see http://entroplet.com/context_salt_model/navigate for interactive model

CSALT model posts

1. CSALT model
2. CSALT and SST corrections
3. Detailed Analysis of CSALT Model
4. Climate Variability and Inferring Global Warming
5. CSALT with CW Hybrid
6. CSALT Ju-Jutsu
7. Orbital forcings in the CSALT model : explain the pause?
8. Tidal Component to CSALT
9. CSALT and the Hale cycle
10. The Cause of the Pause is due to thermodynamic Laws
11. Reverse Forecasting via the CSALT model
12. Projection Training Intervals for CSALT Model
13. Relative strengths of the CSALT factors

FAQ

Here is a basic criticism of CSALT:

”
David Springer | November 20, 2013 at 7:52 am |

CSALT is not a model. A model would reproduce the Southern Oscillation Index history over 130 years not use it as an input. All you are doing is taking a temperature proxy (SOI) and converting it to a temperature.
“

This is instructive and we can use CSALT to explain what a physical model is and how a model can take various forms.  One such model is the ideal gas law. It can be written as:

$$ nRT = PV $$

where

n = moles of gas
R = gas constant
T = temperature
P = pressure
V = volume

The differential form of this is

$$ d(nRT) = d(PV) $$

expanding this

$$ nR dT = V dP + P dV $$

or

$$ dT = 1/(nR) * (V dP + P dV) $$

What this model says is that we can reconstitute the slight variations in temperature via monitoring the slight variations in pressure and volume.

This is essentially what the CSALT model does, but instead of the ideal gas law, we use a more general expression that is closer to a Gibbs free energy formulation described via a variational approach:

The precise structure does not matter as what we are trying to show is how the variation of temperature is reflecting the change of free energy within the system. This emerges via other intensive and extensive parameters of the system, which constitute the CSALT parameters.

Here is another basic criticism of CSALT:

”
NW | November 20, 2013 at 2:54 am |

.. you’ve referred to the high r (or r-squared) of these models several times, as some sort of support for the model. I think this is a mistake, especially in time series situations (which is your situation) where the dependent variable and one of the regressors (temp and co2 in your situation) both have a strong trend. Do you know the famous paper by Granger and Newbold on spurious regression? It is a classic statistics masterwork. Everyone who works with time series data should read it:

http://wolfweb.unr.edu/~zal/STAT758/Granger_Newbold_1974.pdf

Yes, this is published 1974 in an economics journal, but that’s because economists were making this mistake frequently at that time. Both Granger and Newbold’s PhDs were in statistics.
”

The piece that NW is missing is that CSALT is not fundamentally a time-series model. Look at Springer’s criticism, where he doesn’t even call it a model.

It is in fact a variational approach to solving the earth’s average temperature in terms of its governing parameters. Aggregating from every point in time, CSALT is solving the contribution of the various parameters. Any filtering is done to model the lags and latencies that may exist, which is useful.

Another aspect which is overlooked is that the correlation coefficient is used to compare one model against the other. So if you can come up with a better model, we can use your choice of model criteria and figure out which works better. We van use an evaluation criteria such as AIC or BIC.

Also, this points to the danger of using economics models instead of strong physical models.   The paper by the economist Beenstock  (M. Beenstock, Y. Reingewertz, and N. Paldor, “Polynomial cointegration tests of anthropogenic impact on global warming,” Earth System Dynamics, vol. 3, no. 2, pp. 173–188, 2012.) was debunked by  Hendry and Pretis

“We demonstrate major flaws in the statistical analysis of Beenstock et al. (2012), discrediting their initial claims as to the different degrees of integrability of CO2 and temperature”

] D. Hendry and F. Pretis, “Comment on‘ Polynomial cointegration tests of anthropogenic impact on global warming’ by Beenstock et al.(2012)–Some fallacies in econometric modelling of climate change,” Earth System Dynamics Discussions, vol. 4, no. 1, pp. 219–233, 2013.

Rationale

The CSALT model works very well to determine the underlying contribution of the CO2 control knob to recent temperature rise. The contribution of pressure, via SOI, is obviously limited, as it cannot sustain a differential for long, being that it is based completely on atmospheric density and atmospheric height.

So given that the temperature is still rising and the average pressure (i,e SOI) has not gone anywhere, something else is causing the warming. That is likely CO2 as the other parameters of the CSALT model, aerosols, LOD, and TSI are not going anywhere either.

The CSALT model is also so simple that I would teach it as the basis of explaining how one could construct a basic reanalysis model.