The impact of radiosonde data over the ice-free Arctic Ocean on the atmospheric circulation in the Northern Hemisphere

Transcription

1 GEOPHYSICAL RESEARCH LETTERS, VOL. 40, , doi: /grl.50207, 2013 The impact of radiosonde data over the ice-free Arctic Ocean on the atmospheric circulation in the Northern Hemisphere Jun Inoue, 1,2,3 Takeshi Enomoto, 1,4 and Masatake E. Hori 1,2 Received 14 January 2013; revised 29 January 2013; accepted 30 January 2013; published 6 March [1] We investigated the impact of radiosonde data from the ice-free Arctic Ocean obtained by the Japanese R/V Mirai during a cruise in the fall of 2010 on the AFES-LETKF experimental ensemble reanalysis version 2 (ALERA2) data set. The reanalysis used radiosonde data over the ice-free region. Compared with observations, it captured Arctic cyclogenesis along the marginal ice zone, including a tropopause fold, very well. Without the observations, a 5 K cold bias in air temperature was found, suggesting that radiosondes over the Arctic Ocean are vital for reproducing the change in tropopause variability. As a consequence of including the Arctic radiosondes, a tropopause height difference formed and persisted after cyclogenesis, increasing the subpolar jet in ALERA2 by 3% at ı N. The air temperature in the whole troposphere north of 70 ı N showed a cooling in the 2 weeks after cyclogenesis, whereas a warming was observed in the lower stratosphere, reflecting the regional impact of the intensive radiosonde observations. A remote response of the radiosondes over the Arctic Ocean to the midlatitudes was discussed by focusing on the density of observing network and seasonal march of atmospheric circulations. Our results demonstrated that the high-temporal radiosonde observations over the Arctic Ocean can help reduce uncertainty in reanalyses and numerical weather predictions throughout the northern half of the Northern Hemisphere for weeks afterwards. Citation: Inoue, J., T. Enomoto, and M. E. Hori (2013), The impact of radiosonde data over the ice-free Arctic Ocean on the atmospheric circulation in the Northern Hemisphere, Geophys. Res. Lett., 40, , doi: /grl Introduction [2] Radiosonde data over the Arctic Ocean are very limited due to the difficulty of operational observations [Andersson, 2007]. To date, only a few Arctic expeditions have provided data (e.g., the North Pole drifting camps [Kahl et al., 1999;Makshtas et al., 2007], icebreakers [Uttal et al., 2002; Tjernström et al., 2004, 2012; Lüpkes et al., 1 Earth Simulator Center, Japan Agency for Marine-Earth Science and Technology, Yokohama, Japan. 2 Research Institute for Global Change, Japan Agency for Marine-Earth Science and Technology, Yokosuka, Japan. 3 National Institute of Polar Research, Tachikawa, Japan. 4 Disaster Prevention Research Institute, Kyoto University, Uji, Japan. Corresponding author: Jun Inoue, National Institute of Polar Research, 10-3 Midori-cho, Tachikawa, Tokyo , Japan. (inoue.jun@nipr.ac.jp) American Geophysical Union. All Rights Reserved /13/ /grl ], and non-icebreakers [Inoue et al., 2011]). The number of radiosondes launched has been very limited; however, the impact of Arctic radiosonde observations on weather forecasts and reanalysis data has not been fully investigated. [3] The impact of surface observation over the Arctic Ocean on reanalyses has been investigated through experimental ensemble reanalysis and the use of Arctic drifting buoys, which send sea-level pressure data through the Global Telecommunication System (GTS) [Inoue et al., 2009]. As expected, the impact of surface data on the atmospheric circulation is limited to the lower troposphere over the Arctic Ocean. Tjernström and Graversen [2009] confirmed that the SHEBA soundings [Uttal et al., 2002] improve the ERA-40 reanalysis in the SHEBA region. Considering the very limited tropospheric data over the Arctic Ocean, uncertainty regarding atmospheric circulation in mid- and upper troposphere still remains. Although the Northern Sea Route is opening to a greater extent each year due to sea-ice retreat, cyclogenesis along the marginal ice zone has also increased, as described by Inoue and Hori [2011]. To avoid high waves, icing, and sea-ice striking merchant ships, improvements in numerical weather prediction for these extraordinary atmospheric and oceanic situations are urgently required. [4] During the fall of 2010, we made a series of intensive radiosonde observations over the ice-free portion of the Arctic Ocean using the Japanese R/V Mirai. As reported by Inoue and Hori [2011], an upper positive potential vorticity (PV) anomaly was essential for rapid cyclogenesis, which was related to the onset of freezing in that year. Therefore, a good reproduction of upper atmospheric circulation is very important to improve weather prediction. Using the shipboard radiosonde observations and experimental reanalysis, we investigated the impact of the radiosonde data over the Arctic Ocean on the atmospheric circulation over the Northern Hemisphere. 2. Description of Data 2.1. Radiosondes over the Ice-free Arctic Ocean in 2010 [5] The R/V Mirai conducted an Arctic cruise in the Chukchi and Beaufort Seas from 2 September to 16 October During the cruise, 3- or 6-hourly observations were made with a Vaisala RS92-SGPD radiosonde (red dots in Figure 1). All of the radiosonde data were sent to the GTS, likely improving the accuracy of reanalyses over the Arctic Ocean when compared with other years. One of the meteorological outcomes from the cruise was the successful observation of an Arctic cyclone [Inoue and Hori, 2011]. Although this event was triggered by baroclinicity in the lower troposphere near the marginal ice zone, the role of an upper potential vorticity (PV) intrusion was also very 864

2 Figure 1. (a) Sea-ice concentration (%) derived from the Advanced Microwave Scanning Radiometer for Earth Observing System (AMSR-E) on 25 September 2010 with the track of the R/V Mirai (gray line) and radiosonde stations (red dots). Sea-level pressure analysis CTL is shown as contours (hpa). (b) NOAA/AVHRR infrared image at 2341 UTC, 24 September 2010, with temperature difference between analysis CTL and OSE at 300 hpa shown as red isotherms ( ı C). important for its rapid development. Figure 2a shows the time-height cross-section of the air temperature observed by radiosondes launched from the R/V Mirai. Before24 September, tropospheric temperatures were high. After the positive PV anomaly, indicated in Figure 2a by a region of cold air near the tropopause, passed over the R/V Mirai on September, colder temperatures persisted throughout the troposphere. This event marked the onset of freezing for this year. Therefore, the period during the cyclogenesis is considered suitable for an observing-system experiment (OSE) ALERA2 [6] The ALERA2 is an experimental reanalysis (version 2) produced by an ensemble data-assimilation system composed of the Atmospheric general circulation model for the Earth Simulator (AFES, [Ohfuchi et al., 2004]) version 3.6 [Kuwano-Yoshida et al., 2010] and the fourdimensional local ensemble transform Kalman filter (4D- LETKF) [Miyoshi and Yamane, 2007] with covariance localization by distance [Miyoshi et al., 2007]. The ALERA2 data set was generated by an analysis cycle using a 63 member ensemble forecast by AFES at a T119L48 (1 ı in the horizontal and 48 vertical levels) resolution for the period from August 2010 to the present day. A detailed description is presented by Enomoto et al. [2013]. [7] PREPBUFR data sets compiled by the USA National Centers for Environmental Prediction (NCEP) and archived at the University Corporation for Atmospheric Research (UCAR) are used as the source of observational data for ALERA2. For weather balloons (referred to as ADPUPA in NCEP system), the data at 1000, 925, 850, 700, 500, 400, 300, 250, 200, 150, 100, 70, 50, and 10 hpa are used. Reports from aircraft and satellite retrievals are reduced to every fourth value and wind profilers to every third level [St-James and Laroch, 2005]. National Oceanic and Atmospheric Administration (NOAA) daily 1/4 ı Optimum Interpolation Sea Surface Temperature (OISST) version 2 [Reynolds et al., 2007] is used to provide the ocean and ice boundary conditions. The initial conditions are prepared by integration with time-varying prescribed ocean boundary conditions for 20 years. The integration has been performed with AFES at the same resolution (T119L48, 1 ı in the horizontal and 48 vertical levels). Ensemble members are arbitrarily chosen to be the atmospheric states for a particular date and nearby dates from different years. [8] In this study, the ALERA2 is regarded as the control analysis (CTL). In the observing-system experiment (OSE), the same observational data set, except for the radiosonde data obtained by the R/V Mirai north of 70 ı N, is assimilated using the same system. These products include the analysis ensemble mean and analysis ensemble spread of the wind, temperature, specific humidity, geopotential height, vertical velocity at 18 pressure levels between 10 and 1000 hpa, and sea-level pressure. The study period was from 2 September 2010 to 28 October 2010, which covers the duration of the R/V Mirai s fall cruise period. 3. Results 3.1. Local Effects [9] To compare the CTL results with the radiosonde observations, the time-height cross-section of the air temperature at the closest grid to the ship position is shown in Figure 2b. When compared with the observation (Figure 2a), the warm period until 24 September and the cold period in the troposphere after the cyclogenesis were well reproduced in the CTL. Hereafter, we will focus on the difference field between the CTL and the OSE (CTL OSE) to investigate the impact of radiosonde data on the reproduction of atmospheric circulation. Figure 2c shows the temperature difference (CTL OSE) in ensemble means (shading) and the ensemble spread (contour). A significant warming was observed at 300 hpa on 25 September, corresponding to the cyclone event. There is a wide spatial distribution of the warming (exceeding 2 K) with a diameter of 1000 km, which is the same size as the cyclone (Figure 1b). The temperature in the CTL above the tropopause is colder than in OSE when warm events in the stratosphere occurred on 25 September. The difference in the ensemble spread of the air temperature was large in the lower stratosphere during this period (contour in Figure 2c). [10] We also compared the profiles on 25 September to investigate the detailed temperature structures (Figure 2f). The temperature profile in the CTL (red line) between 250 and 400 hpa is relatively close to the observations (black line) and significantly warmer than the OSE (blue line). The 865

3 Figure 2. Time-height cross-sections of air temperature by (a) radiosonde observations and (b) CTL at the closest grid to the radiosonde stations ( ı C). (c) Temperature difference between the CTL and OSE for the ensemble mean (shading) and ensemble spread (contour: enclosed area exceeds 0.5 K), (d) surface heat flux difference (the sum of sensible and latent heat fluxes: W m 2 ) over the Chukchi Sea ( ı W, ı N), and (e) tropopause height difference between the CTL and OSE (m). (f) Air temperature profiles at 1800 UTC on 25 September 2010 by CTL, OSE, and observation. Thin lines are ensemble mean first guesses. difference is more than 5 K in the upper troposphere around 300 hpa. Because the temperature difference between the CTL and OSE at 300 hpa seems to come from the difference in tropopause height, we estimated the tropopause height to be the 3.5 potential vorticity unit (PVU) level [Hoinka, 1998]. The time series of the tropopause height difference is shown in Figure 2e. After 24 September, the height of the tropopause in the CTL was 300 m lower than in the OSE, and this tendency frequently occurred until mid- October, suggesting that the observations after cyclogenesis (Figure 1) contributed to a significant modification of the polar vortex. [11] In the lower troposphere, the observations indicated a transition to a colder air mass at the time of the positive upper PV intrusion (Figures 2a and 2b). CTL is colder than OSE during the latter half of the analysis period, although soundings after this transition show that both the CTL and OSE analyses are colder than the observations (e.g., Figure 2f), which is likely a typical bias over the marginal ice zone arising from the sea-ice parameterization [e.g., Inoue et al., 2011]. This cooling in CTL in the lower troposphere should modify the surface heat fluxes. The difference of the sum of turbulent heat fluxes over the Chukchi Sea (150 ı W 180 ı W, 70 ı N 75 ı N) exceeded 15 Wm 2 on 25 September (Figure 2d), likely enhancing sea-surface cooling. The difference in time averaged heat flux from 25 September to 13 October over the same area was 5.4 W m 2, which corresponds to 7% of the mean surface heat flux in the CTL (77.3 W m 2 ). Because this heat flux difference persisted until the end of the cruise, the reproduction of the vertical structure of the Arctic cyclone from the surface to the stratosphere is very important for understanding the air-sea coupling system during the transitional season, including the onset of freezing. [12] The other interesting feature is that the difference between the first guess (thin lines in Figure 2f) and the analysis field (thick lines in Figure 2f) is very small for both CTL and OSE. These small increments suggest a cumulative effect of the 20 days of soundings prior to 25 September. Specifically, in CTL, the atmospheric circulation at high latitudes was continuously modified during the cruise by our frequent observations from early September, whereas in OSE, the increments were necessarily small because the number of observations sent to the GTS from the Arctic Ocean was originally small Remote Effects [13] To determine the impact of radiosonde observations on reproductions of atmospheric circulation in the midlatitudes, we computed zonally averaged fields focusing on the period after cyclogenesis until 13 October (Figure 3). The tropopause height difference in CTL was negative north of 866

4 Figure 3. Difference fields between the CTL and OSE for zonally averaged (a) tropopause height (m), (b) zonal mean air temperature (shading; K), and (c) zonal wind (shading; m s 1 ). The period averaged is from 24 September to 13 October Isopleths and dots in Figures 3b and 3c denote zonal wind in the CTL and the tropopause defined as 3.5 PVU, respectively. The observational zone is depicted by magenta dashed lines. Figure 4. (a) Temperature difference between the CTL and OSE at 250 hpa ( ı C) and 2 week forward trajectories in the CTL at 6-hourly intervals between 24 September and 13 October 2010 at the R/V Mirai radiosonde stations (gray lines). Operational radiosonde stations north of 20 ı N are depicted by dots (black: twice or more daily; magenta: once daily). (b) Seasonal march of air temperature at 250 hpa in CTL (October September). 867

5 65 ı N, although the observation area was limited to between 70 ı N and 80 ı N (Figure 3a). With regard to the air temperature, a cooling ( 0.2 K) was identified in all of the troposphere north of 65 ı N (Figure 3b), corresponding to the tropopause height difference. The gradients of the anomalies in tropopause height and air temperature were large between 55 ı N and 75 ı N, likely enhancing the subpolar jet ( 0.2 m s 1 ) (Figure 3c). The zonal component of the thermal wind difference ( U T ) in the upper troposphere between 55 ı N and 75 ı N was roughly estimated as U T = (T R d /f CTL T OSE) 0 ln y P 1000 P 300.HereR d (=287 JK 1 kg 1 )isthe gas constant for dry air; f 0 (= s 1 ) is the Coriolis parameter at 65 ı N; T CTL T OSE ( 0.2 K) is the mean temperature difference in the troposphere between the CTL and OSE; y (= m) is the meridional distance between 55 ı N and 75 ı N, and P level is the height at each level (hpa). U T is estimated as 0.24 m s 1, suggesting that zonal winds are accelerated by the cooling identified from our observations. This value agrees with Figure 3c (shading). Although this case is much exaggerated because Arctic cyclogenesis is targeted, the impact of remote radiosonde observations on our understanding of atmospheric circulation in the Northern Hemisphere is likely related to improved reproduction of the polar vortex in the CTL. 4. Discussion [14] With regard to remote effects, the other interesting feature is a cooling ( 0.2 K) in the midlatitudes between 40 ı Nand50 ı N at the 250 hpa level in CTL (Figure 3b). Considering no clear difference in the ensemble increment field at the midlatitudes ( ı C: figure not shown), where there are many regular radiosonde stations over Eurasia and North America, this small amount of cooling might be related to a model bias in ALERA2 (e.g., weighting of observations versus the first guess). However, we tried to interpret this cooling from the viewpoint of impacts of additional radiosonde observations over the Arctic Ocean because the air masses presumably travel from the Arctic to the midlatitudes. [15] Here, we calculated the 2 week forward trajectories at the 250 hpa level every 6 h starting from the position of the ship (Figure 4a). The analysis period was from 24 September (cyclogenesis) to 13 October (the end of radiosonde observations from the ship). Most of the trajectories spread from the Arctic Ocean to the midlatitudes. The cold anomalies were accompanied by air masses crossing the Eurasian continent, where the number of regular sounding stations is relatively small compared with Europe and eastern Asia. It should also be noted that the frequency of observations in eastern Europe and southern Russia is not always twice daily (magenta dots in Figure 4a). Considering that cooling appeared in the downstream region of the data-sparse area (e.g., eastern Europe and Mongolia), the information from our 3- or 6-hourly intense observations over the Arctic Ocean appears to have spread to the midlatitudes, passing through the continental sounding network. [16] However, other factors should contribute to this cooling, in particular over North America where the observing network is dense and uniform. Here, we focus on the seasonal march of air temperature at 250 hpa in CTL from September to October (i.e., the monthly mean field of October September) (Figure 4b). The seasonal cooling is strongest over North America and Eurasia, which is a similar pattern to the cooling at midlatitudes in Figure 4a due to the inclusion of Arctic radiosondes. Considering that the tropopause height climatologically decreases from September to October over the Arctic region, the reduction of the tropopause height in CTL (Figure 3a) is a characteristic of the seasonal march of the atmospheric circulation. Because the rawinsondes reduced the tropopause height on numerous occasions during the latter half of the analysis period (Figure 2e) and the Arctic cyclone event on 24 September marked the onset of freezing for this year, our high-temporal observation after the event likely contributed to improve the evolution of the atmospheric structure during the transitional season. 5. Concluding Remarks [17] An improved weather forecasting capacity over the ice-free Arctic Ocean is vital for safe ship navigation in the Northern Sea Route and Northwest Passage because storms can generate strong winds, high waves, icing on the ship surface, and sea-ice advection. A precise prediction depends on not only a sophisticated model itself but also in situ observations. Buoys and reports from ships are good data sources; however, their impact is limited to the lower Arctic troposphere [Inoue et al., 2009]. In this study, we showed that analysis using 40 days of high-temporal resolution soundings over a portion of the ice-free Arctic Ocean demonstrate their impact on analyses, not just at the local observation site and time but throughout the northern half of the Northern Hemisphere for weeks afterwards. [18] Although yearlong operational soundings over the Arctic Ocean are difficult to conduct politically and logistically, the frequency of observations at the coastal stations (e.g., Russia, Canada, etc.) can possibly be increased. Because the optimization of the Arctic observation network (the number of stations and frequency) has not been fully assessed, a yearlong special field research program under an international framework is urgently required in the near future. [19] Acknowledgments. We are greatly indebted to K. Sato, S. Okumura, S. Sueyoshi, N. Nagahama, A. Doi, and W. Tokunaga for conducting radiosonde observations. The authors would like to thank the crew of the R/V Mirai. Discussion with members of OREDA team was very helpful. Constructive comments from reviewers were also very useful. This work was partly supported by KAKENHI(A) References Andersson, E. (2007), Data assimilation in the polar regions, ECMWF Newsletter, 112, Enomoto, T., T. Miyoshi, Q. Moteki, J. Inoue, M. Hattori, A. Kuwano- Yoshida, N. Komori, and S. Yamane (2013), Observing-system research and ensemble data assimilation at JAMSTEC, in Data Assimilation for Atmospheric, Oceanic and Hydrologic Applications II, edited by Park, S. K., and L. Xu, Springer. in press. Hoinka, K. P. (1998), Statistics of the global tropopause pressure, Mon. Wea. Rev., 126, Inoue, J., and M. E. Hori (2011), Arctic cyclogenesis at the marginal ice zone: A contributory mechanism for the temperature amplification? Geophys. Res. Lett., 38, L12502, doi: /2011gl Inoue, J., T. Enomoto, T. Miyoshi, and S. Yamane (2009), Impact of observations from Arctic drifting buoys on the reanalysis of surface fields, Geophys. Res. Lett., 36, L08501, doi: /2008gl

6 Inoue, J., M. E. Hori, T. Enomoto, and T. Kikuchi (2011), Intercomparison of surface heat transfer near the Arctic marginal ice zone for multiple reanalyses: A case study of September 2009, SOLA, 7, , doi: /sola Kahl, J. D., N. A. Zaitseva, V. Zaitseva, V. Khattatov, R. C. Schnell, D. M. Bacon, J. Bacon, V. Radionov, and M. C. Serreze (1999), Radiosonde observations from the former Soviet North Pole series of drifting ice stations, , Bull. Am. Meteorol. Soc., 80, Kuwano-Yoshida, A., T. Enomoto, and W. Ohfuchi (2010), An improved PDF cloud scheme for climate simulations, Quart. J. Roy. Meteor. Soc., 136, , doi: /qj.660. Lüpkes, C, T. Vihma, E. Jakobson, G. König-Langlo, and A. Tetzlaff (2010), Meteorological observations from ship cruises during summer to the central Arctic: A comparison with reanalysis data, Geophys. Res. Lett., 37, L09810, doi: /2010gl Makshtas, A., D. Atkinson, M. Kulakov, S. Shutilin, R. Krishfield, and A. Proshutinsky (2007), Atmospheric forcing validation for modeling the central Arctic, Geophys. Res. Lett., 34, L20706, doi: /2007GL Miyoshi, T., and S. Yamane (2007), Local ensemble transform Kalman filtering with an AGCM at a T159/L48, Mon. Wea. Rev., 135, , doi: /2007mwr Miyoshi, T., S. Yamane, and T. Enomoto (2007), Localizing the error covariance by physical distances within a Local Ensemble Transform Kalman Filter (LETKF), SOLA, 3, 89 92, doi: /sola Ohfuchi, W., H. Nakamura, M. K. Yoshioka, T. Enomoto, K. Tanaka, X. Peng, S. Yamane, T. Nishimura, Y. Kurihara, and K. Ninomiya (2004), 10-km mesh meso-scale resolving simulations of the global atmosphere on the Earth Simulator: Preliminary outcomes of AFES (AGCM for the Earth Simulator), J. Earth Simulator, 1, Reynolds, R. W., L. Chunying, T. M. Smith, D. B. Chelton, M. G. Schlax, and K. S. Casey (2007), Daily high-resolution-blended analyses for sea surface temperature, J. Climate, 20, , doi: /2007JCLI St-James, J. S., and S. Laroch (2005), Assimilation of wind profiler data in the Canadian Meteorological Centre s analysis systems, J. Atmos. Oceanic Technol., 22, , doi: JTECH Tjernström, M., C. Leck, P. O. Persson, M. L. Jensen, S. P. Oncley, and A. Targino (2004), The summertime Arctic atmosphere: Meteorological measurements during the Arctic Ocean Experiment 2001, Bull. Am. Meteorol. Soc., 85, , doi: /bams Tjernström, M., and R. G. Graversen (2009), The vertical structure of the lower Arctic troposphere analysed from observations and the ERA-40 reanalysis, Quart. J. Roy. Meteorol. Soc., 135, , doi: /qj.380. Tjernström, M., et al. (2012), Central Arctic atmospheric summer conditions during the Arctic Summer Cloud Ocean Study (ASCOS): Contrasting to previous expeditions, Atmos. Chem. Phys. Discuss., 12, , doi: /acpd Uttal, T., et al. (2002), Surface heat budget of the arctic ocean, Bull. Am. Meteorol. Soc., 83(2), , doi: 869