CAJ | 학술논문

将集合预报成员不等权重思想与集合卡尔曼滤波(EnKF)同化方法相结合,利用集合成员的离散度作为权重因子,对EnKF算法优化后的集合成员采用不等权重取平均值,作为同化后的预报值.首先检验了集合离散度和预报误差的相关性,证明将集合离散度作为权重因子的可靠性;利用一个水文过程模型(DHSVM)和实测数据进行了土壤水分的同化变权实验,对EnKF分析和更新后产生的土壤水分集合,分别采用算术平均和变权平均的方法,计算土壤水分预报结果并进行比较.实验表明,集合变权平均法可以进一步提高同化的预报效果.
장집합예보성원불등권중사상여집합잡이만려파(EnKF)동화방법상결합,이용집합성원적리산도작위권중인자,대EnKF산법우화후적집합성원채용불등권중취평균치,작위동화후적예보치.수선검험료집합리산도화예보오차적상관성,증명장집합리산도작위권중인자적가고성;이용일개수문과정모형(DHSVM)화실측수거진행료토양수분적동화변권실험,대EnKF분석화경신후산생적토양수분집합,분별채용산술평균화변권평균적방법,계산토양수분예보결과병진행비교.실험표명,집합변권평균법가이진일보제고동화적예보효과.
The arithmetic average of the ensemble members is usually taken as the final assimilation result of Ensemble Kalman Filter (EnKF).In this study,the weighted average of the forecasting ensembles was used as the assimilated result of EnKF.The dispersion of the ensemble members was considered as the weighting factor.After the EnKF has analyzed and updated each ensemble member,the weighted average of the ensemble members was adopted as the final forecasting results.Since the relationship between the dispersion of ensemble members and the forecast errors was significant,taking the dispersion of the ensemble members as the weighting factor was reasonable.This approach was examined based on a soil moisture assimilation experiment,which assimilated the in situ measurements of the surface soil moisture with a distributed hydrologic soil vegetation model (DHSVM) using EnKF.The results of the weighted average were compared with the results of arithmetic average based on EnKF.This study indicates that the assimilation can be improved significantly using the weighted average of ensemble members.