Modern  data  assimilation  techniques  represent  a  way  to  combine  the  numerical  model and observations together for an optimal estimation of the state of the earth system.  In addition to their vital  role  in  numerical weather prediction and climate  reanalysis,  data  assimilation techniques can  also  be  applied  in  many  aspects of  climate  study,  such  as model  initialization, validation and optimization.  The  paper  gives  a  brief  discussion on the  applications  of  data assimilation in climate modeling with emphases on a few key issues that are closely associated with  the  regional  climate  study  over  the  western  China.  Challenges,  potential  methodologies, recent results and future development are presented.