Advanced Search
Article Contents

Assimilating All-sky Infrared Radiances from Himawari-8 Using the 3DVar Method for the Prediction of a Severe Storm over North China


doi: 10.1007/s00376-020-0219-z

  • Although radar observations capture storm structures with high spatiotemporal resolutions, they are limited within the storm region after the precipitation formed. Geostationary satellites data cover the gaps in the radar network prior to the formation of the precipitation for the storms and their environment. The study explores the effects of assimilating the water vapor channel radiances from Himawari-8 data with Weather Research and Forecasting model data assimilation system (WRFDA) for a severe storm case over north China. A fast cloud detection scheme for Advanced Himawari imager (AHI) radiance is enhanced in the framework of the WRFDA system initially in this study. The bias corrections, the cloud detection for the clear-sky AHI radiance, and the observation error modeling for cloudy radiance are conducted before the data assimilation. All AHI radiance observations are fully applied without any quality control for all-sky AHI radiance data assimilation. Results show that the simulated all-sky AHI radiance fits the observations better by using the cloud dependent observation error model, further improving the cloud heights. The all-sky AHI radiance assimilation adjusts all types of hydrometeor variables, especially cloud water and precipitation snow. It is proven that assimilating all-sky AHI data improves hydrometeor specifications when verified against the radar reflectivity. Consequently, the assimilation of AHI observations under the all-sky condition has an overall improved impact on both the precipitation locations and intensity compared to the experiment with only conventional and AHI clear-sky radiance data.
    摘要: 尽管高时空分辨率的雷达资料能精准地探测出风暴系统的三维结构,但是其有效观测半径有限并且观测时段局限于降水发生之后。作为雷达资料的有益补充,静止卫星辐射亮温观测是能够提供快速演变的对流天气系统结构和环境场信息的主要资料来源。本文以北方的一次强降水过程为例,在WRFDA (Weather Research and Forecasting model data assimilation system)同化系统中实现了Himawari-8 红外辐射率资料(Advanced Himawari Imager,简称AHI)的全空同化。完成了偏差订正模型、晴空条件下的云检测模型以及全空条件下的对称观测误差模型的构建。结果显示,使用云依赖的观测误差模型能改进云雨区AHI卫星资料的有效同化,改善了云量和云高的分析。通过和卫星观测和雷达观测相对比,可以发现,AHI辐射率资料的全空同化相对于晴空同化能够改进大部分地区云雨变量特别是雪水和云中液态水。进一步通过对比各组试验的降水结果可以发现,AHI辐射率资料的全空同化改进本次强降水过程的落区和强度预报效果最为显著。
  • 加载中
  • Figure 1.  Wind field (vector; m s−1) and geopotential height (shaded; gpm) of the ECMWF analyses (ECA) at 500 hPa valid at (a) 0000 UTC 6 July, (b) 0000 UTC 8 July 2018, Wind field (vector; m s−1) and relative humidity (shaded; %) of the ECA at 850 hPa valid at (c) 0000 UTC 8 July 2018, (d) 0000 UTC 9 July 2018.

    Figure 2.  (a) The WRF model domains of D01 and D02, (b) The distribution of the conventional observations valid at0000 UTC 8 July 2018. The numbers of each observation are marked on the right

    Figure 3.  The total observation counts before the quality control (black), over land before the quality control (green), over sea before the quality control (blue) and the total counts after the quality control (red) for channel 10.

    Figure 4.  The mean bias and standard deviation of channel 9 for OMB with no bias correction (OMB-nb) and with bias correction (OMB-wb) from 0000 UTC 6 July 2018 to 0000 UTC 10 July 2018.

    Figure 5.  The OMB distributions normalized by the standard deviation of all the samples (blue curves) or normalized by the binned standard deviation with different ranges of cloud amount (green curves), and the gaussian curves (red curves) for (a) channel 8 (wavelength: 6.2 μm), (b) channel 9 (wavelength: 6.9 μm) and (c) channel 10 (wavelength: 7.3 μm).

    Figure 6.  (a) The observed brightness temperature, the simulated brightness temperature (b) from the background, (c) from the analysis of the all-sky radiance data assimilation, and (d) the difference between the simulated radiance from the analysis and the background (units: K) valid at 0000 UTC on 8 July 2018 for channel 10.

    Figure 7.  Cloud height (units: gpm) for (a) AHI-level2, (b) conv+ahiclr, (c) conv+ahicloud 0000 UTC on 8 July 2018.

    Figure 8.  (a, b) The water vapor, (c, d) the cloud water, and (e, f) the cloud rain, (g, h) the precipitation snow water path at 0000 UTC 9 July 2018 for conv+ahiclr (left panel), and conv+ahicloud (right panel). The locations of Beijing (BJ) and Heibei (HB) province are marked in Fig. 8c.

    Figure 9.  The water vapor flux (shaded; g cm−1 hPa−1 s−1) difference between analyses of conv+ahicloud and conv+ahiclr at 850 hPa at (a) 0000 UTC 7 July 2018, (b) 0000 UTC 8 July 2018. The vectors show the direction and magnitude of the wind from ECMWF analysis. The locations of Beijing (BJ) and Heibei (HB) province are marked in Fig. 9a.

    Figure 10.  (a) The counts of radar observations, (b) bias, and (c) the RMSE against the radar reflectivity for the 12 h forecasts valid from 1200 UTC 6 July 2018 to 0000 UTC 10 July 2018.

    Figure 11.  The RMSE against the observed brightness temperature with different forecast leading time initialized from (a) 0000 UTC on 6 July 2018, (b) 0000 UTC on 7 July 2018, (c) 0000 UTC on 8 July 2018 (d) (a) 0000 UTC on 9 July 2018 for channel 13.

    Figure 12.  24 h accumulated precipitation (units: mm) for (a) precipitation data from CMA, (b) conv+ahiclr experiment, and (c) conv+ahicloud experiment from 1200 UTC 8 July 2018 to 1200 UTC 9 July 2018.

    Figure 13.  FSS (a) with a threshold of 5 mm h−1 as a function of forecast lead time (b) of the 24 h accumulated precipitation for conv+ahiclr and conv+ahicloud experiments averaged from 0000 UTC 7 July 2018 to 1200 UTC 9 July 2018 every 12 hour.

  • Auligné, T., A. P. McNally, and D. P. Dee, 2007: Adaptive bias correction for satellite data in a numerical weather prediction system. Quart. J. Roy. Meteor. Soc., 133, 631−642, https://doi.org/10.1002/qj.56.
    Barker, D., and Coauthors, 2012: The weather research and forecasting model’s community variational/ensemble data assimilation system: WRFDA. Bull. Amer. Meteor. Soc., 93, 831−843, https://doi.org/10.1175/BAMS-D-11-00167.1.
    Barker, D. M., W. Huang, Y. R. Guo, A. J. Bourgeois, and Q. N. Xiao, 2004: A three-dimensional variational data assimilation system for mm5: Implementation and initial results. Mon. Wea. Rev., 132, 897−914, https://doi.org/10.1175/1520-0493(2004)132<0897:ATVDAS>2.0.CO;2.
    Bessho, K., and Coauthors, 2016: An introduction to Himawari-8/9- Japan’s new-generation geostationary meteorological satellites. J. Meteor. Soc. Japan, 94, 151−183, https://doi.org/10.2151/jmsj.2016-009.
    Buehner, M., J. Morneau, and C. Charette, 2013: Four-dimensional ensemble-variational data assimilation for global deterministic weather prediction. Nonlinear Processes in Geophysics, 20, 669−682, https://doi.org/10.5194/npg-20-669-2013.
    Chen, Y. D., X. Xia, J. Z. Min, X.-Y. Huang, and S. R. H. Rizvi, 2016: Balance characteristics of multivariate background error covariance for rainy and dry seasons and their impact on precipitation forecasts of two rainfall events. Meteorol. Atmos. Phys., 128, 579−600, https://doi.org/10.1007/s00703-016-0434-4.
    Chen, Y. D., J. Wang, Y. F. Gao, X. M. Chen, H. L. Wang, and X.-Y. Huang, 2018: Refinement of the use of inhomogeneous background error covariance estimated from historical forecast error samples and its impact on short-term regional numerical weather prediction. J. Meteor. Soc. Japan, 96, 429−446, https://doi.org/10.2151/jmsj.2018-048.
    Chevallier, F., and G. Kelly, 2002: Model clouds as seen from space: Comparison with geostationary imagery in the 11-μm window channel. Mon. Wea. Rev., 130, 712−722, https://doi.org/10.1175/1520-0493(2002)130<0712:MCASFS>2.0.CO;2.
    Chevallier, F., P. Bauer, G. Kelly, C. Jakob, and T. McNally, 2001: Model clouds over oceans as seen from space: Comparison with HIRS/2 and MSU radiances. J. Climate, 14, 4216−4229, https://doi.org/10.1175/1520-0442(2001)014<4216:MCOOAS>2.0.CO;2.
    Dee, D. P., and S. Uppala, 2009: Variational bias correction of satellite radiance data in the ERA-interim reanalysis. Quart. J. Roy. Meteor. Soc., 135, 1830−1841, https://doi.org/10.1002/qj.493.
    Descombes, G., T. Auligné, F. Vandenberghe, D. Barker, and J. Barré, 2014: Generalized background error covariance matrix model (GEN_BE v2.0). Geoscientific Model Development Discussions, 7, 4291−4352, https://doi.org/10.5194/gmdd-7-4291-2014.
    Garcia-Reynoso, A., and M. A. Mora-Ramirez, 2017: Implementation of the unified post processor (UPP) and the model evaluation tools (MET) for WRF-CHEM evaluation performance. Atmósfera, 30, 259−273, https://doi.org/10.20937/atm.2017.30.03.06.
    Geer, A. J., and P. Bauer, 2011: Observation errors in all-sky data assimilation. Quart. J. Roy. Meteor. Soc., 137, 2024−2037, https://doi.org/10.1002/qj.830.
    Geer, A. J., S. Migliorini, and M. Matricardi, 2019: All-sky assimilation of infrared radiances sensitive to mid- and upper-tropospheric moisture and cloud. Atmospheric Measurement Techniques, 12, 4903−4929, https://doi.org/10.5194/amt-12-4903-2019.
    Geer, A. J., and Coauthors, 2017: The growing impact of satellite observations sensitive to humidity, cloud and precipitation. Quart. J. Roy. Meteor. Soc., 143, 3189−3206, https://doi.org/10.1002/qj.3172.
    Harnisch, F., M. Weissmann, and Á. Periáñez, 2016: Error model for the assimilation of cloud-affected infrared satellite observations in an ensemble data assimilation system. Quart. J. Roy. Meteor. Soc., 142, 1797−1808, https://doi.org/10.1002/qj.2776.
    Honda, T., S. Takino, and T. Miyoshi, 2019: Improving a precipitation forecast by assimilating all-sky himawari-8 satellite radiances: A case of Typhoon Malakas (2016). SOLA, 15, 7−11, https://doi.org/10.2151/sola.2019-002.
    Honda, T., S. Kotsuki, G. Y. Lien, Y. Maejima, K. Okamoto, and T. Miyoshi, 2018a: Assimilation of himawari-8 all-sky radiances every 10 minutes: Impact on precipitation and flood risk prediction. J. Geophys. Res., 123, 965−976, https://doi.org/10.1002/2017JD027096.
    Honda, T., and Coauthors, 2018b: Assimilating all-sky Himawari-8 satellite infrared radiances: A case of Typhoon Soudelor (2015). Mon. Wea. Rev., 146, 213−229, https://doi.org/10.1175/MWR-D-16-0357.1.
    Hong, S.-Y., Y., Noh, and J. Dudhia, 2006: A new vertical diffusion package with an explicit treatment of entrainment processes. Mon. Wea. Rev., 134, 2318−2341, https://doi.org/10.1175/MWR3199.1.
    Hunt, B. R., E. J. Kostelich, and I. Szunyogh, 2007: Efficient data assimilation for spatiotemporal chaos: A local ensemble transform Kalman filter. Physica D: Nonlinear Phenomena, 230, 112−126, https://doi.org/10.1016/j.physd.2006.11.008.
    Kleist, D. T., and K. Ide, 2015: An OSSE-based evaluation of hybrid variational-ensemble data assimilation for the NCEP GFS. Part II: 4DEnVar and hybrid variants. Mon. Wea. Rev., 143, 452−470, https://doi.org/10.1175/MWR-D-13-00350.1.
    Köpken, C., G. Kelly, and J.-N. Thépaut, 2004: Assimilation of Meteosat radiance data within the 4D-var system at ECMWF: Assimilation experiments and forecast impact. Quart. J. Roy. Meteor. Soc., 130, 2277−2292, https://doi.org/10.1256/qj.02.230.
    Li, J., P. Wang, H. Han, J. L. Li, and J. Zheng, 2016b: On the assimilation of satellite sounder data in cloudy skies in numerical weather prediction models. J. Meteorol. Res., 30, 169−182, https://doi.org/10.1007/s13351-016-5114-2.
    Li, X., M. J. Zeng, Y. Wang, W. L. Wang, H. Y. Wu, and H. X. Mei, 2016a: Evaluation of two momentum control variable schemes and their impact on the variational assimilation of radarwind data: Case study of a squall line. Adv. Atmos. Sci., 33, 1143−1157, https://doi.org/10.1007/s00376-016-5255-3.
    Liu, C. S., and Q. N. Xiao, 2013: An ensemble-based four-dimensional variational data assimilation scheme. Part III: Antarctic applications with Advanced Research WRF using real data. Mon. Wea. Rev., 141, 2721−2739, https://doi.org/10.1175/MWR-D-12-00130.1.
    Liu, Q. H., and F. Z. Weng, 2006: Advanced doubling-adding method for radiative transfer in planetary atmospheres. J. Atmos. Sci., 63, 3459−3465, https://doi.org/10.1175/JAS3808.1.
    Liu, Z. Q., C. S. Schwartz, C. Snyder, and S.-Y. Ha, 2012: Impact of assimilating AMSU-A radiances on forecasts of 2008 Atlantic tropical cyclones initialized with a Limited-area Ensemble Kalman Filter. Mon. Wea. Rev., 140, 4017−4034, https://doi.org/10.1175/MWR-D-12-00083.1.
    Lorenc, A. C., N. E. Bowler, A. M. Clayton, S. R. Pring, and D. Fairbairn, 2014: Comparison of hybrid-4DEnVar and Hybrid-4DVar Data assimilation methods for global NWP. Mon. Wea. Rev., 143, 212−229, https://doi.org/10.1175/MWR-D-14-00195.1.
    Lu, J. Z., T. Feng, J. Li, Z. L. Cai, X. J. Xu, L. Li, and J. L. Li, 2019: Impact of assimilating Himawari-8-derived layered precipitable water with varying cumulus and microphysics parameterization schemes on the simulation of Typhoon Hato. J. Geophys. Res., 124, 3050−3071, https://doi.org/10.1029/2018JD029364.
    McNally, A. P., J. C. Derber, W.-S. Wu,, and B. B. Katz, 2000: The use of TOVS level-1B radiances in the NCEP SSI analysis system. Quart. J. Roy. Meteor. Soc., 126, 689−724, https://doi.org/10.1002/qj.49712656315.
    Minamide, M., and F. Q. Zhang, 2017: Adaptive observation error inflation for assimilating all-sky satellite radiance. Mon. Wea. Rev., 145, 1063−1081, https://doi.org/10.1175/MWR-D-16-0257.1.
    Minamide, M., and F. Q. Zhang, 2018: Assimilation of all-sky infrared radiances from Himawari-8 and impacts of moisture and hydrometer initialization on convection-permitting tropical cyclone prediction. Mon. Wea. Rev., 146, 3241−3258, https://doi.org/10.1175/MWR-D-17-0367.1.
    Mlawer, E. J., S. J. Taubman, P. D. Brown, M. J. Iacono, and S. A. Clough, 1997: Radiative transfer for inhomogeneous atmospheres: RRTM, a validated correlated-k model for the longwave. J. Geophys. Res., 102, 16 663−16 682, https://doi.org/10.1029/97JD00237.
    Okamoto, K., 2017: Evaluation of IR radiance simulation for all-sky assimilation of himawari-8/AHI in a mesoscale NWP system. Quart. J. Roy. Meteor. Soc., 143, 1517−1527, https://doi.org/10.1002/qj.3022.
    Okamoto, K., Y. Sawada, and M. Kunii, 2019: Comparison of assimilating all-sky and clear-sky infrared radiances from Himawari-8 in a mesoscale system. Quart. J. Roy. Meteor. Soc., 145, 745−766, https://doi.org/10.1002/qj.3463.
    Otkin, J. A., R. Potthast, and A. S. Lawless, 2018: Nonlinear bias correction for satellite data assimilation using Taylor series polynomials. Mon. Wea. Rev., 146, 263−285, https://doi.org/10.1175/MWR-D-17-0171.1.
    Parrish, D. F., and J. C. Derber, 1992: The National Meteorological Center's spectral statistical-interpolation analysis system. Mon. Wea. Rev., 120, 1747−1763, https://doi.org/10.1175/1520-0493(1992)120<1747:TNMCSS>2.0.CO;2.
    Qin, Z. K., X. L. Zou, and F. Z. Weng, 2013: Evaluating added benefits of assimilating goes imager radiance data in GSI for coastal QPFs. Mon. Wea. Rev., 141, 75−92, https://doi.org/10.1175/MWR-D-12-00079.1.
    Roberts, N. M., and H. W. Lean, 2008: Scale-selective verification of rainfall accumulations from high-resolution forecasts of convective events. Mon. Wea. Rev., 136, 78−97, https://doi.org/10.1175/2007MWR2123.1.
    Sawada, Y., K. Okamoto, M. Kunii, and T. Miyoshi, 2019: Assimilating every-10-minute himawari-8 infrared radiances to improve convective predictability. J. Geophys. Res., 124, 2546−2561, https://doi.org/10.1029/2018JD029643.
    Schmetz, J., P. Pili, S. Tjemkes, D. Just, J. Kerkmann, S. Rota, and A. Ratier, 2002: An introduction to meteosat second generation (MSG). Bull. Amer. Meteor. Soc., 83, 977−992, https://doi.org/10.1175/1520-0477(2002)083<0977:AITMSG>2.3.CO;2.
    Schmit, T. J., M. M. Gunshor, W. P. Menzel, J. J. Gurka, J. Li, and A. S. Bachmeier, 2005: Introducing the next-generation advanced baseline imager on GOES-R. Bull. Amer. Meteor. Soc., 86, 1079−1096, https://doi.org/10.1175/BAMS-86-8-1079.
    Schmit, T. J., P. Griffith, M. M. Gunshor, J. M. Daniels, S. J. Goodman, and W. J. Lebair, 2017: A closer look at the ABI on the GOES-R series. Bull. Amer. Meteor. Soc., 98, 681−698, https://doi.org/10.1175/BAMS-D-15-00230.1.
    Shen, F. F., and J. Z. Min, 2015: Assimilating AMSU‐a radiance data with the WRF hybrid En3DVAR system for track predictions of Typhoon Megi (2010). Adv. Atmos. Sci., 32, 1231−1243, https://doi.org/10.1007/s00376-014-4239-4.
    Shen, F. F., D. M. Xu, and J. Z. Min, 2019: Effect of momentum control variables on assimilating radar observations for the analysis and forecast for Typhoon Chanthu (2010). Atmospheric Research, 230, 104622, https://doi.org/10.1016/j.atmosres.2019.104622.
    Stengel, M., P. Undén, M. Lindskog, P. Dahlgren, N. Gustafsson, and R. Bennartz, 2009: Assimilation of SEVIRI infrared radiances with HIRLAM 4D-Var. Quart. J. Roy. Meteor. Soc., 135, 2100−2109, https://doi.org/10.1002/qj.501.
    Sun, J. Z., H. L. Wang, W. X. Tong, Y. Zhang, C.-Y., Lin, and D. M. Xu, 2016: Comparison of the impacts of momentum control variables on high-resolution variational data assimilation and precipitation forecasting. Mon. Wea. Rev., 144, 149−169, https://doi.org/10.1175/MWR-D-14-00205.1.
    Tewari, M., and Coauthors, 2004: Implementation and verification of the unified NOAH land surface model in the WRF model. In: Paper 14.2A, 20th conference on Weather Analysis and Forecasting/16th conference on numerical weather prediction, Vol 1115, pp 6.
    Thompson, G., R. M. Rasmussen, and K. Manning, 2004: Explicit forecasts of winter precipitation using an improved bulk microphysics scheme. Part I: Description and sensitivity analysis. Mon. Wea. Rev., 132, 519−542, https://doi.org/10.1175/1520-0493(2004)132<0519:EFOWPU>2.0.CO;2.
    Tiedtke, M., 1989: A comprehensive mass flux scheme for cumulus parameterization in large-scale models. Mon. Wea. Rev., 117, 1779−1800, https://doi.org/10.1175/1520-0493(1989)117<1779:ACMFSF>2.0.CO;2.
    Wang, P., J. Li, B. Lu, T. J. Schmit, J. Z. Lu, Y.-K. Lee, J. L. Li, and Z. Q. Liu, 2018a: Impact of moisture information from advanced Himawari imager measurements on heavy precipitation forecasts in a regional NWP model. J. Geophys. Res., 123, 6022−6038, https://doi.org/10.1029/2017JD028012.
    Wang, Y. B., Z. Q. Liu, S. Yang, J. Z. Min, L. Q. Chen, Y. D. Chen, and T. Zhang, 2018b: Added value of assimilating himawari-8 AHI water vapor radiances on analyses and forecasts for “7.19” severe storm over North China. J. Geophys. Res., 123, 3374−3394, https://doi.org/10.1002/2017JD027697.
    Wu, Y. L., Z. Q. Liu, and D. Q. Li, 2020: Improving forecasts of a record-breaking rainstorm in Guangzhou by assimilating every 10-min AHI radiances with WRF 4DVAR. Atmospheric Research, 239, 104912, https://doi.org/10.1016/j.atmosres.2020.104912.
    Xu, D. M., Z. Q. Liu, X.-Y. Huang, J. Z. Min, and H. L. Wang, 2013: Impact of assimilating IASI radiance observations on forecasts of two tropical cyclones. Meteorol. Atmos. Phys., 122, 1−18, https://doi.org/10.1007/s00703-013-0276-2.
    Xu, D. M., J. Z. Min, F. F. Shen, J. M. Ban, and P. Chen, 2016: Assimilation of MWHS radiance data from the FY-3B satellite with the WRF Hybrid-3DVAR system for the forecasting of binary typhoons. Journal of Advances in Modeling Earth Systems, 8, 1014−1028, https://doi.org/10.1002/2016MS000674.
    Yang, C., Z. Q. Liu, F. Gao, P. P. Childs, and J. Z. Min, 2017a: Impact of assimilating GOES imager clear-sky radiance with a rapid refresh assimilation system for convection-permitting forecast over Mexico. J. Geophys. Res., 122, 5472−5490, https://doi.org/10.1002/2016JD026436.
    Yang, J., Z. Q. Zhang, C. Y. Wei, F. Lu, and Q. Guo, 2017b: Introducing the new generation of Chinese geostationary weather satellites, fengyun-4. Bull. Amer. Meteor. Soc., 98, 1637−1658, https://doi.org/10.1175/BAMS-D-16-0065.1.
    Zhang, Y. J., F. Q. Zhang, and D. J. Stensrud, 2018: Assimilating all-sky infrared radiances from GOES-16 ABI using an Ensemble Kalman filter for convection-allowing severe thunderstorms prediction. Mon. Wea. Rev., 146, 3363−3381, https://doi.org/10.1175/MWR-D-18-0062.1.
    Zhuge, X., and X. Zou, 2016: Test of a modified infrared-only ABI cloud mask algorithm for ahi radiance observations. J. Appl. Meteorol. Climatol., 55, 2529−2546, https://doi.org/10.1175/JAMC-D-16-0254.1.
    Zou, X. L., Z. K. Qin, and F. Z. Weng, 2011: Improved coastal precipitation forecasts with direct assimilation of GOES-11/12 imager radiances. Mon. Wea. Rev., 139, 3711−3729, https://doi.org/10.1175/MWR-D-10-05040.1.
  • [1] Bo LIU, Juan HUO, Daren LYU, Xin WANG, 2021: Assessment of FY-4A and Himawari-8 Cloud Top Height Retrieval through Comparison with Ground-Based Millimeter Radar at Sites in Tibet and Beijing, ADVANCES IN ATMOSPHERIC SCIENCES, 38, 1334-1350.  doi: 10.1007/s00376-021-0337-2
    [2] Andrei Filei, Aleksei Sorokin, Olga Girina, 2024: Retrieval of Volcanic Sulphate Aerosols Optical Parameters from AHI Radiometer Data, ADVANCES IN ATMOSPHERIC SCIENCES.  doi: 10.1007/s00376-024-3105-2
    [3] HUO Juan, LU Daren, 2010: Preliminary Retrieval of Aerosol Optical Depth from All-sky Images, ADVANCES IN ATMOSPHERIC SCIENCES, 27, 421-426.  doi: 10.1007/s00376-009-8216-2
    [4] Chao LIU, Shu YANG, Di DI, Yuanjian YANG, Chen ZHOU, Xiuqing HU, Byung-Ju SOHN, 2022: A Machine Learning-based Cloud Detection Algorithm for the Himawari-8 Spectral Image, ADVANCES IN ATMOSPHERIC SCIENCES, 39, 1994-2007.  doi: 10.1007/s00376-021-0366-x
    [5] Yunji ZHANG, Eugene E. CLOTHIAUX, David J. STENSRUD, 2022: Correlation Structures between Satellite All-Sky Infrared Brightness Temperatures and the Atmospheric State at Storm Scales, ADVANCES IN ATMOSPHERIC SCIENCES, 39, 714-732.  doi: 10.1007/s00376-021-0352-3
    [6] Fabien CARMINATI, Stefano MIGLIORINI, 2021: All-sky Data Assimilation of MWTS-2 and MWHS-2 in the Met Office Global NWP System., ADVANCES IN ATMOSPHERIC SCIENCES, 38, 1682-1694.  doi: 10.1007/s00376-021-1071-5
    [7] ZHENG Xiaogu, WU Guocan, ZHANG Shupeng, LIANG Xiao, DAI Yongjiu, LI Yong, , 2013: Using Analysis State to Construct a Forecast Error Covariance Matrix in Ensemble Kalman Filter Assimilation, ADVANCES IN ATMOSPHERIC SCIENCES, 30, 1303-1312.  doi: 10.1007/s00376-012-2133-5
    [8] Xiaoli XIA, Jinzhong MIN, Feifei SHEN, Yuanbing WANG, Chun YANG, 2019: Aerosol Data Assimilation Using Data from Fengyun-3A and MODIS: Application to a Dust Storm over East Asia in 2011, ADVANCES IN ATMOSPHERIC SCIENCES, 36, 1-14.  doi: 10.1007/s00376-018-8075-9
    [9] Rui LI, Jiheng HU, Shengli WU, Peng ZHANG, Husi LETU, Yu WANG, Xuewen WANG, Yuyun FU, Renjun ZHOU, Ling SUN, 2022: Spatiotemporal Variations of Microwave Land Surface Emissivity (MLSE) over China Derived from Four-Year Recalibrated Fengyun 3B MWRI Data, ADVANCES IN ATMOSPHERIC SCIENCES, 39, 1536-1560.  doi: 10.1007/s00376-022-1314-0
    [10] Rong KONG, Ming XUE, Edward R. MANSELL, Chengsi LIU, Alexandre O. FIERRO, 2024: Assimilation of GOES-R Geostationary Lightning Mapper Flash Extent Density Data in GSI 3DVar, EnKF, and Hybrid En3DVar for the Analysis and Short-Term Forecast of a Supercell Storm Case, ADVANCES IN ATMOSPHERIC SCIENCES, 41, 263-277.  doi: 10.1007/s00376-023-2340-2
    [11] GAO Feng*, Peter P. CHILDS, Xiang-Yu HUANG, Neil A. JACOBS, and Jinzhong MIN, 2014: A Relocation-based Initialization Scheme to Improve Track-forecasting of Tropical Cyclones, ADVANCES IN ATMOSPHERIC SCIENCES, 31, 27-36.  doi: 10.1007/s00376-013-2254-5
    [12] Zhiyong MENG, Eugene E. CLOTHIAUX, 2022: Contributions of Fuqing ZHANG to Predictability, Data Assimilation, and Dynamics of High Impact Weather: A Tribute, ADVANCES IN ATMOSPHERIC SCIENCES, 39, 676-683.  doi: 10.1007/s00376-021-1362-x
    [13] Yong LI, Siming LI, Yao SHENG, Luheng WANG, 2018: Data Assimilation Method Based on the Constraints of Confidence Region, ADVANCES IN ATMOSPHERIC SCIENCES, 35, 334-345.  doi: 10.1007/s00376-017-7045-y
    [14] ZHU Jiang, LIN Caiyan, WANG Zifa, 2009: Dust Storm Ensemble Forecast Experiments in East Asia, ADVANCES IN ATMOSPHERIC SCIENCES, 26, 1053-1070.  doi: 10.1007/s00376-009-8218-0
    [15] FU Weiwei, ZHOU Guangqing, WANG Huijun, 2004: Ocean Data Assimilation with Background Error Covariance Derived from OGCM Outputs, ADVANCES IN ATMOSPHERIC SCIENCES, 21, 181-192.  doi: 10.1007/BF02915704
    [16] Bohua Huang, James L. Kinter III, Paul S. Schopf, 2002: Ocean Data Assimilation Using Intermittent Analyses and Continuous Model Error Correction, ADVANCES IN ATMOSPHERIC SCIENCES, 19, 965-992.  doi: 10.1007/s00376-002-0059-z
    [17] Banglin ZHANG, Vijay TALLAPRAGADA, Fuzhong WENG, Jason SIPPEL, Zaizhong MA, 2015: Use of Incremental Analysis Updates in 4D-Var Data Assimilation, ADVANCES IN ATMOSPHERIC SCIENCES, 32, 1575-1582.  doi: 10.1007/s00376-015-5041-7
    [18] Jidong GAO, Keith BREWSTER, Ming XUE, 2006: A Comparison of the Radar Ray Path Equations and Approximations for Use in Radar Data Assimilation, ADVANCES IN ATMOSPHERIC SCIENCES, 23, 190-198.  doi: 10.1007/s00376-006-0190-3
    [19] Fuqing ZHANG, Meng ZHANG, James A. HANSEN, 2009: Coupling Ensemble Kalman Filter with Four-dimensional Variational Data Assimilation, ADVANCES IN ATMOSPHERIC SCIENCES, 26, 1-8.  doi: 10.1007/s00376-009-0001-8
    [20] Xiaogu ZHENG, 2009: An Adaptive Estimation of Forecast Error Covariance Parameters for Kalman Filtering Data Assimilation, ADVANCES IN ATMOSPHERIC SCIENCES, 26, 154-160.  doi: 10.1007/s00376-009-0154-5

Get Citation+

Export:  

Share Article

Manuscript History

Manuscript received: 03 July 2020
Manuscript revised: 02 November 2020
Manuscript accepted: 02 December 2020
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Assimilating All-sky Infrared Radiances from Himawari-8 Using the 3DVar Method for the Prediction of a Severe Storm over North China

    Corresponding author: Dongmei XU, xdmjolly@gmail.com
  • 1. The Key Laboratory of Meteorological Disaster, Ministry of Education (KLME)/Joint International Research Laboratory of Climate and Environment Change (ILCEC)/Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters (CIC-FEMD), Nanjing University of Information Science & Technology, Nanjing 210044, China
  • 2. Heavy Rain and Drought-Flood Disasters in Plateau and Basin Key Laboratory of Sichuan Province, Chengdu 610225, China
  • 3. National Center for Atmos. Res., Boulder, Colorado, 80301, USA
  • 4. Institute of Urban Meteorology, China Meteorological Administration, Beijing 100089, China

Abstract: Although radar observations capture storm structures with high spatiotemporal resolutions, they are limited within the storm region after the precipitation formed. Geostationary satellites data cover the gaps in the radar network prior to the formation of the precipitation for the storms and their environment. The study explores the effects of assimilating the water vapor channel radiances from Himawari-8 data with Weather Research and Forecasting model data assimilation system (WRFDA) for a severe storm case over north China. A fast cloud detection scheme for Advanced Himawari imager (AHI) radiance is enhanced in the framework of the WRFDA system initially in this study. The bias corrections, the cloud detection for the clear-sky AHI radiance, and the observation error modeling for cloudy radiance are conducted before the data assimilation. All AHI radiance observations are fully applied without any quality control for all-sky AHI radiance data assimilation. Results show that the simulated all-sky AHI radiance fits the observations better by using the cloud dependent observation error model, further improving the cloud heights. The all-sky AHI radiance assimilation adjusts all types of hydrometeor variables, especially cloud water and precipitation snow. It is proven that assimilating all-sky AHI data improves hydrometeor specifications when verified against the radar reflectivity. Consequently, the assimilation of AHI observations under the all-sky condition has an overall improved impact on both the precipitation locations and intensity compared to the experiment with only conventional and AHI clear-sky radiance data.

摘要: 尽管高时空分辨率的雷达资料能精准地探测出风暴系统的三维结构,但是其有效观测半径有限并且观测时段局限于降水发生之后。作为雷达资料的有益补充,静止卫星辐射亮温观测是能够提供快速演变的对流天气系统结构和环境场信息的主要资料来源。本文以北方的一次强降水过程为例,在WRFDA (Weather Research and Forecasting model data assimilation system)同化系统中实现了Himawari-8 红外辐射率资料(Advanced Himawari Imager,简称AHI)的全空同化。完成了偏差订正模型、晴空条件下的云检测模型以及全空条件下的对称观测误差模型的构建。结果显示,使用云依赖的观测误差模型能改进云雨区AHI卫星资料的有效同化,改善了云量和云高的分析。通过和卫星观测和雷达观测相对比,可以发现,AHI辐射率资料的全空同化相对于晴空同化能够改进大部分地区云雨变量特别是雪水和云中液态水。进一步通过对比各组试验的降水结果可以发现,AHI辐射率资料的全空同化改进本次强降水过程的落区和强度预报效果最为显著。

    • Remote sensing observations are important in terms of improving the numerical weather prediction (NWP), especially for domains lacking conventional observations (McNally et al., 2000; Xu et al., 2016) through data assimilation (DA). It has been shown that advanced geostationary satellites are able to provide nearly continuous four-dimensional information about the atmosphere, similarly to radar observations. Geostationary satellites are able to monitor the severe weather with higher spatial and temporal resolutions. In recent years, new generation geostationary weather satellites were launched that are superior to their predecessors in terms of temporal and spatial resolutions for more spectral bands (Schmit et al., 2005; Bessho et al., 2016; Yang et al., 2017b). Most of them, such as the FY-4 series of the China Meteorological Administration (Yang et al., 2017b), Himawari-8/9 of the Japan Meteorological Agency (Bessho et al., 2016), and the Geostationary Operational Environmental Satellite (GOES)-R series of the National Aeronautics and Space Administration (NASA) (Schmit et al., 2005) have been recently investigated in both global and regional NWP models.

      In the global NWP models, the use of geostationary satellite observations is well demonstrated. For example, based on the four dimensional variational (4DVar) system, the European Centre for Medium-Range Weather Forecasts (ECMWF) assimilates Meteosat visible and infrared imager (MVIRI) onboard Meteosat-7 (Köpken et al., 2004) and the Spinning Enhanced Visible and InfraRed Imager (SEVIRI) on board Meteosat-8 (Schmetz et al., 2002). In the National Centers for Environmental Prediction (NCEP) global data assimilation system, the GOES imager observations are assimilated using three dimensional variational assimilation (3DVar). For regional NWP models, Stengel et al. (2009) explored the impact of SEVIRI assimilation with the 3DVar system (10 km) and 4DVar system (22 km) with the France/Aldin model respectively. Zou et al. (2011) and Qin et al. (2013) assimilated GOES imager with the NCEP Gridpoint Statistical Interpolation (GSI) 3DVar scheme using 9 and 10 km resolutions, respectively, for improving the track and intensity forecast for Hurricane Debby (2012) as well as the precipitation skills over the coastal areas of the Gulf of Mexico. Wang et al. (2018a) assimilated the layer precipitable water (LWP) retrieved from the Advanced Himawari-8 imager (AHI) for local storms. The impact of the cumulus parameterization schemes on assimilating the AHI LWP data are also studied in Lu et al. (2019). In the Weather Research and Forecasting Data Assimilation (WRFDA) system (Barker et al., 2012), Yang et al. (2017a) assimilated GOES imager radiances for convection-permitting (4 km) forecasts over Mexico. Wang et al. (2018b) investigated the added value of clear-sky AHI radiances assimilation on the analyses and forecasts for a severe storm using the WRFDA.

      The above studies prove that NWP can be improved by assimilating the Infrared (IR) Radiances data from geostationary satellites measurements that are sensitive to the mass and thermal state, such as temperature, pressure and wind. However, less appreciated is the benefit of satellite data measuring atmospheric moisture, clouds and precipitation (Geer et al., 2017, 2019), since it is not possible to improve the model initial conditions by holding cloud constant in DA. It is also very desirable to fully utilize the cloud information from the IR channels, since the cloud development is observed earlier by the IR channels than precipitation by the radar observations. Thus, the problem of under-utilizing humidity and precipitation information from IR channels needs to be addressed (Chevallier et al., 2001; Chevallier and Kelly, 2002) through all-sky IR radiance assimilation. Because of the uncertainties and complexities of the cloudy radiance simulation in the radiative transfer models and the analysis techniques with clouds considered, the all-sky assimilation of IR channels is among one of the most challenging issues in DA (Li et al., 2016b). Generally, the difficulties exist in the zero-gradient problems for the clear background, the complex statistics of the situation dependent background error, the observation errors and the simulation accuracy in the radiative transfer models. Recently, Minamide and Zhang (2017) investigated an adaptive observation error inflation method to limit the erroneous analysis increments due to large representativeness errors, as is often the case for cloud-affected radiances for GOES-16. Harnisch et al. (2016) investigated a unified method to assimilate the two humidity channels at 6.2 and 7.3 μm from Meteosat Second Generation (MSG) considering the nonlinear impact of clouds on both observation and model forecast as well as the systematic deficiency of model and observation operator. Based on a local ensemble transform Kalman filter (LETKF, Hunt et al., 2007) method, Okamoto (2017) evaluated the all-sky Himawari-8/AHI IR radiance assimilation in a mesoscale NWP model with 5-km resolution using LETKF after studying the observation error models. Okamoto et al. (2019) further compared the assimilation clear-sky and all-sky IR radiances for Himawari-8/AHI in a mesoscale system using a single humidity band. Improvements for temperature, water vapor, wind, and rainfall are found in their study because of the universal coverage of Himawari-8/AHI. They also proposed adaptive bias correction schemes by setting different predictors for all-sky IR radiance assimilation. Similar studies include that of Sawada et al. (2019), who assimilated Himawari-8/AHI radiances every 10-minutes to improve convective predictability. Zhang et al. (2018) assimilated all-sky IR radiances from GOES-16 ABI using EnKF for 1 km model convection-permitting severe thunderstorms prediction for a tornado case with an adaptive observation and first guess error inflation methods. They found that by removing spurious clouds, good estimation of ice particle and liquid particle fields could be analyzed. Honda et al. (2018a) conducted Himawari-8/AHI radiances assimilation every 10 minutes to improve convective predictability. The assimilation of all-sky Himawari-8 AHI radiance for the moisture and hydrometeor initialization for TC predictions were also studied with promising results (e.g., Honda et al., 2018b, 2019; Minamide and Zhang, 2018).

      This study aims to evaluate the effects of assimilating the clear-sky and all-sky IR radiances from Himawari-8 for a severe storm over north China in the framework of WRFDA using the 3DVar. The differences between the clear-sky and all-sky AHI radiance DA analyses and forecasts are explored to provide some guidelines for the usage of AHI radiance data with high spatial and temporal resolutions.

      This paper is organized as follows. The introduction of the WRFDA-3DVar system, the AHI IR radiance assimilation methods, and the background error modeling method are provided in section 2. Section 3 describes the experiment design and DA configurations. The results are illustrated in Section 4 before concluding in section 5.

    2.   Methodologies
    • The cost function is the core of the 3DVar data assimilation technique. For the traditional 3DVar method of the WRFDA (Barker et al., 2012), the prescribed cost function is designed to achieve the optimal analysis through minimization iteratively as:

      where ${{{y}}_{{\rm{o}}}} $ is the observation vector, x and xb represent the analysis and first guess (background) states, respectively, B is the background error covariance assigning the forecast error, the observation error is described by the observation error covariance matrix R, and H stands for the observation operator mapping the variables in model space to the variables in observation space.

    • The Community Radiative Transfer Model (CRTM: Liu and Weng, 2006) is the forward operator applied in our DA experiments. CRTM is enabled to simulate both the clear-sky and all-sky radiance. The temperature and moisture profiles from the model are the essential inputs for the clear-sky radiance simulation, while extra hydrometeor inputs (such as the cloud rainwater mixing ratio) are also needed for the all-sky radiance simulation. To conduct satellite DA, the corresponding tangent linear model and adjoint model are required in addition to the forward operator. CRTM has proven to be quite efficient with the Jacobian model included for calculating the Jacobians.

      As is well known, it is necessary to correct the systematic biases before DA (Dee and Uppala, 2009) by modifying the observation operator with the bias considered (Liu et al., 2012; Xu et al., 2013; Otkin et al., 2018). The modification includes four potentially state-dependent predictors (200–50 hPa, 1000–300 hPa layer thicknesses, surface skin temperature, and total column water vapor) along with their coefficients, as well as a constant component of total bias. The variational bias correction [VarBC: (Auligné et al., 2007)] scheme is employed in this study. The bias information for the water vapor channels is estimated based upon the departure (observation minus simulated clear-sky radiance from the model background) information of the two week-long period from 0000 UTC 25 June 2018 to 0000 UTC 9 July 2018. Then, the bias correction is performed with VarBC, so that the correction coefficients of the predictors are updated along with the variational analysis during the minimization procedure. As is known, the OMB (observation minus background) errors under the all-sky condition are from multiple unconfirmed sources (Sawada et al., 2019). It is rather complicated to consider and classify the error sources specifically for all sky pixels. Thus, the bias correction coefficients applied for the all-sky AHI IR radiance assimilation in this study are only estimated based on the clear sky pixels excluding the cloud sky pixels. Therefore, it is reliable to directly ingest the bias correction coefficients under clear-sky conditions to represent those under all-sky conditions.

      For the all-sky AHI IR radiance DA in our current study, no quality control procedure is conducted, since the situation-dependent observation error model is applied. It should be pointed out that the observation error model may not be able to include all the error sources from the observations. Nevertheless, it is of interest to validate the observation error model by considering all kinds of observations without any quality control procedure. For the clear-sky AHI IR radiance DA, strict quality control procedures are required since the retention of erroneous AHI IR radiance data after quality control is likely to ruin the analyses. Erroneous data after quality control tend to further affect the design of the bias correction procedures. Clear-sky AHI IR radiance DA uses a range of quality controls, including:

      (1) removing pixels where satellite zenith angels are larger than 65°;

      (2) removing pixels where channels are over mixed surface types;

      (3) removing pixels where the standard deviation of the neighboring 9 observation points for channel 13 (10.4 μm) is larger than 2 K;

      (4) removing pixels with a gross check, namely |OMB| ≥8 K;

      (5) eliminating data if bias-corrected innovation is larger than 3 times the observation error under clear-sky conditions. The clear sky observation errors will be described in section 2.3;

      (6) eliminating cloudy pixels with the cloud detection schemes that will be introduced as below.

      Clear-sky AHI IR radiance data assimilation relies on a cloud detection technique to separate the cloudy AHI IR radiances from the clear-sky AHI IR radiance before assimilation, allowing a better use of available data. The cloud-detection algorithm (named ZZ-cloud detection scheme hereafter) developed by Zhuge and Zou (2016) are applied to search different cloud cases to achieve a composite cloud mask (Wu et al., 2020). In this study, the ZZ-cloud detection scheme is enhanced in the framework of WRFDA system initially. With the 10 tests in the ZZ-cloud detection algorithm, the metrics provided by different tests indicate the signals for discriminating the existence of the cloud.

    • Among the key issues of the all-sky AHI IR radiance data assimilation technology, the correct estimation of the observation errors has an important impact on the quality of the analysis because the observation error directly affects the determination of the weights for the observations in the analysis procedure. The observation errors should be assigned as situation-dependent, considering the representativeness errors that are smaller in clear-sky conditions and larger in the presence of clouds and precipitation (Geer and Bauer, 2011). The observation errors should also be channel dependent since the detection characteristics and the extent of their sensitivities to various cloud and precipitation conditions are different. Based on the observation error model in Harnisch et al. (2016), the observation error related statistics are obtained by calculating the OMB difference from 0000 UTC 4 July 2018 to 1800 UTC 7 July 2018 every 6 hours by using the 3 h forecast as the background. The brightness temperature differences of the clear-sky and the all-sky AHI IR radiance simulated with the CRTM are first calculated from the background. The differences are then binned according to the brightness temperature simulated from the background under the all-sky condition. The clear-sky AHI IR radiance simulation from CRTM needs to remove all hydrometeors while keeping temperature and moisture profiles unchanged. The corresponding brightness temperature value is determined as the brightness temperature threshold (BTlim) of each channel, when the brightness temperature difference is less than a certain value (0.1 K). Once the BTlim of the three water vapor channels are determined, the averaged cloud amount Ca can be estimated based on both the background and the observation, indicated as:

      Here Obs is AHI IR radiance observation and Bak is the simulated all-sky AHI IR radiance from the background. Cxb and Cy are determined from the background and the observation respectively. The parallel relationships between the observation errors and the cloud amount are estimated based on a large amount of OMB samples (observation minus the simulated all-sky AHI IR radiance from the background). The four parameters – the minimum observation errors (SDmin), the maximum observation errors (SDmax), the minimum cloud amount Ca1, and the maximum cloud amount Ca2 – are determined from the statistics. With the four determined parameters, it is convenient to estimate the cloud amount dependent observation errors for any given Ca as follows,

    • The climatological background error statistics are computed offline with the commonly used National Meteorological Center (NMC) method (Parrish and Derber, 1992). A number of 24 h and 12 h forecasts are launched at different times over roughly one month in July of the previous year in 2017 (Chen et al., 2018). Pairs of forecasts valid at the same time with different forecast lead times are extracted by the NMC method to represent the difference through averaging them over the time and domain, for example with 24 h and 12 h forecasts, respectively. Both 24 h and 12 h forecasts contain errors, while the truth is never known. Further, the B matrix is decomposed as B=UUT with the preconditioning transform applied as Uv=xxb (Barker et al., 2004). Here x and xb is the analysis and background respectively, while v represents the control variable.

      The Control Variable option 7 (CV7) is applied in this study to include the zonal wind component, the meridional wind component, the full temperature, the full surface pressure, and the pseudo-relative humidity as the control variables (Chen et al., 2016; Li et al., 2016a; Sun et al., 2016; Shen et al., 2019). In addition to the five commonly used control variables, another five hydrometeor-related control variables (snow: qs, rainwater: qr, cloud liquid: qc, ice: qi, and graupel: qg mixing ratio) are also considered using GEN_BE v2.0 software developed by Descombes et al. (2014). These hydrometeor related control variables represent the microphysical variables for both ice and liquid particles in the cost function so as to effectively assimilate the cloudy AHI IR radiance observations.

    3.   Experiment Setup
    • On 17 October 2014, the Meteorological Satellite Center of Japan Meteorological Agency launched the Himawari-8 satellite successfully into geosynchronous orbit. The AHI instrument onboard Himawari-8 has an observing capability significantly improved compared with its predecessor Himawari-7 series satellites, and is similar to the Advanced Baseline Imager (ABI) onboard U.S. GOES-R satellites in terms of the hardware configurations (Schmit et al., 2005, 2017). Full-disk imagery is provided every 10 minutes by AHI, covering Japan and China. There are 16 channels including IR channels (nadir resolution 2 km) and near-IR and visible channels (nadir resolution 0.5 or 1 km). Among them, channel 7 to channel 16 are IR channels with central wavelengths ranging from 3.9 μm to 13.3 μm. The specifications of the channel information appear in Bessho et al. (2016). In this study, the three water vapor channels of AHI IR radiance are assimilated, which are sensitive to middle-to-upper tropospheric humidity and less so to the lower troposphere moisture. The three water vapor channels include channel 8 (6.2 μm), channel 9 (6.9 μm), and channel 10 (7.3 μm). For clear-sky IR radiance assimilation, these three channels are applied in Wang et al. (2018b), For all-sky AHI IR radiance assimilation, only one humidity channel is assimilated in Minamide and Zhang (2018) and Honda et al., (2018b) to exclude the potential correlations among humidity channels. However, the idea of assimilating two humidity channels are stated in Okamoto (2017) with 6.2 μm and 6.9 μm and in Sawada et al., (2019) with 6.2 μm and 7.3 μm, respectively, for mesoscale systems. In Okamoto (2019), radiances from three water vapor channels and one window channel were assimilated with promising results. In addition, a 9 km resolution thinning mesh is applied to reduce the impact of the potential correlations in the radiance observations, following Wang et al. (2018b). It should be emphasized that 9 km may not be large enough to avoid the inter-channels correlations, especially for the AHI all sky data assimilation. The study of the impact of the thinning mesh on the results from the clear sky and all sky AHI data assimilation is another interesting topic, which is beyond the scope of this study and is planned for future studies.

    • In this study, a severe storm event over north China that occurred during 6−10 July 2018 was selected. The plots for the meteorological factors and weather situations are firstly examined before discussing the model configuration and the experimental design. There is an upper-level low pressure system over northwestern China at 500 hPa at 0000 UTC 6 July 2018 as the main factor for the persistent precipitation in north China. The upper-level low pressure system propagated to north China and became stronger on 8 July and 9 July 2018.Water vapor transport by the southwesterly flow near the subtropical high is strong at 850 hPa for the entire period, leading to a corresponding torrential rain (Fig. 1).

      Figure 1.  Wind field (vector; m s−1) and geopotential height (shaded; gpm) of the ECMWF analyses (ECA) at 500 hPa valid at (a) 0000 UTC 6 July, (b) 0000 UTC 8 July 2018, Wind field (vector; m s−1) and relative humidity (shaded; %) of the ECA at 850 hPa valid at (c) 0000 UTC 8 July 2018, (d) 0000 UTC 9 July 2018.

      The numerical simulations are based on ARW (Advanced Research WRF) of WRFV4.0.3. The simulation area center over north China is determined as 37°N, 105°E in Fig. 2a with 649 × 500 grid points (9 km) for domain 1. For domain 2, the grid points are 550 × 424 (3 km). The domains are configured with vertical structure of 57 unequally spaced sigma (non-dimensional pressure) levels with 10 hPa as the cloud top. The initial conditions and boundary conditions for the numerical simulations are from ECMWF analyses with 0.25° × 0.25° resolutions. The following parameterizations are used: New Thompson microphysics scheme (Thompson et al., 2004), YSU planetary boundary layer scheme (Hong et al., 2006), NOAH land surface scheme (Tewari et al., 2004), New Tiedtke scheme cumulus scheme for the larger domain (D01) (Tiedtke, 1989), and the Rapid Radiative Transfer Model (RRTM) longwave and shortwave radiation schemes (Mlawer et al., 1997).

      Figure 2.  (a) The WRF model domains of D01 and D02, (b) The distribution of the conventional observations valid at0000 UTC 8 July 2018. The numbers of each observation are marked on the right

      The cold-start and subsequent cycling data assimilation experiments are completed for the 4-day from 0000 UTC 6 July to 0000 UTC 10 July 2018. A 6 h spin-up run is initialized as the cold start from 1800 UTC to 0000 UTC for each day. Then, the data assimilation and forecast are conducted every 3 hours for the rest of the day. A 24 h deterministic forecast is initialized from the final analysis of each day.

      Two DA experiments are designed to investigate the impact of AHI IR radiances on the analyses and forecasts for the severe storm in north China. The first experiment was the benchmark experiment named as “conv+ahiclr” that assimilated the conventional observations in both the larger domain (D01) and the smaller domain (D02) and the AHI IR radiances under clear-sky conditions in D02. Conventional observations include aircraft reports, ship reports, sounding reports, satellite cloud wind data, GPS data, and ground stations provided by the China Meteorological Administration (CMA). A snapshot of the distribution of the conventional data is provided in Fig. 2b.

      The second experiment named as “conv+ ahicloud” is similar to the first experiment except for assimilating all-sky AHI IR radiance observations in D02. In D02, conventional observation and all-sky AHI IR radiance data are assimilated in two steps. The analysis with the assimilation of conventional observations from D02 is provided as the first guess for the all-sky AHI IR radiance data assimilation procedure.

    4.   Results
    • Figure 3 provides the number of observations from channel 8 before and after the quality control procedure. It is seen that most of the AHI IR radiance data are located over land. With the cloud detection and other quality control procedures, about 1/4 of the total AHI IR radiance observations are kept in the DA system. The counts of the assimilated AHI IR radiance observations are more than 4000 for most of the data assimilation cycles.

      Figure 3.  The total observation counts before the quality control (black), over land before the quality control (green), over sea before the quality control (blue) and the total counts after the quality control (red) for channel 10.

    • The bias correction coefficients of each predictor are estimated using OMB from 0000 UTC on 25 June 2018 to 0000 UTC on 9 July 2018 every 3 hours. The background is interpolated from the ECMWF analyses with 0.25° × 0.25° resolution. The statistically obtained bias correction coefficients are then used as the static coefficients for the first data assimilation cycle. For the clear-sky AHI IR radiance DA, the coefficients for the bias correction are updated by Variational Bias Correction scheme (VARBC) during the minimization, while those for the all-sky AHI IR radiance DA are kept constant. The bias is estimated with the obtained coefficients associated with the corresponding predictors for each DA cycle. Figure 4 shows time series of the 3 hourly OMB statistics with and without bias corrections for channel 9 of the AHI clear-sky radiance simulation from 0000 on 6 July 2018 to 0000 on 10 July 2018 by the conv+ahiclr experiment. Clearly, the bias correction scheme effectively reduces the standard deviation and the mean bias for almost all the analysis times, indicating the validity and rationality of the variational bias correction scheme. For the all-sky data assimilation, it is not straightforward to validate the bias correction scheme by the results of the all-sky bias correction, since the mean and standard deviation of the OMB in all-sky data assimilation include mixed information besides the systematic bias.

      Figure 4.  The mean bias and standard deviation of channel 9 for OMB with no bias correction (OMB-nb) and with bias correction (OMB-wb) from 0000 UTC 6 July 2018 to 0000 UTC 10 July 2018.

    • As was described in section 2.3, there is no significant difference for the radiative effect between the clear-sky and all-sky simulated radiance when the brightness temperature is greater than BTlim. The brightness temperature thresholds BTlim of channels 8, 9, and 10 are 233.86 K, 242.51 K, and 253.93 K respectively. As is well known, channel 10 is more affected by clouds, since the weighting function shows a maximum at a relatively low height. Thus, the brightness temperature threshold BTlim of channel 10 is higher than the other two channels.

      Figure 5 shows the OMB distributions normalized by the constant standard deviation of all the samples (blue curve). The statistical results are based on the OMB samples from 0300 UTC on 4 July 2018 to 2100 UTC on 7 July 2018 every 6 hours. The background field for calculating of OMB is the 3 h forecasts initialized from 0000 UTC on 4 July 2018 to 1800 UTC on 7 July 2018 every 6 hours. The distributions of the OMB normalized by the binned standard deviation with different ranges of cloud amount instead of using the constant standard deviation are also presented (green curve). The statistics are obtained from the same OMB samples in Fig. 5 for the three water vapor channels. The red curves represent the Gaussian distribution. Note that the frequency of the negative OMB is significantly larger than that of the positive OMB for the blue line with the regular normalization. Following the dynamical normalization achieved by taking the varying cloud amount into account, the frequency of OMB in the negative region is quite comparable to that of the positive OMB, matching the Gaussian curves better. Approaching the Gaussian distribution for the OMB distribution is expected to lead more accurate analyses.

      Figure 5.  The OMB distributions normalized by the standard deviation of all the samples (blue curves) or normalized by the binned standard deviation with different ranges of cloud amount (green curves), and the gaussian curves (red curves) for (a) channel 8 (wavelength: 6.2 μm), (b) channel 9 (wavelength: 6.9 μm) and (c) channel 10 (wavelength: 7.3 μm).

    • Figures 6a, 6b, and 6c show the distribution of the observed AHI radiance along with the simulated radiance from the background and the analysis after the all-sky AHI IR radiance DA respectively, valid at 0000 UTC on 8 July 2018 for channel 10. The difference between the simulated radiance from the analysis and the background is also provided in Fig. 6d. As previously mentioned, channel 10 of AHI radiance is rather sensitive to clouds and precipitation. It is noted that clouds mainly exist in the southeast portion of the domain. The brightness temperature simulated with the background is significantly higher than the observed brightness temperatures in most areas, indicating that the hydrometeors in the background state are generally less than the actual situation. Following the all-sky AHI IR radiance assimilation, the amount of cloud is effectively simulated by moistening the background for most of the domain (Figs 6c, 6d). The results indicate that the assimilation of AHI water vapor radiance improves the analysis of temperature, humidity, and hydrometeor information, eventually leading to a better simulation of the AHI IR radiance.

      Figure 6.  (a) The observed brightness temperature, the simulated brightness temperature (b) from the background, (c) from the analysis of the all-sky radiance data assimilation, and (d) the difference between the simulated radiance from the analysis and the background (units: K) valid at 0000 UTC on 8 July 2018 for channel 10.

      Figure 7a shows the cloud height from the AHI level-2 cloud product provided by Japan Aerospace Exploration Agency (JAXA). Figures 7b and 7c present the diagnosed cloud height from the analyses in conv+ahiclr and conv+ahicloud respectively. The Unified Post Processor (UPP) software (Garcia-Reynoso and Mora-Ramirez, 2017) is employed to diagnose the cloud height taking the water vapor, the cloud water, the cloud ice, and the precipitation snow into account from the NWP model. It is found that the cloud height is overestimated near (39°N, 110°E) in conv+ahiclr. Similarly, the cloud amount is underestimated in the east of the model domain near the area of (40°N, 116°E) from conv+ahiclr. Generally, the cloud amount and the cloud height in the analysis with the all-sky AHI IR radiance data assimilation match the AHI level-2 cloud products better. It indicates that the clouds are probably effectively analyzed by fully utilizing the hydrometeor information in both the radiative transfer model and the cost function for the all-sky AHI IR DA.

      Figure 7.  Cloud height (units: gpm) for (a) AHI-level2, (b) conv+ahiclr, (c) conv+ahicloud 0000 UTC on 8 July 2018.

    • Figure 8 shows the analysis increments of the water vapor (qv), cloud water (qc), cloud rain (qr), and precipitation snow (qs) water path from the conv+ahiclr and conv+ahicloud. Noticeable water vapor increments exist in Beijing and Hebei provinces. Additional outstanding increments of the water vapor are observed from conv+ahicloud northwest of Beijing close to the region of Hebei and Inner Mongolia with the all-sky DA of AHI IR radiance in Fig. 8b. As expected, no visible hydrometeor increment is found in conv+ahiclr, since no hydrometeor is considered in the observation operator in the clear-sky AHI IR radiance data assimilation used in WRFDA. The qc increment exists in the conv+ahicloud experiment, whereas the increment of qr from conv+ahicloud is relatively smaller since precipitation is not observable by IR channels in most cases. The increment of qs from conv+ahicloud is distinct, indicating the existence of snow-phase clouds from the analysis. This large increment of qs explains the noticeable adjustment of the cloud height in the conv+ahicloud experiment in Fig. 8, since qs is one of the key inputs of UPP.

      Figure 8.  (a, b) The water vapor, (c, d) the cloud water, and (e, f) the cloud rain, (g, h) the precipitation snow water path at 0000 UTC 9 July 2018 for conv+ahiclr (left panel), and conv+ahicloud (right panel). The locations of Beijing (BJ) and Heibei (HB) province are marked in Fig. 8c.

      The differences of water vapor flux (WVF) at 850 hPa between analyses of conv+ahicloud and conv+ahiclr are illustrated in Fig. 9 along with the winds from the ECMWF analysis. Figure 9a shows the difference (conv+ahicloud minus conv+ahiclr) in the WVF valid at 0000 UTC on 8 July 2018. Compared with the conv+ahiclr, conv+ahicloud generally shows an increase of the WVF to the north and west of Hebei and Beijing, leading to overall intense model-simulated precipitation north of Beijing, especially at 0000 UTC on 8 July 2018. Water vapor can be transported from southwest to the rainfall area over Beijing area, since WVF represents the direction and magnitude of the water vapor transport to maintain a persistent heavy rainfall. It is consistent with Fig. 9 showing existence of rainfall from south to north in Beijing.

      Figure 9.  The water vapor flux (shaded; g cm−1 hPa−1 s−1) difference between analyses of conv+ahicloud and conv+ahiclr at 850 hPa at (a) 0000 UTC 7 July 2018, (b) 0000 UTC 8 July 2018. The vectors show the direction and magnitude of the wind from ECMWF analysis. The locations of Beijing (BJ) and Heibei (HB) province are marked in Fig. 9a.

    • The absolute bias and RMSE of the 12 h forecasts against the radar reflectivity (RF) are shown in Fig. 10 initialized from 1200 UTC on 6 July 2018 to 0000 UTC on 10 July 2018 every 12 hours. The radar RF observations are from 8 coastal radars in Handan of Hebei, Zhangbei of Hebei, Chengde of Hebei, Beijing, Tianjing, Shijiazhuang of Hebei, Qinhuangdao of Hebei, and Cangzhou of Hebei with total observations counts over 30000 (Fig. 10a). Note that the results of applying the all-sky AHI IR radiance data assimilation are superior to those of using the clear-sky AHI IR radiance data assimilation when verified against the radar reflectivity in Fig. 10 for both the bias and the RMSE. The RMSE is reduced for all of the forecasts while the improvements for the bias exist for most times. The reduced bias in Fig. 10b indicates an improvement forecast by moisturizing the background after the all-sky AHI IR radiance data assimilation.

      Figure 10.  (a) The counts of radar observations, (b) bias, and (c) the RMSE against the radar reflectivity for the 12 h forecasts valid from 1200 UTC 6 July 2018 to 0000 UTC 10 July 2018.

      Figure 11 provides the RMSE against the observed brightness temperature for the window channel 13. The RMSEs are calculated with different forecast lead times initialized from 0000 UTC on 6 July, 0000 UTC on 7 July, 0000 UTC on 8 July, and 0000 UTC on 9 July 2018 respectively. Consistent reduced RMSEs are found from the conv_ahicloud as expected. The reduction of RMSE persists for different forecast lead times although the largest reduction is observed at the analysis time.

      Figure 11.  The RMSE against the observed brightness temperature with different forecast leading time initialized from (a) 0000 UTC on 6 July 2018, (b) 0000 UTC on 7 July 2018, (c) 0000 UTC on 8 July 2018 (d) (a) 0000 UTC on 9 July 2018 for channel 13.

    • Figure12 presents the 24 h accumulated precipitation from 1200 UTC 8 July 2018 to 1200 UTC 9 July 2018 for the two DA experiments. The results initialized from other data assimilation cycles are not shown since similar results are found in terms of the accumulated precipitation. Generally, the precipitation pattern is more compact from conv+ahicloud than that from conv+ahiclear. As can be seen here, there is a rain band extending from the south to the north of Beijing marked as A in the observations. It is apparent that there is a rain band through Beijing from the conv+ahicloud (Fig. 12c), at least to a certain extent, while for the clear-sky AHI radiance data assimilation experiment, there is no match for such a rain band.

      Figure 12.  24 h accumulated precipitation (units: mm) for (a) precipitation data from CMA, (b) conv+ahiclr experiment, and (c) conv+ahicloud experiment from 1200 UTC 8 July 2018 to 1200 UTC 9 July 2018.

      In order to quantitatively evaluate the performance of precipitation prediction, the Fractional Skill Scores (FSS, Roberts and Lean, 2008) scores for the 1h accumulated rainfall calculated with a threshold of 5 mm are calculated as a function of forecast leading time in Fig. 13a averaged from the forecasts initialized from 0000 UTC 7 July 2018 to 1200 UTC 9 July 2018 every 12 hours. The influence distance of the neighborhood used in this study is 9 km. The verification domain is a sub-domain of D02 shown in Fig. 12 for the calculating of the FSS. It is found that the FSS decreasees with the forecast leading time. The FSS from conv+ahicloud are consistently higher than those from conv+ahiclr. In addition, the FSSs for the 24 h accumulated precipitation are calculated for the 5 mm, 10 mm, 15 mm, and 25 mm thresholds, respectively, averaging over the same period mentioned above. It can be seen from Fig. 13b that for most of the precipitation thresholds, the scores from the DA experiment with all-sky AHI IR radiance are superior to those from the clear-sky AHI IR radiance DA experiment. The conv+ahicloud experiment yields higher scores than conv+ahiclr experiment does for the thresholds of 10 mm, 15 mm, and 25 mm.

      Figure 13.  FSS (a) with a threshold of 5 mm h−1 as a function of forecast lead time (b) of the 24 h accumulated precipitation for conv+ahiclr and conv+ahicloud experiments averaged from 0000 UTC 7 July 2018 to 1200 UTC 9 July 2018 every 12 hour.

    5.   Conclusion
    • The effects of assimilating the infrared radiances from Himawari-8 data are studied with WRFDA for a severe storm over northern China. Both clear-sky and all-sky AHI IR radiance data assimilation (DA) experiments are designed. AHI radiance preprocessing is conducted including the bias corrections, cloud detection for clear-sky AHI IR radiance DA and observation error modeling for all-sky AHI IR radiance DA. AHI radiance data are fully applied without any quality control for all-sky AHI radiance data assimilation.

      With the cloud dependent observation error model, the analyzed all-sky AHI IR radiances fit the observations more closely, further improving the cloud heights. Compared to the clear-sky AHI IR radiance assimilation experiment, the all-sky AHI IR radiance assimilation is able to adjust all types of hydrometeor variables, especially for cloud water and precipitation snow. It is found that assimilating all-sky AHI IR radiance in addition to just conventional data is able to improve the hydrometeors when verified against the radar reflectivity. The assimilation of AHI data under the all-sky condition has an overall positive impact on both precipitation locations and precipitation intensity compared to the experiments with only conventional and clear-sky AHI IR radiance data.

      In future work, the statistical model of the background error covariance can be further enhanced to take multivariate correlations into account. In addition, the ensemble-variational DA method (Shen et al., 2015; Xu et al., 2016) is expected to provide better analyses through the use of the flow-dependent background error covariances. Four-dimensional ensemble-variational DA methods (Buehner et al., 2013; Liu and Xiao, 2013; Lorenc et al., 2014; Kleist and Ide, 2015; Shen et al., 2019) are planned to assimilate AHI IR radiance observations every 10 minutes for severe storm events. Other surface channels are also needed for investigation in order to perform careful treatment of the surface emissivity. How to effectively combine the radar observations and all-sky AHI IR radiance observations simultaneously is another interesting aspect in our future research.

      Acknowledgements. This research was primarily supported by the Chinese National Natural Science Foundation of China (G41805016, G41805070), the Chinese National Key R&D Program of China (2018YFC1506404, 2018YFC1506603), the research project of Heavy Rain and Drought-Flood Disasters in Plateau and Basin Key Laboratory of Sichuan Province in China (SZKT201901, SZKT20 1904), the research project of the Institute of Atmospheric Environment, China Meteorological Administration, Shenyang in China (2020SYIAE02, 2020SYIAE07).

Reference

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return