I don't know how you could get that. You can never get more then two UTF-8 bytes for a Unicode character. When I run that code the bytes in the resulting array are -16, -97, -104, -125. But that's the decimal representation assuming the byte value is signed. The hexadecimal string representation of those bytes is F0, 9F, 98, 83.
Okay, you're using Java 1.4, which means that you have to use the UTF-16 encoding of the character (as you did) rather than using the character directly, which Java 5 allows you to do.
At any rate it seems that you are generating something which appears to be a UTF-8 version of that character in some way, at least it converts back to the character via new String(bytearray, "UTF-8"). However I still think you need to explain your original problem, rather than trying to discuss a (possibly) failed solution to that unknown problem.
Let me be more clear, then. The problem I am asking about is the problem to which "I must obtain the result of the third column, strarting from the value of the first one" is your idea of a solution. There may be better ways of solving that unknown problem, but we can't know until we know what that problem is.