- Data assimilation of multi-source remote sensing observations
1.1 Quality control and data assimilation of the national radar reflectivity
The 3d isometric CAPPI products generated by over 200 Doppler weather radars were utilized to generate the national-wide radar reflectivity mosaic data, which was assimilated over the national 9km domain following a series of quality control and pre-processing procedures. The strategy of cycling assimilation of radar mosaic data was also modified to accommodate single assimilation, i.e., the data were only assimilated during the forecast phase. By adjusting the variance and length scales of the humidity control variable in background error covariance, the estimated moisture analysis effectively reduced the precipitation overprediction produced by the cycling assimilation of radar mosaic data.
1.2 Quality control and assimilation of wind profile observations
The observations of the nationwide profiler are subjected to a two-step QC procedure based on the iterated reweighted minimum covariance determinant (IRMCD) of multivariate data. On the basis of wind profile data, the U/V wind data assimilation test and test effect evaluation results were conducted in the CMA-BJ, and the assimilation application of the national wind profile radar data was implemented in the CMA-BJ V2.0 regional rapid hourly update cycle high-resolution numerical model system.
- Optimization of Physical Parameterization Schemes
3.2 Cloud radiative forcing scheme
The adoption of the new cloud radiation scheme modifies the scalability of cloud cover diagnosis in the radiation process. Following the adjustment, the amplitude of downward shortwave radiation decreases by half. Correspondingly, the warm bias of the 2-m temperature forecast was partially corrected, and the wetter boundary layer was favorable for the formation of more low clouds, resulting in less downward short-wave radiation reaching the surface, which is advantageous for the systematic improvement of the warm bias of the surface temperature forecast.
3.3 Optimization of the PBL and surface-layer schemes
The land-air exchange coefficient has a significant impact on the land-air coupling strength within the surface-layer scheme. In order to reduce forecast bias, the results indicate that the YSU PBL scheme is superior to ACM2 in describing the temperature and humidity profile within the boundary layer. Consequently, by replacing ACM2 with YSU and adjusting the energy exchange coefficient within PBL, the upward sensible and latent heating during the day were increased. Nevertheless, the 2-m temperature forecasts differed for various land-use types, whereas the 2-m humidity increased globally, with the increase being most pronounced in the forest region.
3.4 Update of vegetation coverage dataset
As an important input of land surface features, vegetation coverage has a significant impact on the simulation results of land surface models. After updating the dataset on vegetation coverage, the vegetation coverage of farmland and grassland areas in summertime North China increased to a certain extent, resulting in a decrease in upward sensible and an increase in latent heating of the surface during the day. The systematic warm and dry bias of the afternoon surface during the warm season was partially corrected.
3.5 Soil types and update NOAH's new table of soil hydraulics parameters
After updating the soil properties and soil hydrological parameters, the surface soil temperature in North China decreased and the relative humidity became dry, the sensible heat output from the land to the atmosphere decreased and the latent heat output increased, and the temperature and relative humidity in the lower atmosphere decreased and increased, respectively. Changes in the lower atmosphere reduce the forecasted surface temperature and humidity deviation in northern China.
3.6 Optimization of cumulus convection parameterization scheme
There are two major updates in the new Tiedtke scheme: (1) grid-distance dependent convective adjustment time; (2) grid-distance dependent conversion of cloud water to rain; (3) no middle-level convections are excited when the air layer is saturated. In addition, by turning off the shallow convective process in the scale-adapted New Tiedtke cumulus cloud convective parameter scheme, the large area of overprediction of small amount precipitation is improved.
- Hourly rapid updated cycling forecast system
4.1 Incremental Analysis update (IAU)
The incremental analysis update (IAU) method was developed for high frequency noise control and dynamic balance of hourly rapid update cycle system. This method can suppress false high frequency noise resulting from data assimilation and initial value imbalance between wind and pressure, hydrometeors, and dynamic fields. Within the time window centered on the assimilation moment, the IAU method was implemented by progressively incorporating analysis increments into the model's integration procedure. Therefore, the initial imbalance can be effectively identified and suppressed, and model spin-up/spin-down time is reduced.
4.2 Large-scale global forecast blending
To generate the analysis background, a dynamic blending scheme was created by combining large-scale information from global forecasts with small-scale information extracted from a regional model (DFB, dynamic forecast blending). Calculating the kinetic energy spectrum and error patterns from the global and regional models, respectively, yields the flow-dependent blending scales.
4.3 A catch-up hourly cycling strategy
The operation hourly cycling framework is divided into two sections: catch-up cycle and update forecast. In the catch-up cycling part, the availability and coverage of GTS, local, and remote sensing observations from multiple sources during the time window are taken into account in their entirety. On the basis of real-time monitoring of the timeline of observation arrival, the optimal observation cutoff time of the hourly rapid update cycle is determined so that the analysis component can make maximum use of real-time observations.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner