一个使用mv和参数扩展的bash shell循环可以做到这一点：

for file in *:[[:digit:]][[:digit:]][[:digit:]][[:digit:]][[:digit:]][[:digit:]][[:digit:]][[:digit:]][[:digit:]][[:digit:]]
do
  prefix=${file%:*}
  suffix=${file#*:}
  mv -- "${file}" "${prefix}:${suffix:0:4}-${suffix:4:2}-${suffix:6:2}_${suffix:8:2}"
done

此循环将选取与模式匹配的每个文件：

• *--任何内容
• :--冒号
• [[:digit:]][[:digit:]][[:digit:]][[:digit:]][[:digit:]][[:digit:]][[:digit:]][[:digit:]][[:digit:]][[:digit:]]-- 10位数字

...然后通过在所需位置插入短划线和下划线对其进行重命名。
我仔细地选择了循环的通配符，这样它就可以匹配“输入”文件，而不是重命名的文件。如果实际的文件名有边缘情况，导致通配符失败（并因此再次重命名文件），请根据需要调整模式。

# !/bin/bash

strindex() { 
  # get position of character in string
  x="${1%%"$2"*}"
  [[ "$x" = "$1" ]] && echo -1 || echo "${#x}"
}

get_new_filename() {
  # change filenames like:  SURFACE_OBS:2019062218
  # into filenames like:    SURFACE_OBS:2019-06-22_18

  src_str="${1}"

  # add last underscore 2 characters from end of string
  final_underscore_pos=${#src_str}-2
  src_str="${src_str:0:final_underscore_pos}_${src_str:final_underscore_pos}"

  # get position of colon in string
  colon_pos=$(strindex "${src_str}" ":")

  # get dash locations relative to colon position
  y_dash_pos=${colon_pos}+5
  m_dash_pos=${colon_pos}+8

  # now add dashes in date
  src_str="${src_str:0:y_dash_pos}-${src_str:y_dash_pos}"
  src_str="${src_str:0:m_dash_pos}-${src_str:m_dash_pos}"
  echo "${src_str}"
}

# accept path as argument or default to /tmp/baz/data

target_dir="${1:-/tmp/baz/data}"

while read -r line ; do
  # since file renaming depends on position of colon extract 
  # base filename without path in case path has colons 
  base_dir=${line%/*}
  filename_to_change=$(basename "${line}")
  echo "mv ${line} ${base_dir}/$(get_new_filename "${filename_to_change}")"

  # find cmd attempts to exclude files that have already been renamed 
done < <(find "${target_dir}" -name 'SURFACE*' -a ! -name '*_[0-9]\{2\}$')