For now, I have simply estimated an anthropogenic component with a linear trend and a starting value. At some point in the future I might use the greenhouse gas index or some other estimate of CO2 concentration. There are several papers which have estimated the human influence on global temperature, either through modeling or by removing natural factors from temperature data.
Temperature observations come from the NASA GISTemp dataset.
Solar activity generally follows a ~11yr cycle from peak to peak which has an estimated 0.1-0.2C impact on global temperature. I calculated a mean value for TSI over the period and converted monthly averages to anomalies. TSI anomalies had an amplitude of almost 2 W/m2, so I used a short term temperature sensitivity of 0.1 C/[W/m2] for solar activity to match 0.2C impact. Foster and Rahmstorf (2011) estimated about 1 month for the temperature lag to solar activity, which I assumed for default.
The NASA page for volcanic aerosol data indicated a value of -23 for effective efficacy of AOD. This is the W/m2 change in forcing for each unit of AOD. I assumed the same short term temperature sensitivity for volcanic activity as used for solar activity. Foster and Rahmstorf (2011) estimated about 6 month lag for volcanic activity's influence on temperature.
Because ENSO is not really a climate forcing, it's effects on global temperature are more of an apparent effect. I used the same short term temperature sensitivity as with the other indeces but had to estimate the efficacy-equivalent based upon the observed temperature data used (ENSO affects each temperature dataset differently). A default value of 4 months was used for the lag based upon numerous sources, including Foster and Rahmstorf (2011).
I estimated the warming influence from human greenhouse gas emissions to be about 0.18C/decade based upon some recent papers I read and also the estimate from Foster and Rahmstorf (2011). A starting anomaly of +0.12C was used to better match the observed data.
Please note: because this is not a sophisticated climate model and slow response warming from the oceans is not simulated, the temperature sensitivities used are likely not the same as temperature sensitives estimated by other sources. The sensitivity used for this tool is for the short-term impact only, and I have seen the true sensitivity estimated as 7x the change that I use by default (roughly 0.7 C/[W/m2] instead of 0.1).
Click here to change the model parameters.
Look at the resulting graphs here.
Foster and Rahmstorf (2011) also put much more effort into finding the true global warming signal in the observed temperature record. See: http://skepticalscience.com/foster-and-rahmstorf-measure-global-warming-signal.html
Thompson (2008) also attempted to remove the climate variability from ENSO. See: http://www.skepticalscience.com/el-nino-southern-oscillation.htm