Too many singleton alignments found by bwa

[weibominingdata]

I was using bwa to map the high coverage paired-end sequenced trio data from 1000 genome. I found that singleton alignments (one end map, the other not for the paired-end read) appear along the whole chromosome 1. I did not expect this.

I am wondering that anybody has done the similar thing before? Have you observed the similar phenomenon about singleton alignment?

BTW: default parameters values for bwa were applied to do the mapping.