Operational National Weather Service (NWS) meteorologists who issue forecasts and warnings currently employ two dimensional displays of analyses, remote sensing imagery and numerical model output via the Advanced Weather Interactive Processing System (AWIPS). Although the graphical displays provided by AWIPS are much advanced from earlier methods employed by meteorologists (hand drawn/plotted maps and facsimile charts), the method of visualization has changed little since shortly after the turn of the century and remains two-dimensional.
In contrast to operational use, the research community (among others), has been actively using 3D visualization applications for some time. An example is the University of Wisconsin-Madison Space Science and Engineering Center's Vis5D which has been in use in one iteration or another for nearly a decade.
Forecast Systems Laboratory (FSL) in Boulder, CO has been supporting development of 3D visualization software and applications since 1990, though until recently their emphasis as been on research application (McCaslin et al., 2000). This changed in 1997 as the move was made at FSL to develop the Display Three-Dimensional (D3D) software to determine if 3D visualization of data sets and variables from meteorological numerical models would offer added value in an operational forecast setting. Real-time forecast exercises utilizing this system were conducted in 1998 and 1999. This author participated in the latter exercise.
Subsequently, the D3D software was considered mature enough for limited distribution to a few National Weather Service (NWS) field offices where sufficient prior knowledge of the software existed. This has allowed the software to be used and evaluated in a real-time forecasting environment at the NWS office in Dodge City, KS.
A number of issues surfaced once the software was in place in an operational environment, the most alarming of which was difficulty in getting operational forecasters to actually use the new application for numerical model interpretation during the forecast process. This is partly due to natural human resistance to change, but also due to the more mature and thus more refined two dimensional methods of data interrogation which define the present paradigm.
This talk will center on the challenge this change of paradigms brings to the operational environment. Discussion will follow concerning; why this transition will be difficult, what can be done to mitigate the difficulty, why it must occur, and what the future of meteorological visualization might look like.
2. Examining the Paradigms
Capra (1988) notes that the term "paradigm" in science represents "a conceptual framework shared by a community of scientists and providing them with model problems and solutions". He also refers to paradigm as, "the totality of thoughts, perceptions, and values that form a particular vision of reality - a vision that is the basis of the way a society organizes itself". As such, paradigms are not necessarily the truth, but are rather models of reality within a frame of reference. When that frame of reference changes, it is called a paradigm shift.
The old frame of reference for operational meteorologists, reinforced by many years of repetition, has been 2D visualization of 3D and 4D conceptual models of the atmosphere. Nearly all operational meteorology methods rely upon this tenet despite meteorologists certain knowledge that the atmosphere is a 3D entity and that forecasting involves 4D integration of the same. The bulk of meteorological texts, teaching, and training materials are steeped in this tradition.
When confronted with 3D visualization of the atmosphere then, it is small wonder that many operational forecasters find themselves facing a vision of reality which is outside of their experience and education. This is a new paradigm and thus a paradigm shift is imminent.
Kuhn (1970) observes that when paradigms shift, they are inevitably met with resistance. He provide the following very human reactions to evidence of a new paradigm;
1) We are literally unable to "see" the evidence of the new paradigm
even though it is right in front of our "eyes" (ie. The evidence does not
fit into the old framework so it is psychologically denied.).
2) We see the evidence but rationalize it away.
3) We see the evidence and distort it to fit our old paradigm.
4) We see the evidence but declare that it is inaccurate (ie, Perhaps something is wrong with the tools of measurement or with the person presenting the new information).
Furthermore, when a paradigm shifts, everyone goes back to zero, thus our sanity is often riding upon our paradigms and we are reluctant to give them up without a fight (Barker, 1992).
Human Reactions to Change
Whatever kinds of change people encounter, there are certain patterns of response that occur and re-occur. Blanchard (1992) enumerated what he sees as the seven dynamics of change which seem imminently applicable to this particular shift in paradigms. They are worth listing here, as an understanding of them is critical for managers and trainers in assisting people change paradigms with a minimum of stress. They are:
I. People feel awkward or ill-at-ease and self-conscious.
II. People initially focus on what they have to give up.
III. People will feel alone even if everyone else is going through the same change.
IV. People can handle only so much change.
V. People are at different levels of readiness for change.
VI. People will be concerned that they don't have enough resources.
VII. If you take the pressure off, people will revert to their old behavior.
Unless these social dynamics are mitigated then, change to a new paradigm is unlikely regardless of its potential value to the science and the scientist.
In his on line article concerning "Real Change", Rick Tate, of the international consulting firm "Innovative Thinking, Inc.", warns that, "Unless we fully understand the theory, set of assumptions or thinking that we held true when we created practices and procedures that we use presently, we will be forever condemned to create different versions of what we have always done in the future. The result... no real change, just different manifestations of what we always used to do." Sadly, this seems a very apt description of the current state of 2D visualization of atmospheric models in meteorology.
In short then, it is a normal human reaction for operational forecasters faced with a change such as 3D visualization (however much greater its potential may be) to counter it by reinventing more methods of traditional 2D visualization.
4. Mitigating Change
Thankfully, there are available resources for mitigating social and organizational change of this nature. Indeed, there has very recently emerged an entire body of research and study dedicated to the social aspects of computerization known as Social Informatics (SI). As recently as 1992, Greenburg and Thimbleby characterized the science of Human Computer Interaction (HCI) as a "weak" science in that it showed a sparsity of theories, risky hypotheses, and difficulty in substantiating experiments through replication. Since then, however, any number of societies have emerged (Society on Social Implications of Technology, SSIT; International Federation for Information Processing, IFIP; Special Interest Group on Computers And Society, SIGCAS) which are dedicated to the study of human interaction with technology. Much of this rubric is so new as to be virtually unknown to most meteorologists and to most management within meteorological organizations. The sobriquet SI itself was only defined in early 1996!
This study is in and of itself a new paradigm with SI and HCI scholarly literature published largely on line rather than in traditional paper journals or volumes. Nonetheless, there is a considerable amount of material available today regarding human transition to new technologies and in particular new paradigms in visualization.
Bob Lewis (2001), a management consultant for the Perot Corporation provides an excellent essay on facilitating change within an organization. He notes that managers, who are in the best position to facilitate change, are often the prime offenders. They often perceive change as a threat to their positions. Employees resist change largely due to the absence of information. In order to facilitate change, he says, "All you have to do is make sure the average employee will benefit from it, communicate 'a lot', involve them, and keep change-resisting managers where they can't sabotage the effort."
In the case of 3D visualization, then, it would appear that frequent encouragement and involvement of operational forecasters in the new idiom should be the keys to changing the paradigm. From the preceding thoughts it is evident that this level of commitment can only be achieved providing management strongly endorses it. But management need not be the only facilitator in moving into this new paradigm. The change can be greatly assisted by the educational community in exposing meteorology students at all levels to as much 3D visualization as possible during the teaching process. This will also help clarify in the minds of the future forecasters a true 3D visualization of the conceptual models of atmospheric processes rather than relying upon the old paradigm of mentally constructing such models from numerous 2D "slices".
5. Strengths and Weaknesses
With any emerging technology such as 3D visualization, there are both strengths and weaknesses. D3D is not exempt from this. However, as we have seen over and over in meteorology (and most other sciences), as the paradigm develops, it is continually enhanced with innovative improvements which are largely driven by the users themselves.
Current weaknesses lie in such areas as (but are not limited to); increasing resolution of the visual displays, optimization of physical resources (the users interaction with the data sets), finding the most effective ways to visualize numerical and non-numerical information, finding the most efficient methods of direct user manipulation of the visualizations and data sets, and finding effective ways to simultaneously share the visualizations over numerous platforms in a network environment (Hibbard, 1999).
In using the D3D operationally, this forecaster finds one of its greatest strengths to be displays of meteorological parameters (both measured and derived) in the form of isosurfaces throughout the atmospheric volume.
For example, traditional interrogation of vorticity advection in the 2D paradigm requires the user to evaluate a number of 2D plan view representations in order to determine if the vorticity advection is increasing or decreasing with height (thereby determining its contribution, negative or positive, to vertical motion). On the other hand, an isosurface of positive vorticity advection in the D3D quickly evaluates this parameter in a single image. Best of all, a time lapse of the same shows the forecaster in a matter of moments its 4D aspects. This tremendous strength of the 3D visualization process also holds true for many other derived meteorological variables such as omega, equivalent potential vorticity, thermal and moisture advection, just to name a few.
Among the other major strengths of the D3D over the D2D noted by McCaslin et.al. (2000) are; the ability to quickly construct 2D slices and to effortlessly move them throughout the volume, and the ability to display atmospheric sounding variables throughout the volume using a sounding tool which can be quickly moved throughout the volume.
This forecaster is convinced that the primary effort at this time should be to;
1) Get a usable version of the software in front of operational forecasters.
2) encourage them to use it and document their impressions in order to begin further developing the paradigm.
6. What the Future Holds
The future of operational meteorology is bright indeed if one allows the imagination free rein. Although 3D visualization is relatively infantile, a bit if exposure to it opens hitherto undreamed of possibilities.
As time allows the inevitable improvement in technology to resolve the current problems outlined by Hibbard (1999), several startling new concepts come immediately to mind.
1) New and more effective methods of user manipulation of the data sets will emerge. Virtual goggles and gloves come immediately to mind as replacements for the age old keyboards and mice. Touch screen systems are also a possibility. Without a great deal of difficulty (and given workstation hardware capable of the task), it is possible to imagine a forecaster virtually entering a 3D model forecast atmosphere, manipulating the initialization to better fit observational data (also displayed within the 3D volume) and then re-running the atmospheric model locally and in real-time (and iteratively) until it arrives at the desired final result.
2) As the NWS moves into the concept of a common digital data base through IFPS, it is easy to envision collaborative sharing of a 3D version of this data base between many forecasters in real-time. They should be able to do so while simultaneously conferring amongst themselves over the desirability of various changes made. This will, of course, await the development and implementation of networks with sufficient band-width to support massive amounts of simultaneous data, voice and video transfer in real-time.
3) As difficult as the above changes may be to accept, an even more difficult hurdle will appear (at least in the view of the author). As 3D visualization and data set manipulation lead to more automated output of various forecast products, and as the forecasters become more deeply immersed within the virtual volume to produce their final product, management will have less and less hard evidence of the productivity of the employee. It is conceivable that at least some of the traditional roles for managers will disappear. This will be a bitter pill for middle and upper level management to swallow and will likely be firmly resisted. But that represents an entirely different paradigm shift which beyond the scope of the present discussion.
What should be clear from the above discourse is that the science of computer 3D visualization has already presented meteorologists with intriguing if somewhat revolutionary new ways of doing their jobs. What should also be clear is that 3D visualization of traditional meteorological variables represents a true shift in paradigm for the operational forecaster who has, as yet, been kept rather remote from its development.
Given this new paradigm, it will be incumbent upon management to take the lead in facilitating the change by deeply involving the operational forecaster in the change process and frequently encouraging that change regardless of what their own doubts may be. This will require a considerable amount of vision on the part of management, a commodity usually scarce at that level in many organizations. Educational institutions can greatly reduce the problem by exposing their students to the new paradigm early on so that new forecasters arrive in the operational environment expecting to find such methods already in place.
Existing limitations and difficulties in 3D visualization such as the D3D, as with all previous forms of computer systems and visualization paradigms, can and will mature rapidly if, and only if, they are put to use in the operational environment. This will allow the end users to become immediately involved in providing constructive feedback concerning needed improvements to the software designers.
Finally, vision of the benefits of the new paradigm and where it may lead must be kept in the forefront of all concerned. Both managers and operational forecasters must be regularly encouraged to "think outside the box".
References furnished upon request.