A Temperature Measurement Issue

One of the major issues in the climate change discussions is the reliability of the historical temperature records. The sceptics pounce on this one, and frankly, I have to give them some credit, because I've spent a good deal of time looking into the temperature data, and understanding some of the nuances isn't necessarily easy. In fact, I'm not convinced that the scientific community at large understands this as well as they need to. This is not to say that I think the carbon issue is a non-issue, but I would not be surprised if a significant part of the measured warmth is not due to greenhouse gases but to other things with the measurement problem being one of a myriad of outstanding analysis wrinkles.

The graph above is a rather extreme case in point. This is a plot of the outside temperature data as measured at our house and nearby. The light blue line is the temperature measurement from the thermometer associated with my generator, and a reading is taken every 15 minutes. The outside temperature is measured mainly to make sure the fuel balance is correct. I'm using the data to help understand the expected heating load.

The data are from 31-Mar-2009. Last spring I was outside and remember seeing that the sun was shining directly on the box that is attached to the back of my house and which contains the temperature reader. Most of the year this side gets shaded by three trees, but in the spring, prior to the leaves coming, the sun will peek through to the back side of the house in the late afternoon. When that happens, the process of radiative heating warms up the box beyond the ambient air temperature. I remember at the time thinking I needed to check the data for this, and well, I guess the effect is pretty big.

The other data are from the Weather Underground site. There are data from literally thousands of personal weather stations available there now, and I picked the closest stations to my house that had data on the appropriate date. There are two green-colored datasets, one of which is missing most of the data (and is the straight line across the plot). The second one is closer to the beach, and that is the primary reason it is lower in temperature. The others are further from the shore and more representative of our experience--except for the late afternoon when the sun hits our thermometer.

Alas, another thing that needs to be checked and corrected.

Just for the record, no I don't think all the NOAA (and other scientifically oriented) thermometers drift by as much as 20 degrees F, but I would not completely rule out issues at 1/20th that size, and this comment is based on not this particular graph but on my own experience dealing with NOAA data as well as other International data. Yes, there is a whole sub-industry involved in this academically, and yes they do good work, but I don't think the issue is as closed as many believe it is.