Why does GATK get so many wrong SNPs?

[SoFoie]

I used GATK4 detection to obtain a large number of SNPs, but I found many wrong SNPs when I randomly selected some SNPs to verify the accuracy of the detection. For example, I got a site (Chr01_202655, alleles: C/T) in sample 1 which is C. But when I check it in the sorted bam file using the "samtools tview", I find that it is a heterozygote site (alleles: Y). And these wrong SNPs found are often concentrated in some areas. I don't know why this is happening?