How to Give Up Correctly

First uploaded on 2023/09/29
Last updated on 2024/12/08
Copyright(C)2023-2024 jos <jos@kaleidoscheme.com> All rights reserved.

In writing this site, I try to avoid quoting articles by authors other than myself as much as possible, because I claim the copyright on this site as a whole. However, I would like to quote here a part of a statement published by an author other than myself. The reason for this is that I think it represents the thinking of meteorologists working on cumulus parameterization to date.

"Although the importance of cumulus parameterization for the practical problems of weather and climate prediction has always been well recognized, the scientific basis for cumulus parameterization has not attracted sufficient interest on the part of the modeling community. Due to this imbalance, cumulus parameterization has not become a mature scientific subject yet, ..."

REVIEW ARTICLE
The Cumulus Parameterization Problem: Past, Present, and Future
Arakawa (2004)

He recognizes the importance of cumulus parameterization, but admits that it does not currently work at all. However, he does not claim that the current approach to cumulus parameterization is itself at fault, but that the reason is the lack of a sufficient scientific basis, observations and theory, for the DD method. But is that really the case?

When he wrote that review, we had already published our first paper on RDC. Unfortunately, he did not seem to realize that RDC is a new approach that replaces the traditional dynamical method.

[NOTE]
Since this article is 20 years old, you might think that things have changed by now. However, if our RDC theory is correct, their DD approach should not have made much progress, even over such a long period of time.

[NOTE]
For the record, we have no intention of attacking or criticizing the DD research community, much less Dr. Arakawa. We are simply discussing the fact that the DD method is physically incompatible with the RDC scheme, which we advocate throughout this website.

In fact, 30 years ago, I was in exactly the same situation that they are still in. In an atmospheric physics lab that I was lucky enough to get into, I was assigned the topic of radiation-convection interactions to contribute to a WCRP project at the time. I investigated the correlation between physical quantities related to cumulus convection and mass/heat/water vapor transport in the equilibrium atmosphere, obtained by long-time integration of a numerical model, DCM, of the radiation-convection atmosphere. However, I could NOT find ANY significant results between them AT ALL. While the students around me were producing one result after another on mesoscale phenomena using more complex dynamical models, I could not explain the transport from the inside to the outside of cumulus clouds dynamically, even though I was dealing only with moist convection together with upward longwave radiation transfer in a comparatively simple two-dimensional vertical atmosphere.

I tried every analysis method known to me and still nothing worked. After more than half a year of analysis, I GAVE UP. DCM was the first numerical model I had ever created, so there must have been a mistake somewhere. I thought that the fact that I was able to integrate the radiation-convection model in 10-second increments over 100 years must have been a fluke. Nevertheless, I was satisfied because I had mastered the basics of the CFD (Computational Fluid Dynamics) technique, which was my primary goal. Now that I have mastered CFD, the most complex of the numerical methods in physics, I should be able to work with most other numerical models in the future. I was sure that I would be able to find a place to work with this skill somewhere else.

[NOTE]
In our lab, our advisor had an unspoken rule that we had to build our own numerical models from scratch. There was no inheritance of models. This meant that the user of the model coincided with the implementer of the model and was familiar with every detail of the model. At first glance, this policy seems to reduce the short-term efficiency of research, but we can now see that it has the highest educational efficiency and increases the long-term efficiency of research.

I had two advisors at the time, specialists for the atmospheric dynamics and planetary thermodynamics, respectively. So I decided to report honestly to them,
"I can't find any correlation between the dynamical and physical quantities of cumulus convection and the transport around it."
The great thing about two guys was that instead of accusing me of incompetence, they gave me advice that opened up new avenues,
"If it's so hard to analyze from inside the cumulus, why not from the outside?"

[NOTE]
Sadly, they have passed away young. One of them, when he was still alive, advised me to explain our RDC theory to the world in some way. This motivated me to publish this website, even though it took so long.

The previous method of studying the transport from the inside to the outside of cumulus clouds focused on the dynamical quantities inside the cumulus clouds during convective events, but was changed to focus on the physical quantities outside the cumulus clouds. Since there is almost no difference between the presence and absence of convective events outside the cumulus, I decided to ignore the convective events and study the equilibrium state, the time mean field. What happened then? A very clear correlation appeared between the mean field outside the cumulus and the transport from the inside to the outside of the cumulus! The flow field could be consistently explained, starting from the radiative cooling. There were no mistakes in the implementation of DCM, and it was correct that the analysis of the cumulus dynamics yielded no results. The outflow from the inside of the cumulus cloud to the outside was caused by a mechanism quite different from the dynamic mechanism that everyone believed. Thus, the RDC theory was constructed from the point of complete abandonment. If I had stuck to the dynamical properties of the cumulus interior, I would still be repeating the same analysis without any results today, and this site of commentary on the RDC theory would not exist.

Let us return to the current cumulus parameterization. Although they admit that the reality of the cumulus parameterization is not as expected, they still believe that their dynamical approach itself is correct. They say that the reason it does not work is that observations and theory are not yet sufficient. They seem to be clinging to a method that doesn't give the right answer. How can they be so confident?

Many of them are experts in meteorology, especially in meteorologic dynamics. As you know, meteorologic dynamics has achieved a great deal and is practically used to predict disasters. At the synoptic, meso, and cumulus scales, the respective dynamics are almost well established at the operational level. It would be strange not to think that the transport from the interior to the exterior of cumulus clouds can be explained by the same meteorologic-dynamical methods. Their confidence in climatology comes from their successful experience.

But let us reframe the problem. Weather forecasts determined by the motion of high and low pressure systems are understood in terms of synoptic dynamics, torrential rainfall in the mid-latitudes is understood in terms of mesoscale dynamics, and precipitation forecasts associated with tropical regions and summer cumulus clouds are understood in terms of cumulus dynamics. Each system of dynamics is valid only within the range of the physical scale it covers and is not applicable to phenomena at other physical scales.

In contrast, the transport from the inside to the outside of a cumulus cloud is different. The source of the outflow is on the cumulus scale, but the outflow itself has a wide mesoscale or synoptic-scale range. They try to understand the outflow by linking it to the dynamics of smaller cumulus clouds. This is fundamentally wrong. The direction of flow over different physical scales is determined by the second law of thermodynamics. In nature, the peak of the energy spectrum moves from larger wavelength phenomena to smaller wavelength phenomena, never the other way around. In other words, when transport occurs from within a cumulus to the outside, it must be determined by a large area of the atmosphere over which the outflow occurs, and never by the forcing due to cumulus convection, which has only a small time and space scale.

The problem is that many researchers have used the vague term cumulus parameterization to describe all attributes of cumulus. Even in the quote at the top, the term is used to describe parameterization for both meteorology and climatology. The motion of the disturbance itself, as well as the wind and precipitation that occur within it, are meteorologic phenomena and can already be explained by the dynamics at the respective physical scales. Only the transport from the inside to the outside of the cumulus cloud is a phenomenon that spans different physical scales and should be treated as a climatological phenomenon (transport in the equilibrium atmosphere, due to RDC) and clearly distinguished from the others (transport by dynamical forcing, due to dynamics at the respective scales).

The DD method has been worked on by the elite of meteorologic dynamics for half a century, with no hopeful results. It's a mistake to assume that even if they can't do it now with the DD method, they will be able to do it someday with the same method, or someone else will. If they can't do it now, no one will ever be able to do it in the future. They should be honest about the results and acknowledge the incompetence of DD. They should stop making assumptions and immediately abandon the DD method and their sense of mission altogether. It's the same story as the fact that although various substances could be made by chemistry, gold could not be made by the chemical method of alchemy. Making gold would require nuclear fusion, a non-chemical process. Sometimes the right way is something completely different from the way it has always been done. Isn't that what makes science so interesting and exciting? Insisting on a wrong method will not produce any results and will hide the right way from us.

I am not attacking the cumulus parameterization researchers. I am advising them as a pioneer who made the exact same mistake. Fortunately, I was able to give up correctly. We were able to explain the cumulus outflow using the RDC mechanism by abandoning the dynamic property explanation. Why not do the same? This is not an analogy. I am talking about the exact same physical object, the flow out of the cumulus cloud. It is not because the time is not right that they are not getting the right answer. It is because the method itself is wrong. I hope that some of you will be wise enough to correctly give up the DD method and try the RDC scheme to find the answer.