Like I said, put your pan of water in the pool and measure its temperature vs the temperature of the pool and tell me that you have the same evaporative rate.
Of course you don't, but then you beleive that altering past temperature data down in order to make the present appear warmer isn't terribly flawed either.
More measurements are not necessary because as I stated, you are dealing with a contained system with predictable variables. If, in reality, you had a leak, you wouild know in the first day using my method. The other two days are simply to determine whether the leak is in the plumbing or the shell. Overcomplicating a simple system is as bad as oversimplyfing a complicated system.
Of what value is a graph of temperature anomolies claimed to be accurate to tenths of a degree if one doesn't know the actual temperature to within 2 degrees? Can you say mental masturbation?
Because if you don't know the absolute temperature within tenths of a degree, you have no basis to claim warming or cooling to within tenths of a degree. If your margin of error is 2 degrees, it is rediculous to believe you can make accurate statements regarding anomolies. Again, mental masturbation. There is a reason that the models are so poor at either reflecting present reality or predicting future conditions. You are describing the thinking that goes into present modelling and you are also describing why present modelling is such an abject failure.