After a lot of testing and also reading through the forum.
I may figure out there a lot of parameter can fluence to the results, but still can not to figure out how to have a “perfect trend”.
And also the interpolation method and DEM model can influence to the results.
SRTM 3Sec or SRTM 1Sec, BICUBIC or BISINC_5_POINT or BILINEAR_INTERPOLATION.
Some of them can be considering:
setparm(‘unwrap_time_win’, 24)
setparm(‘unwrap_grid_size’, 10)
setparm(‘unwrap_gold_n_win’,8)
setparm(‘scn_time_win’,50)
setparm(‘unwrap_gold_n_win’,32)
setparm(‘merge_resample_size’,4)
My other result
Finally, I found a discussion that may help to improve results.