Since a significant number of negative stations are available I decided to combine them and see how the various models would deal with this contrary data set. The normalized, composite trend for this data set indicates a -0.07 C/decade temperature drop in the default 1950 to 2013 time frame. As expected, most of the climate models made positive trend predictions and did not provide a good result relative to the negative composite data. But, it was interesting to discover that several of the AR5 models predicted negative temperature trends with the Historical Natural scenario. The attached image shows a screen shot of the Climate Time Series Browser with these negative trend predictions. One model even predicted a negative trend with the Historical model scenario. The prediction for the GISS-E2-H model in this scenario was -0.0015 C/decade. It would be interesting to understand more details on the specific models to gain insight into how they make predictions for such an unusual data set
There is likely no wide significance to any of these findings. Looking at this unusual data set was helpful in increasing my understanding of how the climate browser application functioned. It was also interesting to consider how biased selection of weather stations might impact the model results and potential conclusions on temperature trends.
Image of negatively trending US meteorological stations and negative model predictions.
Image of WV meteorological stations with negatively trending sites (blue and green) located along the Ohio River.
