Changing matrices into single lines of data

I am sure there is an awk command that can do this, I am not skilled enough in awk to come up with a solution. In the meantime, you could use something like this:

#!/bin/bash

csplit -z tosplit /661/ {*}

for file in xx*; do
    newName=$(egrep -o '2[0-9]{7}' $file)
    mv $file $newName.txt
done
rm -rf xx*

Where tosplit is this file (your example file):

661###############20160315###
###########################
###########################
661###############20160316###
###########################
661###############20160317###
###########################

After running this script (in the same directory as the tosplit file) I get three files:

ls 2016031*
20160315.txt  20160316.txt  20160317.txt

...looking like this:

cat 20160315.txt 
661###############20160315###
###########################
###########################

cat 20160316.txt 
661###############20160316###
###########################

cat 20160317.txt 
661###############20160317###
###########################

You can possibly(?) use csplit to name the files as well, but that too is above my paygrade!