大气科学
大氣科學
대기과학
Chinese Journal of Atmospheric Sciences
2015年
6期
1165-1178
,共14页
刘德强%冯杰%李建平%王金成
劉德彊%馮傑%李建平%王金成
류덕강%풍걸%리건평%왕금성
GRAPES_MESO%时间步长%空间分辨率%预报效果
GRAPES_MESO%時間步長%空間分辨率%預報效果
GRAPES_MESO%시간보장%공간분변솔%예보효과
GRAEPS_MESO%Time-step size%Spatial resolution%Model prediction
基于GRAPES区域中尺度数值预报系统(GRAPES_MESO),针对700 hPa、500 hPa和200 hPa的位势高度场H,温度场T,风场纬向分量U,经向分量V和地面降水场,在给定的模式物理过程下,分别考察了时间步长和空间分辨率对于模式预报效果的影响。研究结果表明,空间分辨率(0.3°×0.3°)相同时,各变量在不同层次的预报几乎都存在最优时间步长使得预报技巧最高,初步说明最优时间步长理论在复杂的偏微分方程组中的适用性。随后,将空间分辨率为0.3°×0.3°时最优时间步长(240 s)的预报结果与当前业务中(空间分辨率为0.15°×0.15°、时间步长为90 s)的预报结果进行比较,发现前者的变量H、T、U、V和地面降水场的预报技巧均高于后者,表明并不是空间分辨率越高预报效果越好。
基于GRAPES區域中呎度數值預報繫統(GRAPES_MESO),針對700 hPa、500 hPa和200 hPa的位勢高度場H,溫度場T,風場緯嚮分量U,經嚮分量V和地麵降水場,在給定的模式物理過程下,分彆攷察瞭時間步長和空間分辨率對于模式預報效果的影響。研究結果錶明,空間分辨率(0.3°×0.3°)相同時,各變量在不同層次的預報幾乎都存在最優時間步長使得預報技巧最高,初步說明最優時間步長理論在複雜的偏微分方程組中的適用性。隨後,將空間分辨率為0.3°×0.3°時最優時間步長(240 s)的預報結果與噹前業務中(空間分辨率為0.15°×0.15°、時間步長為90 s)的預報結果進行比較,髮現前者的變量H、T、U、V和地麵降水場的預報技巧均高于後者,錶明併不是空間分辨率越高預報效果越好。
기우GRAPES구역중척도수치예보계통(GRAPES_MESO),침대700 hPa、500 hPa화200 hPa적위세고도장H,온도장T,풍장위향분량U,경향분량V화지면강수장,재급정적모식물리과정하,분별고찰료시간보장화공간분변솔대우모식예보효과적영향。연구결과표명,공간분변솔(0.3°×0.3°)상동시,각변량재불동층차적예보궤호도존재최우시간보장사득예보기교최고,초보설명최우시간보장이론재복잡적편미분방정조중적괄용성。수후,장공간분변솔위0.3°×0.3°시최우시간보장(240 s)적예보결과여당전업무중(공간분변솔위0.15°×0.15°、시간보장위90 s)적예보결과진행비교,발현전자적변량H、T、U、V화지면강수장적예보기교균고우후자,표명병불시공간분변솔월고예보효과월호。
This study considered the impacts of time-step size and spatial resolution on the prediction skill of the Global/Regional Assimilation and Prediction System (GRAPES) mesoscale numerical forecast system (GRAPES-MESO) for a given parameter set. The forecasts of geopotential height (H), temperature (T), and the zonal (U) and meridional (V)components of wind at 700, 500, and 200 hPa, were assessed, as well as surface precipitation. The results showed that, at a spatial resolution of 0.3°×0.3°, the prediction skill of almost all variables, including H, T, U and V, in the three vertical layers were optimized at a particular time step of approximately 240 s. This raises the possibility of an optimal time-step size for a particular spatial resolution, and the explanation for this relationship might be related to the computational uncertainty principle. The operational forecasts based on a spatial resolution of 0.15°×0.15° and a time-step size of 90 s were also compared with the best results obtained previously, in which the spatial resolution was 0.3°×0.3° and the time step was 240 s. The latter configuration possessed higher skill than the operational forecasts for all variables, indicating that the prediction quality may not be significantly improved by an increase in the spatial resolution of the model.