make incorrect values as nan

I have a data set of temperature which has an usual range between 0 and 20 degreesC. Some of the data is incorrect and gives readings which are obviously wrong. So, I want to write a command which states that if a reading is 1 degreeC different to the one prior to it, the reading should be changed to a nan.

How is this possible?

Answer 1

[Andrew Newell]

Here is one way:

I = find(diff(T)>1);
T(I+1)=NaN;

EDIT: And here is another using logical indexing:

T([1 diff(T)>1]) = NaN;

[Walter Roberson]

Possibly >= instead of > for the comparisons.