to be effective in both Antactic and patial Actic egions by subsequent studies. Unfotunately, with cuently available millimete adiomete channel combinations, the etievable IWV content can not exceed 6kg/m 2 using Method of Miao. In this pape a new goup of channel fequencies aound 183GHz ae established fo a futue AMSU-B-like adiomete, which is optimized fo lage etievable IWV ange in pola egions. Results show the etievable IWV ange has been expanded. The atmosphee pofiles being used in the optimization pocess ae in situ adiosonde data fom IGRA, suggesting the optimized channel combination is not only wokable in theoy but also can be put into pactice fo IWV etieval in pola egions. Keywods: - channel fequency optimization, millimete-wave adiomete, integated wate etieval in pola egions, Method of Miao, passive micowave emote sensing fequencies (89, 1, 183+/-7, +/-3, +/-1 GHz, etc.) is pimaily tageting fo wate vapo pofile etieval in the low and mid-latitude aeas, and may not be optimized fo pola egions. It is known that, the fathe a channel is located fom 183GHz absoption peak, the less sensitive and moe toleant it is to wate vapo. So in ode to expand the etievable IWV ange with Method of Miao, it is necessay to find possible doublesideband channel combinations located fathe to 183GHz than available SSM/T2, AMSU-B channels (the fathest being 183+/-7GHz). In this pape, up to +/-37 GHz aound 183GHz ae consideed and clea sky condition is assumed though out the simulation. 1 Intoduction Numeic modeling studies have shown that pola egion is one of most sensitive egions on Eath to global climate change [1]. And being the most impotant geen house gas in the pola egions [2], wate vapo does not only povide cucial infomation about climate change but also foms a feedback loop to it [3], so it is impotant to undestand the quantity and distibution of atmospheic wate vapo. Method of Miao has been poved to be effective fo IWV etieval in both Antactic and patial Actic egions [4], []. Unfotunately, howeve, this method is bound to some limitations, one of them is that, with cuently available adiomete channel combinations (SSM/T2 and AMSU-B, etc.), the etievable IWV content can not exceed 6kg/m 2 [6], which is geneally satisfied within the Antactic Region. But when tying to use Method of Miao to the whole Actic Region, one poblem appeas: in the Actic Region, the IWV content is usually highe than that of Antactic Region (See Fig.1); etieval of IWV can only be done at cetain limited aeas. The most impotant eason fo this limitation is the selection of SSM/T2 and AMSU-B channel 2 Radiative Tansfe Model and Dataset 2.1 Radiative Tansfe Model The adiative tansfe model used in this pape is VDISORT [7], which is last modified by the autho in Apil 2. The pogam is witten in Fotan 9, and compiled with Compaq Visual FORTRAN 6.6 unde Windows platfom. The millimete-wave absoption module in the pogam is adapted fom Rosenkanz [8]. ISSN: 179-2769 1 ISBN: 978-96-474-3-7
Poceedings of the 4th WSEAS Intenational Confeence on REMOTE SENSING (REMOTE'8) 6 2.2 Dataset IWV () The pola egion adiosonde pofile data in this wok ae adapted fom IGRA (Integated Global Radiosonde Achive). IGRA is new adiosonde dataset fom the National Climate Data Cente (NCDC). IGRA contains quality-assued data fom 11 diffeent souces. Rigoous pocedues ae employed to ensue pope station identification, eliminate duplicate levels within soundings, and select one sounding fo evey station, date, and time [9].VDISORT equies the numbe of levels fo input atmosphee pofiles to be constant, and in the topdown ode. So we select those pofiles fom IGRA whose numbe of levels 3 and 4, sot the levels in the ight ode and tuncate the top ones, leaving a constant 2 levels pe pofile. Finally we select 1412 pofiles within the Actic Cicle (Latitude 66.) (See Fig.2) in the yea of 27, and 813 pofiles within the Antactic Cicle (Latitude -66.) in the yea of 2~27. The eo intoduced by the level tuncation pocess is expected to be small, because the wate vapo mixing atio of top levels being tuncated is. (See Fig.3) (a) 2 (b) 1 4 1 2 1 2 3 4 6 7 8 91 12 Month 1 2 3 4 6 7 8 91 12 Month Fig.1 Mean IWV monthly distibution calculated fom IGRA fo (a) Antactic Region and (b) Actic Region E27 E E18 E9 3 Methodology Fig.2 Actic weathe stations 3.1 2-channel combinations Height (Km) As stated in the intoduction, if we want to expand the etievable IWV ange using Method of Miao, possible channels should be located fathe to 183GHz absoption peak than available wate vapo pofiling channels. Hee, up to +/-37GHz unpolaized channels ae consideed (183.31 +/-1, +/-2,,+/-37GHz). The bightness tempeatues (Tb) of coesponding channels with diffeent suface emissivities ( ε =.8,.7,.7) and scan angles (scan angle = 1., 4 degees) ae calculated using the VDISORT Radiative Tansfe Model. The bightness tempeatue fo each sideband is obtained by a thee fequency point (f-1ghz, f, f+1ghz, f is the cente fequency of a sideband) aveage to simulate a 2GHz bandwidth, and the bightness tempeatue fo a double-sideband channel is obtained by the aveage of two sidebands. 3 3 2 2 2 2 1 1 1 1 1 2 2 3 Tempeatue (K).1.2 Wate vapo mixing atio (g/kg) Fig.3 Tuncated pofiles fo Actic Region (1412 pofiles) Then a subtaction between two channels fom the 37 double-sideband channels is pefomed (See Fig.4). Thee ae 666 such subtactions (select-2 combination out of 37). Two nomenclatues ae defined: ISSN: 179-2769 16 ISBN: 978-96-474-3-7
Poceedings of the 4th WSEAS Intenational Confeence on REMOTE SENSING (REMOTE'8) Fom the 2-channel combinations that meet ou citeia, a select-2 combination is pefomed again to achieve the final 4-channel combinations (1328 in all). One thing to be noted is with the two-step pocedue, 3-channel combinations used by Miao ae completely included in the moe genealized 4- channel combinations. Fig.4 a 2-channel combination: Tb(183+/-7GHz) Tb(183+/-37GHz) Fig. Oiginal channel combinations of SSM/T2 used by Miao [1] (1) Maxima (K): max bightness tempeatue diffeence between two channels, coesponding ε =.8, (2) Satuation Point (Kg/m 2 ): IWV value, whee bightness tempeatue diffeence is, coesponding ε =.8. ε =.8 is used to define Maxima and Satuation Point because the emissivity fo sea ice is close to.8 [1], [3]. Thee ae 4 2-channel combinations meet the citeia: (1) Maxima exceeding 3K and (2) Satuation Point exceeding 6Kg/m 2. The citeia ae so selected because (1) 3K assues the Miao s plot would not be too flat (See Fig.), (2) 6Kg/m 2 fits the expanded etievable ange. The citeia ae poved to be athe obust by late calculations, i.e. small change of these citeia does not alte the final optimized channel combinations. 3.2 Genealized 4-channel combinations 3.3 Retieved IWV fo Each 4-channel Combination Befoe calculating the etieved IWV, a.k noise is intoduced into each double-sideband channel. The noise is simulated by a Gaussian distibuted andom numbe with mean of., and standad deviation of.. The pofiles with IWV in the ange of 6~16Kg/m 2 fom the dataset ae selected to test the 4-channel combinations. The ange is so selected because the oiginal SSM/T2 channel combinations cove etievable ange ~6Kg/m 2, and the expanded ange is moe inteested to us. Also, Lage IWV than 16Kg/m 2 can be etieved by othe means, o even pofiling is possible. In pactice, the exact ε of the ice coveed suface is unknown, so ε =.8,.7,.7, which ae close to ice, ae used in the testing. With some compute effots, the etieved IWVs fo each 4-channel combination ae calculated; ms and bias between the actual IWV and etieved IWV ae eadily obtained. The final ms and bias ae obtained by the aveage of 1 ecalculations, esults show lage numbe of ecalculation does not impove thei accuacy. We confine the ms of the etieved IWV within 1.4 Kg/m 2, thee ae 1617 possible 4-channel combinations left (See Fig.6), the one with smallest ms among them happen to be a 3-channel combination: (183+/-37GHz 183+/-1GHz) / (183+/-37GHz 183+/-6GHz) (See Fig.8). Possible eason fo this coincidence is +/-37GHz is the most toleant channel to wate vapo, and it fits the demand of IWV ange expansion. Thee is a special 3-channel combination: (183+/-37GHz 183+/-17GHz ) / (183+/-37GHz 183+/-7GHz) (See Fig.9), which is consist of fequencies eithe aleady in use by othe adiometes (183+37 GHz = 22GHz, 183+/-7GHz, 183+17GHz 2GHz, 183-17GHz = 166GHz) o close to them (183-37GHz = 146GHz 1GHz). And as a matte of fact the 183+37GHz channel has aleady been tested by field measuement, and it shows the potential of impovement fo the integated wate vapo above ice suface [1], [11]. ISSN: 179-2769 17 ISBN: 978-96-474-3-7
Poceedings of the 4th WSEAS Intenational Confeence on REMOTE SENSING (REMOTE'8) +/ GHz 37 29 21 13 4 channel combinations The scatte plots fo etieved IWV against the actual IWV fo these two combinations ae shown in the following figues (See Fig.1, 11). 2 1.4 1.3 1.3 1.2 1.2.2 ms bias.2 1 2 4 6 8 1 12 14 1617 Fig.6 4-Channel combinations and thei ms and bias +/ GHz 37 29 21 13 1.4 1.3 1.3 1.2 1.2.2 3 channel combinations ms bias.2 1 2 4 6 8 1 12 146 Fig.7 3-channel combinations included in the 1617 possible 4-channel combinations and thei ms and bias Retieved IWV 1 1 ms=1.239 bias=.3 1 1 2 Actual IWV Fig.1 IWV Scatte Plot fo a 3-channel combination: (183+/-37GHz 183+/-1GHz) / (183+/-37GHz 183+/-6GHz) Retieved IWV 2 1 1 ms=1.317 bias=.68 1 1 2 Actual IWV Fig.11 IWV Scatte Plot fo a 3-channel combination: (183+/-37GHz 183+/-17GHz) / (183+/-37GHz 183+/-7GHz) Fig.8 Miao s Plot fo a 3-channel combination: (183+/-37GHz 183+/-1GHz) / (183+/-37GHz 183+/- 6GHz) Fig.9 Miao s Plot fo a 3-channel combination: (183+/-37GHz 183+/-17GHz) / (183+/-37GHz 183+/-7GHz) 4 Discussions The IWV etieval pefomance of 1617 possible 4- channel combinations is quite compaable (See Fig. 6, 1, 11 and Appendix), thus we popose the final optimized channel combination which has best compatibility with cuently available adiometes. And consideing the economics of spacebone platfom, the included 3-channel combinations ae ou pioity choices. The selection of optimized channel combinations is obust, i.e. in cetain fequency anges all channel fequency points ae applicable to fom the combinations, and IWV etieval pefomance of these combinations is compaable. With Scan angle = 1. degees fo Actic pofiles, the fequency ange fo channel 1 is [28, 37]; the fequency ange fo channel 2 is [14, 37]; the fequency ange fo channel 3 is [11, 21]; the fequency ange fo channel 4 is [4, 8]; (See Fig.6). ISSN: 179-2769 18 ISBN: 978-96-474-3-7
Poceedings of the 4th WSEAS Intenational Confeence on REMOTE SENSING (REMOTE'8) Calculation esults show the optimized channel combinations ae applicable to both Actic and Antactic Regions with both scan angles (1., 4 degees). And the IWV etieval pefomance unde diffeent conditions ae compaable, it suggests the optimized channel combinations ae suitable fo spacebone platfom and can be put into pactice fo IWV etieval in pola egions. othe usages and the IWV etieving pefomance, a moe optimized channel combination in geneal usage is still pending. 6 Acknowledgement The authos wish to thank the National Natual Science Foundation of China fo the suppot unde gants #447661 and #421. Conclusion and Futue Wok In this pape, optimized channel combinations fo IWV etieval in the pola egions with Method of Miao ae poposed. Optimization is mainly tageting fo the expansion of etievable IWV ange. Results show the uppe etievable IWV ange expands fom 6Kg/m 2 to 16Kg/m 2. And consideing the compatibility with cuently available adiomete channels, an optimized 3-channel combination (183+/-37GHz 183+/-17GHz) / (183+/-37GHz 183+/-7GHz) is the best tade-off choice. Thee ae othe appoaches fo the integated wate vapo etieval, like 1) using gound-based GPS eceives. K.-P. Johnsen et al. have studied the possibility fo the Antactic Region in the yea 24; the annual mean value of the CHAMP/GPS IWVs ove Antactica (between day 134 of 21 and day 134 of yea 22) is about 1.6+/-1.7kg [12]. Since we did not conside the suface emissivity vaiance in this pape, the ms hee is smalle; moe wok should be done to compae the two methods; and 2) IWV etieval done by neual net wok seems pomising [13], compaison needs to be done with moe ealistic modeling and even field measuement; 3) Gound based millimete-wave adiomety fo the measuement of low amounts of pecipitable wate vapo has the best accuacy among cuently available methods, the uncetainty the full ange of PWV expected in the Actic Region is within % [14], but it is limited by its spatial coveage and tempoal continuity. Studies show suface emissivity vaiance has consideable impact on IWV pefomance [3]. In this pape, emissivities fo diffeent channels ae deemed to be the same, moe ealistic suface model will be added in the futue. All channels being simulated ae unpolaized, polaization will be consideed in the futue. The geneal usefulness (wate vapo pofiling ability, pefomance unde cloudy conditions, offeing othe weathe poducts, etc.) of the newly poposed channels still needs to be studied, and it s possible that we have to find a compomise between 7 Appendix Some channel combinations deived in pat 3.4 and thei ms, bias (Kg/m 2 ); focal point (K); egession coefficients (Actic Region and scan angle = 1. degees) ae listed below: Combinations ms bias Focal Point C C1 #37#1 #37#6 1.24 -. (11., 2.9) 21.7 1.8 #37#17 #37#7 1.32 -.7 (8.8, 2.) 24.6 1.7 #37#17 #37#4 1.38 -.12 (16.4, 2.4) 21.7 11.8 #37#17 #37# 1.29.3 (13.3, 2.3) 22. 12.6 #37#17 #37#6 1.27 -.3 (1.8, 2.2) 23.4 13.9 #37#12 #37#7 1.37 -. (9.3, 4.4) 19.9 2. #37#13 #37#7 1.31 -.2 (9.2, 3.8) 2.9 22.2 #37#14 #37#7 1.28 -.3 (9.1, 3.2) 21.8 19.9 #37#1 #37#7 1.27 -.4 (9., 2.7) 22.8 18.2 #37#16 #37#7 1.29 -.6 (8.9, 2.4) 23.7 16.8 #37#18 #37#7 1.36 -.8 (8.8, 1.8) 2.4 14.8 Refeences: [1] R. G. Bay, M. C. Seeze, J. A. Maslanik, and R.H. Pelle, The Actic Sea Ice-Climate System: Obsevations and Modeling, Review of Geophysics, Vol. 31, NO.4, 1993, pp.397-422 [2] S. L. Mooe, Detemination of integated wate vapo ove Antactica utilizing Special Senso Micowave Tempeatue-2 (SSM/T2) data, Ph.D. thesis, Dept. Meteoology, Texas A&M Univ., College Station, TX, 1997 [3] J.-P. Blanchet, E. Giad, Wate vapotempeatue feedback in the fomation of continental Actic ai: its implication fo climate, The Science of the Total Envionment Vol.16/161, 199, pp. 793-82 [4] J. R. Wang, P. E. Racette, and M. E. Tiesky, Retieval of pecipitable wate by the millimete-wave imaging adiomete in the Actic egion duing FIRE-ACE, IEEE Tansactions Geoscience and Remote Sensing, Vol. 39 NO.3 21, pp 69-6. ISSN: 179-2769 19 ISBN: 978-96-474-3-7
Poceedings of the 4th WSEAS Intenational Confeence on REMOTE SENSING (REMOTE'8) [] N. Selbach, Detemination of column wate vapo and suface emissivity of sea ice at 89GHz, 17GHz and 183GHz in the Actic winte, Ph.D. Thesis, Dept. Physics, Univ. Bemen, Bemen, Gemany, 23 of Atmospheic and Ocean Technology, Vol.22, 2, pp. 317-337 [6] J. Miao, Retieval of atmospheic wate vapo content in pola egions using spacebone micowave adiomety, Ph.D. Thesis, Dept. Physics, Univ. Bemen, Bemen, Gemany, May 1998 [7] F. M. Schulz, K. Stamnes, F. Weng, VDISORT: An Impoved Genealized Discete Odinate Method fo polaized (Vecto) Radiative Tansfe, J. Quant. Radiat. Tansfe Vol.61, NO1, 1999, pp.1-122 [8] P. Rosenkanz, Absoption of micowaves by atmospheic gases, Atmospheic Remote Sensing by Micowave Radiomety, ch. 2. New Yok: Wiley, M. A. Janssen, Ed. 1993 [9] I. Due, R. S. Vose, and D. B. Wuetz, Oveview of the Integated Global Radiosonde Achive, Jounal of Climate: Vol. 19, NO. 1, 26, pp. 3-68 [1] J.R. Wang, and W. Manning, Retievals of Low Integated Wate Vapo Using MIR and SSM/T-2 Measuements, IEEE Tansactions on Geoscience and Remote Sensing, Vol.41 NO 3, 23, pp. 63-639 [11] J.R. Wang, P. Racette, M.E. Tiesky, and W. Manning, Retievals of Column Wate Vapo Using Millimete-wave Radiometic Measuements, IEEE Tansactions on Geoscience and Remote Sensing, Vol. 4, NO.6, 22, pp. 122-1229 [12] K.-P. Johnsen, J. Miao, S.Q. Kidde, Compaison of Atmospheic Wate Vapo ove Antactica Deived fom CHAMP/GPS and AMSU- B Data, Physics and Chemisty of the Eath Vol.29, 24, pp. 21-2 [13] P. Basili, S. Bonafoni, V. Mattioli, F. Pelliccia, A. Sepolla, E. Bocci, P. Ciotti Development of a neual netwok fo pecipitable wate vapo etieval ove ocean and land, 28 Micowave Radiomety and Remote Sensing of the Envionment, 28, pp. 1-4 [14] P. Racette et al., Measuement of Low Amounts of Pecipitable Wate Vapo Using Gound-Based Millimetewave Radiomety, Jounal ISSN: 179-2769 2 ISBN: 978-96-474-3-7