Project Title:

Small Business Innovative Research Phase I:

Virtual Atmospheric Laboratory

 

Final Project Report

Oct. 1998

 

 

 

 

for

National Science Foundation

 

 by

MESO, Inc

185 Jordan Road

Troy NY 12180

 

 

 

 

 

 

Award No. DMI-9760520

 

Award Dates:  1/1998   to  6/1998

 

 

 

 

 

 

 

 

1.   Phase I Objectives

                The research and educational process in atmospheric science has been hampered by the fact that the atmosphere cannot be brought into the research lab or classroom to study and manipulate in the manner that is often done in many of the other sciences.  The overall objective of this research is to provide students, at various educational and experience levels, with an inquiry-based capability to study and manipulate a virtual earth-atmosphere system.  The Virtual Atmospheric Laboratory (VAL) software system would provide a mechanism to develop problem-solving skills while bringing a unique experimental capability to the atmospheric sciences.

                A primary objective of the Phase I research was to establish the technical and educational feasibility for many educational levels of the VAL.  A major component necessary to meet this objective was building and evaluating a working prototype.  The secondary objective of Phase I was to demonstrate the feasibility of porting the VAL from UNIX to Windows and Macintosh platforms.  The major objectives of the Phase I research were divided into the following specific task objectives:

(1)   establish educational objectives and associated requirements for a university prototype VAL;

(2)   build a working conceptual prototype of the VAL to operate on a university UNIX workstation;

(3)   estimate the instructional utility of the VAL concept by evaluating it as a teaching tool in a university undergraduate meteorology course;

(4)   determine the computational feasibility of hosting the VAL on low-cost consumer-oriented computational platforms such as the Apple Power Macintosh and Intel Pentium-based PCs running a Microsoft Windows operating system;

(5)   estimate the instructional utility for the middle school, high school, non-science major at the college level, and professional forecaster training by a group evaluation of the prototype by instructors at each of the educational and training levels;

(6)   determine the educational objectives and associated functionality to be included in the fully-developed commercial version of the VAL for each educational level; and

(7)   design the functional and software structure of the commercial VAL.

2.   Research Summary

2.1 Description and Findings Objectives 1 - 3: University Prototype VAL      

2.1.1 Description of Research

The first three objectives were addressed through the development and evaluation of a working VAL prototype for the Synoptic Meteorology Laboratory course taught by Dr. James Moore at Saint Louis University.  First, Dr. Moore, his graduate teaching assistants and MESO personnel established the educational objectives and associated requirements for the university-level VAL prototype.  This included identifying the meteorological concepts that would be taught and developing experiments and experimental designs to address those concepts.  For the students using this prototype, the specific educational objectives involved understanding the parameters that play a critical role in two fundamental meteorological concepts: (1) the formation and evolution of a thermally direct circulation, such as a land/sea breeze; and (2) the geostrophic adjustment process. 

Second, MESO designed and built a working prototype of the VAL.  It was installed on the UNIX-based Saint Louis University Meteorology computer laboratory workstation network.  The prototype VAL software system was an integration of: (1) an idealized data ingestor; (2) a set of databases describing basic properties of the earth's surface such as terrain elevation, land/water distribution, land use class, soil texture class, climatological density of vegetation, etc.; (3) an input data exploration module;  (4) a complete 3-D mesoscale model; (5) a model experimentation module; and (6) an experiment visualization module.  The working prototype contained all the conceptual features that will be included in the fully-developed system.  However, it only included a subset of the sophisticated model components, interactive options and educational levels envisioned for the fully-developed system. 

Third, the students investigated the formation and evolution of a thermally direct circulation by using the VAL.  This was done as part of the in-class instruction for a two-week period.  An evaluation, by both the instructors and students, was done to assess the utility of the prototype VAL and to suggest needed improvements to make the VAL more useful as a teaching tool.  The geostrophic adjustment process on the VAL was investigated and evaluated by both MESO and Dr. Moore.

 

2.1.2 Findings of Research Objectives

Overall, the prototype performance of the VAL met the educational objectives very well.  Particularly impressive was the fact that it was possible to configure the mesoscale model in such a way as to be able to run in only 30-35 minutes yet still maintain the physics and dynamics required to ensure realistic experiments.  The VAL interfaces worked well for the setup and execution of the experiments and the visualization capability met the minimum requirements to meet the objectives.  However, there is a need to make the visualization capability more comprehensive and robust.  This would be accomplished by expanding the number of parameters and visualization options that are possible in the analyzed output.  The actual student input interfaces and output can be viewed at http://www.borg.com/~sandi/val.html.

2.2 Description and Findings Objective 4:Computational Feasibility

2.2.1 Description of Research

The fourth objective was the determination of the computational feasibility of hosting the VAL on low-cost consumer-oriented computational platforms such as Apple Power Macintosh and Intel Pentium-based PCs.  There were two distinct segments in this research area.  First, the mesoscale model was ported to a variety of computer platforms by MESO scientists and benchmark testing was accomplished.  Second, the interface and visualization design requirements for a Macintosh and Windows based environment had to be determined.  This evaluation was undertaken by Yaros Communications, Inc. (YCI), the creators of Weatherschool â, who are experts in the field of middle-school PC-based weather instructional software.  YCI created a mockup Macintosh and Windows based PC version of the VAL interfaces and visualization in order to determine the feasibility as well as the tools needed to accomplish Phase II.

2.2.2 Findings of Research

YCI found that the explosion in computer power during the recent years has given rise to a number of data visualization tools, such as Director (available from Macromedia, Inc.), that permit a user to interactively examine multi-dimensional datasets and even animate their evolution.  In addition, it was determined that there are several software packages that could be used to create interfaces and visualization techniques that would allow the same interactive capability that currently exists in a UNIX-based environment.  The results indicated that the model could be scaled and configured in such a way as to offer many levels of sophistication that could be called upon as required according to the concept being taught and the instructional level of the student.  The major conclusion from the benchmarking study is that affordable computational resources are now clearly available to run the VAL efficiently for nearly every level of educator as well as the general public.  The benchmarking showed that many experiments deemed important by educators could be run on a Pentium II 300 MHz PC in less than 5 minutes.  Even quite complicated experiments could be executed in 30-40 minutes.  Only the most complicated experiments, covering large areas on the order of a continent or larger, would require runtimes of several hours.  These could be run overnight for use the next day. 

2.3 Description and Findings Objectives 5 -6: Broad Market Feasibility

2.3.1 Description of Research

Objectives 5 and 6 were addressed by determining the potential instructional utility, educational objectives and associated functionality for the middle school, high school, non-science college major, and professional forecaster training potential markets.  This investigation was done by: (1) demonstrations of a mock-up VAL, (2) interviews and/or written questionnaires; and (3) hands-on evaluation of the university prototype at a workshop at Saint Louis University by a group of instructors from each of the educational areas.  The information gained formed the basis of a design of the commercial VAL for objective 7.

2.3.2 Findings of Research  

The VAL was determined to have great potential to fill important educational and training requirements for each of the targeted areas.  It was concluded that most of the meteorological objectives and concepts are common among the various groups.  What differentiates the groups is the cognitive ability of the students comprising the groups and the required level of understanding of a given concept within each group.  The major educational objectives stated by each targeted educational group involved the student developing understandings of the primary atmospheric processes and interactions.  It was concluded that the VAL would be a useful tool to teach most, if not all, of the underlying physical processes of the atmosphere.  This was concluded because the atmospheric model used in the VAL captures all the atmospheric processes.  The VAL would be particularly effective to teach objectives that involve the integration of multiple interrelated concepts, which is often the case in the earth and atmospheric sciences.  From an educational perspective, the VAL supports the fundamental objectives of the educational community to: (1) see the student as a "problem-solver" with substantial open-ended inquiry and critical thinking skills rather than one who merely recites facts obtained from an instructor; (2) utilize the computer as a "tool" to allow inquiry-based learning to take place, and (3) integrate concepts and knowledge from a variety of fields into an interdisciplinary problem-solving perspective.  Another major conclusion is that it is desirable for educational software to allow the student to progress at a pace commensurate with the student's abilities.  This is a fundamental attribute of the VAL.

2.4 Description and Findings Objective 7: Software and Functional Design

2.4.1 Description of Research 

There were two components to the objective 7 research.  The first component involved integrating and evaluating all the findings from objectives one through six into the design of the functional and software structure of the commercial VAL.  The second component involved examining the methods needed to integrate many technological components to meet the educationally required objectives.  These components include: (1) sophisticated physics-based atmospheric models; (2) visualization tools which permit the user to interactively examine and animate complex multi-dimensional datasets; (3) graphical user interfaces which allow the user to easily navigate through complex information systems; and (4) hypertext instructional databases which permit the user to pursue threads of knowledge through the use of words and phrases which are linked to other parts of a database or even other databases.  The advanced state of each of these technologies makes the VAL a realistic possibility.  It is anticipated that the level of technology in each of these areas will continue to advance rapidly.  Therefore, it is imperative that the VAL be designed to have a flexible structure to take advantage of these future advancements.

2.4.2 Findings of Research Objective 7

The research determined that in order for the VAL to be an effective educational tool, the following design attributes of the VAL are essential: (1) it must be designed so it can be housed on a low-cost computational platform that is widely available to each class of user; it is essential the software be computationally and cost efficient on both computer workstations and low cost "consumer-oriented" platforms such a Windows PC and PowerMac; (2) it must have a sophisticated "point and click”, menu-driven graphical user interface; and (3) it must permit the user to smoothly transition from an intuitive level to a quantitative/mathematical level.

There were several major conclusions on the design of the VAL functionality and software.  The primary requirement of the VAL functionality is for the student interfaces to be intuitive and easy to use.  Because of the overwhelming number of choices possible in simulating the atmosphere, the interface needs to be designed to provide only the most relevant choices to the student.  These choices would be based on the inputs provided by the student that controls the design of a particular experiment.  For example, it was deemed reasonable to require a student to hypothesize that a specific parameter had more influence on a particular atmospheric phenomena than another parameter from a list including, but not limited to, temperature, moisture, terrain type, land use, or winds.  Then the student would be allowed to set the specifics for the experiments from tailored input windows, defaulting the values of those elements that are irrelevant to teaching the concept.  It was deemed unreasonable, as well as confusing, to require students to control every aspect of the atmosphere and surface condition for every experiment.  Another major requirement for the VAL would be to present the experimental output in the most meaningful and easy-to-visualize format to the student for a particular experiment.  This involves both sophisticated visualization techniques such as 3-D animation as well as numerical analysis of the output in the form of tables or graphs.  The specific method of presenting the material was found particularly important to the student at the middle-school level.  Students of this level typically would not have the ability to conceptualize 3-D abstract concepts without the use of highly sophisticated and tailored visualization of the experiment output.

The overall conclusions of the research found that there are significant research and development tasks required in Phase II in order to create a fully-developed commercial VAL.  From the technical perspective, research is needed in the following areas;  (1) porting the model and developing the interfaces for the three major platforms environments (UNIX, Windows and Macintosh);  (2) integrating the various levels of model (1-D, 2-D and 3-D) in order to meet the speed of execution requirements necessary to give students quick feedback on the experiments; and  (3) integrating all the display capabilities needed for the VAL to visualize the experimental output. 

It was also determined that in addition to the technical research the following educationally based research is required: (1) investigation of the issue of designing a “one size fits all” VAL that can be applied to all educational levels versus tailoring a separate VAL to meet the needs of each market niche; (2) determination of how the VAL can be incorporated into the current science programs in the educational and training market; (3) establishment of which meteorological topics, concepts and processes the VAL will add value over traditional lecture and demonstration methods; (4) determination of what graphics are needed in order to "explain" in the best way possible for the various education levels (grade 5 through graduate school); (5) an objective comparison of the understanding of certain meteorological concepts by students who are taught using traditional methods with those using the VAL as part of their instruction.

2.5 Commercial Potential

The research determined that there are three distinct markets for the VAL.  One market is primarily focused on teaching the atmospheric and related sciences at some education level from grade five through the college-graduate student.  The second market is focused on training people in applying knowledge of the atmosphere to practical problems.  The third market comprises individuals of the general public who are interested in meteorology.

The educational market comprises those individuals and institutions involved in various instructional levels in the area of atmospheric and related science.  At the college level, the potential market comprises a significant fraction of the over 3706 two- and four-year U.S. colleges (over 14 million students), since many colleges offer a general "understanding the weather" type of course.  Majors in meteorology are offered at 58 four-year U.S. colleges (Pendleton, 1994).  In addition to the college market, virtually every one of the 106,000 public and private grade school systems with over 46 million students teach weather concepts at some point in the students' education.  This means that during any given school year approximately five million middle- through high-school students study weather.  Weather is most commonly taught as a topic during grades 5-8.  However, weather is becoming an increasingly important subject at the high school level as it is being taught with increased emphasis as part of various earth science curricula.  The government reported that 23% of high school students took earth science courses in 1994.  It should be noted that unless otherwise stated all statistics are from the National Center For Education Statistics (1997) or World Almanac and Book of Facts (Famighetti, 1998).

There are two distinct parts to the weather training market: (1) the government weather training programs and (2) the private sector weather services.  The government is divided between the military weather services and the National Weather Service (NWS) which total approximately 300 locations worldwide and 6,100 meteorologists (Kutscher, 1994) that would use the VAL as part of their formal and in-service training programs.  Weather training personnel from both the military and the NWS indicated that there would be an increased emphasis on onsite training using self teaching learning tools like the VAL.  The personnel who evaluated the VAL felt that it offered a unique training tool that would shorten the learning time for new personnel who arrive on site and provide an unparalleled educational capability.  The private sector can be divided into (1) the broadcast industry with over 1092 U.S. television stations (Infoplease Almanac, 1998) that employ weather personnel and (2) approximately 1000 private weather forecasting services.  Individuals from the broadcast industry and other private weather services were interviewed and given a "laptop" demonstration of the VAL.  In all cases, the interviewees felt that the VAL, with some slight modification to meet industry needs, would be a very useful training tool for their operation.

There would seem to be potential in the general public market, but it is hard to quantify.  It is estimated that in 1997 simulation software accounted for $7,146,000 sales and scientific educational software (not specifically bought by schools) accounted for $5,254,000 (PC Data, Inc., 1997).  However, since there is no software that approximates the sophistication of the VAL, it is hard to estimate the general public VAL market.  Of note is the popularity of more general scientific simulation programs, such as Maxis' "SimEarth".  It was concluded that one way to break into all of the markets was to provide a no cost interactive but limited demonstration of the software to the public as is done with many other software products.

In addition to the value analysis, a cost analysis was performed for each of the educational institutions.  In all the cases an acceptable site license price range for the VAL was deemed to be $300-$500.  Anything over $1000 was clearly considered unacceptable.  Individual licenses would need to be in the $30 - $50 range.

3.     Evaluation of VAL in a University Setting

3.1 Establishment of University VAL Objectives and Design Requirements

The objective of this component of the Phase I project was the specification of the educational objectives and the determination of the VAL design functionality requirements for the university prototype VAL. This was primarily addressed by testing the prototype with the assistance of Dr. James Moore of the Department of Earth and Atmospheric Science (EAS) at Saint Louis University (SLU).  Two courses were selected in the atmospheric science curriculum at SLU: Synoptic Meteorology II and Dynamic Meteorology II.  These were deemed the most logical courses in which to test the VAL, as they are taken by senior meteorology students during their last semester of college.  The senior level students would have the necessary background course work that would enable them to create testable hypotheses, run the experiments, interpret the results and provide feedback to the researchers.  In Synoptic Meteorology II, a large section of the course centers on topics in both mesoscale meteorology and numerical weather prediction, thus the VAL was an excellent tool to tie these concepts together.  In Dynamic Meteorology II students learn about energy transfer in the atmosphere, barotropic instability, and baroclinic instability, which are easily addressed within the VAL.  Thus, these courses provide many topics on which to base VAL experiments for open‑ended inquiry.

3.1.1 Identification of Objectives

There were many ways in which the VAL could be used in the University setting.  A choice had to be made among an array of possible concepts in the atmospheric sciences.  The syllabi of the Synoptic Meteorology II and Dynamic Meteorology II courses were evaluated to determine which topics would lend themselves naturally to use of the VAL. The topics that were considered were:

(1) Frontogenesis (development of a direct thermal circulation);

(2) Lake/sea breezes (development of a direct thermal circulation);

(3) Cyclogenesis and baroclinic instability;

(4) Geostrophic adjustment (forced direct and indirect circulations related to upper‑level jet streaks);

(5) Low‑level jet development related to changes in the planetary boundary layer;

(6) Lake‑effect precipitation; and

(7) Studying the effects of increased soil moisture or deforestation.

For this first phase of the project it was decided to focus on two fundamental concepts: (1) direct thermal circulations, and (2) the geostrophic adjustment process related to upper‑level jet streaks.  The reason these topics were selected is that they both demonstrate important and theoretically well understood concepts in weather forecasting and analysis.  It was important to start with concepts that are well-defined so the VAL could be evaluated for both its accuracy and its educational utility. The direct thermal circulation is the most fundamental and intuitive of atmospheric motions, with the motion being caused by the warmer air rising and cooler air sinking. Another basic circulation type is a dynamically forced coupled circulation associated with the geostrophic adjustment process, which results in either a direct (indirect) thermal circulation in the entrance (exit) region of an upper-level jet streak.  The indirect circulation forces the cooler air to rise and the warmer air to sink.  The indirect thermal circulation is very important to many meteorological situations, such as synoptic-scale cyclogenesis, but it is not as intuitive as the direct thermal circulation and can be a challenge to teach.

In these two experiments the educational objectives were to have students (1) understand what parameters play critical roles in developing the vertical circulation, (2) see the time scale of the circulations and the effect on the sensible weather, (3) visualize the depth, strength, and lateral extent of the vertical circulations and their change with time as the Coriolis effect takes hold, and (4) test the sensitivity of the circulations to small changes in key variables.  An important added benefit of this exercise was that the students would also become familiar with how to set up and run a numerical model, albeit some of the details of the model would remain "hidden."  The ability of the VAL to change one parameter and see its effect on the subsequent run of the model, as compared to the control experiment, is unique in meteorological education.

An important point is that the VAL experiments were to be integrated into a course, i.e., they were run only after lectures describing both the Mesoscale Atmospheric Simulation System (MASS) model and the physics behind the formation of a sea breeze were given in class following the presentation in Atkinson, 1981.  It is envisioned that this would be the case for most of the other modules that are created for the VAL.  Thus, the paradigm was to give a qualitative discussion of the problem, then present a quantitative discussion of the key parameters, followed by the VAL experiments that show the complex non‑linear interactions among those parameters. 

3.1.2  Selection of Experiments

The prototype VAL was early in its design, so only two experiments using the VAL were developed.  The sea breeze direct thermal circulation and forced direct and indirect thermal circulations in the vicinity of upper‑level jet (ULJ) streaks were chosen as the meteorological phenomena to be investigated with the prototype VAL.  The sea breeze experiments demonstrate how unequal heating generates a direct thermal circulation.  The geostrophic adjustment experiments demonstrate how divergence/convergence in the entrance and exit regions of an ULJ streak help create direct and indirect thermal circulations critical to the production of precipitation.  It was commented by several evaluators that it would seem more appropriate to evaluate a "basic" concept such as a cold front that is more widely taught and understood.  This misconception highlights the value of the VAL.  Many of the concepts in meteorology such as the cold front are based on necessarily oversimplified conceptual models of very complex phenomena that are really the product of many interactions at numerous scales.  The VAL can certainly be used to investigate experimentally and reveal the true nature of complex phenomena such as a front, however it was felt that it was essential to begin with a system that operates at one meteorological scale in order be able to understand how the VAL was working.  More complex topics could then be investigated.  

In order to achieve the Phase I goal of demonstrating feasibility, it was deemed best to first run a simulation experiment which was driven by local surface forcing instead of one driven by transient atmospheric features.  The main reason was the greater simplicity in generating initial conditions and the interpretation of the results by the students.  Thus, it was decided to implement the sea breeze experiment first because it is a very basic well‑defined and physically understood atmospheric circulation system, which exemplifies the concept of a direct thermal circulation.  The shallow, low‑level, mesoscale sea breeze circulation is resolvable on a 10-km grid and has been well simulated by the MASS model.  More importantly, there are easily definable parameters that the student can adjust to test the effect they have on the development of the sea breeze.  Another appealing aspect of the sea breeze experiment is that it is easy to document with sensible weather elements, such as temperature, dew point, wind direction and speed, clouds and precipitation.  It is also easy to depict with satellite imagery and links to a well‑established theoretical foundation found in the literature and textbooks.

The second experiment (designed but not implemented in class due to time limitations) was performed independently by the researchers.  It centered on understanding the geostrophic adjustment processes that take place in the vicinity of an ULJ streak.  The VAL student in this case can "control" parameters such as the magnitude of the jet streak, the isotach gradients in the entrance and exit regions, the curvature of the flow, low and mid‑level static stability, and latitude of the jet streak to see the effect each parameter has on the ageostrophic flow forming the exit and entrance regions of the jet streak.  All of these parameters play important roles in controlling the depth, strength, and lateral dimensions of the secondary ageostrophic circulations associated with the ULJ.  Students in both Synoptic Meteorology II and Dynamic Meteorology II learn a great deal concerning the theoretical foundation of these circulations and would greatly benefit  by performing experiments to investigate their characteristics.  Experiments dealing with this geostrophic adjustment process could easily be integrated into either course. 

The SLU professors who taught with the VAL found it most effective that the VAL experiments be conducted only after a detailed lecture and discussion on the theory of the physical processes involved. Thus, the VAL is used as a tool to reinforce and visualize concepts introduced in the lectures and during map discussions. The sea breeze lecture that was developed for integration with the VAL experiments can be seen at http://www.borg.com/~sandi/val.html.

3.2  Construction of University VAL Prototype

A limited but completely functional VAL was designed and implemented successfully at SLU.  The design and development of the university prototype was accomplished through a cooperative effort between SLU and MESO.  The design of the VAL prototype focused on the concept of the sea breeze experiment.  However the prototype VAL is robust enough to perform numerous experiments.

 3.2.1  Design Issues

It was decided that the Phase I prototype would be an HTML (HyperText Markup Language)‑based VAL that could be ported to any UNIX‑based machine with a web server.  The reason for this decision was both educational and technical.  Educationally, the decision was based on the success of such web-based programs as the numerical methods course of the University of British Columbia (Allen et al. 1997) and the "Kids as Global Scientist" course offered as part of the University of Michigan under the "One Sky Program" (Songer et al. 1998).  Technically the decision was made because a web-based approach requires a minimum amount of development to produce a product that could test the educational utility of the VAL.  The plan for the ultimate commercial version of the VAL is to create a self-contained program that is installed in one step and incorporates all the features of the prototype VAL.

The university VAL prototype was designed with an HTML-based web browser user interface that operated under an Apache web server. The Apache web server was selected because it is widely used, well documented, highly portable to many platforms and in the public domain. GEneralized Meteorological analysis PAcKage (GEMPAK, Koch et al. 1983) was selected as the analysis and visualization software package because MESO personnel have extensive experience with GEMPAK, it is available at SLU and widely used at the university level.  In addition it is extremely versatile in its visualization capabilities.

There are three steps in running the VAL: (1) setting up the experiment, (2) running the experiment, and (3) analyzing the results.  Each step has its specific design issues. The overriding issues were to allow the student to interact with the mesoscale model and control the experiment without having to worry about any technical details of running the model and be able to run the model without getting mired in the many pre‑processing options.   The two main issues of setting up the experiment were (1) to give the student a user-friendly interface, and (2) to include enough options to effectively control their experiment.   The issues of running the model were primarily technical in nature that centered on finding the fastest model configuration that gave acceptable results.   The main issue impacting the analysis of the data was how to visualize the model output. 

At the university level it seemed reasonable that the student should know enough about the controllable parameters to select realistic values.  However to ensure that realistic selection occurs, the VAL has built‑in values to tell when the user is way out of bounds.  For senior meteorology students, it also seemed reasonable to allow a dozen or so selectable parameters (Table 3-1) to initialize the model. 

 

Table 3-1.  List of selectable VAL initialization parameters for the sea breeze experiments.

 

Year

Terrain Type

Wind Curvature

Month

Land Use

Wind Direction at low-level

Day

Coastline

Wind Direction at Mid-Level

Start Time

Land-Water Distribution

Wind  Direction at High-Level

Latitude

Water Temp

Wind Speed at low-level

Longitude

Sounding Type

Wind Speed at Mid-Level

Experiment

Low-level inversion strength

Wind  Speed at High-Level

 

Another significant issue was to determine the best type of experiments to run for instructional purposes, real or idealized.  Both types of simulations are possible with the VAL and the Phase I investigators ran samples of each type in order to evaluate the instructional utility of each.  From the evaluation, it was decided to limit the in‑class experiments to the idealized cases for several reasons:  (1) idealized cases can be made physically and dynamically consistent; they also allow complete control of what factors will influence the experiment and permit the experimenter to isolate cause and effect.  Real cases always involve many interacting local and atmospheric features that move through the model domain, making it impossible to ever completely isolate cause and effect for any given factor; (2) real cases always have some inertia due to the antecedent conditions, which will cause unpredictable results;  (3) the real land surfaces are extremely complex which make it very difficult to isolate the impact of specific changes to the surface; and (4) the display of the results in a real case is very difficult to interpret because of its complex nature.

In order to create idealized experiments, MESO developed a scheme that allows the student to select from a set of parameters to create and control an idealized atmosphere and earth surface setting.  This is an additional capability to the existing method of ingesting real data also available to the VAL. The input parameter data is used by the VAL data preprocessor to construct a balanced model atmosphere and realistic but idealized earth surface. In the case of the sea breeze experiment the grid created was divided evenly as a water and land surface.  A web page that enabled the VAL-user to easily input initial conditions and make changes to the data set for "what if" analysis was created. The web interface was designed so that the interface code would test each user response to make sure that no selection is "out of bounds," e.g., a sea surface temperature of 100° C.

A significant issue facing the design of the VAL was to establish a MASS model configuration which would run the experiments fast enough to allow the results to be available to students and instructors for proper integration into the course.  It is acceptable at this level to prepare and run the experiment during one class period and examine the results during the next class period.  However, the desire was to be able to run the experiments and have the students analyze the results within a single lab period.  To meet this goal the experiments had to run in less than 1 hour.  The Department of Earth and Atmospheric Sciences recently acquired a moderately fast (300 MHz) SUN Ultra workstation, which was able to run an 18‑hour experiment in less than 40 minutes.  This facilitated the running of the model in a near real‑time environment so the students could look at the output the same day.  Each student had the results of their experiment stored in their directory under a unique file name so that they could examine the results at some later date, and also run new cases, which would stay separate from the previous experiments.

The final design component of the VAL was the capability of analyzing and examining the experiment results. It was decided to use the VAL browser interface to access pre-selected GEMPAK calculation and analysis routines.  MESO programmers in consultation with SLU personnel designed a web page that allowed easy perusal of the GEMPAK displays of the experimental results.

The result of this design was that the prototype VAL was a self‑contained web site in which the student could set up and run the experiment and then examine the model output with a web browser.  The only "external" component of this system is the GEMPAK software, which is available at most universities which are members of the Unidata consortium.

3.2.2 Programming Issues

Several programming issues were faced when the VAL was constructed and implemented.  The development of the Phase I prototype system was an HTML-based system done under an X‑Windows UNIX workstation environment.  The reasons for this were: (1) the MASS model was already configured to execute in this environment; (2) the components needed to construct the interfaces and visualize the data in a UNIX environment already existed; and (3) university meteorology departments generally utilize UNIX workstations for instructional and research purposes.  All of these factors permitted greater efficiency in the use of the Phase I resources and put the focus on the establishment of the feasibility and value of the concept and not on technical implementation issues.

The modeling component of the VAL (including data ingestors) is written in FORTRAN.  The structure of this code was largely preserved.  The only significant modifications were made to the data ingestor, by adding the synthetic initialization capability, and adding a VAL-user interface to the MASS.  In order to keep the Phase I development costs within the budget constraints, the display capabilities of the data exploration module was provided through GEMPAK.  The GEMPAK functionality was seamlessly integrated with the other components of the VAL so that the prototype system behaved as a coherent package and not as a series of independent software modules.  This will be the starting point for the development of the experimentation and instruction modules in Phase II. The system was designed so that it can be ported efficiently to the desired low‑cost consumer‑oriented platforms during Phase II.

3.2.3 VAL Prototype Description

A series of HTML pages that were used to activate appropriate Common  Gateway Interface (CGI) scripts were developed and implemented.  Users first select the type of experiment they would like to run.  For the Phase I in‑class use the ideal sea breeze case was the only allowable option.  After selecting the type of experiment, the student then sets the controllable meteorological parameters from a menu-based list of 18 parameters and configuration options such as the start time and date, the land water distribution, and shape of the coastline.  An example of the user interface for the configuration phase is shown in Figure 3.1.

Through the interface, the user is able to choose one of several idealized atmospheric profiles to form the initial state of the base atmospheric sounding and can then modify the base sounding in several ways.  The generic base soundings (McLatchey et al. 1972) are tropical, mid-latitude summer, mid-latitude winter, subarctic summer and subarctic winter.  The low-level temperature may be adjusted by an arbitrary amount.  The adjustment will apply at ground level and the temperature profile will be modified in the lowest 150 mb to blend into the base sounding.  The water vapor mixing ratio can be modified in such away as to preserve the relative moisture lapse rate of the original sounding.  A weak (2.5° C over one model vertical level) or strong (5° C) temperature inversion may be added to the base sounding at 900 mb.

There are several ways for the user to add cloud cover to the initial state.  For example, a single model layer of cloud cover may be added at either 2 km or 13 km above the ground.  The cloud layer may be thin (liquid water path of 10 g m‑2) or thick (200 g m‑2).  The relative humidity may be specified (overriding the relative humidity in the idealized base states) in three broad layers of the atmosphere (below 800 mb, between 800 and 450 mb, and above 450 mb).  Similarly, the wind direction and speed may be specified in these same three layers center of the layers and the winds are vertically interpolated between so that they vary smoothly in the vertical.

Once the configuration is complete, the student submits the entries and a quality check of the input is performed.  The student is then given the opportunity to either accept or reenter the input data.  Once the student is satisfied that everything is correct the execution phase of the experiment is initiated.  It typically took about 40 minutes for 18-hour simulation to be completed.

At the start of the execution phase, PREPSYN, a code module developed specifically for the VAL interprets the user input and configures the model initialization dataset to reflect the choices of the user.  The VAL graphical user interface restricts the choices of many of these options to a reasonable subset for practical purposes, although the underlying PREPSYN module can accept a much wider range of inputs.  Once PREPSYN has produced a model initialization file, the execution of the model simulation is initiated.  After the model's number-crunching is complete, the user can activate a web page where options are provided for displaying the MASS model output. The MASS model output is converted to GEMPAK-format as part of the post-processing so that the graphics of the forecast could be displayed.  In the prototype the students were able to choose from a pre-selected set of fields at 3-hour intervals through the use of the web interface.   The fields included many traditional meteorological displays such as mean sea level pressure, wind vectors, relative humidity as well as a few advanced diagnostic parameters such as frontogenesis and moisture advection.

 

Figure 3.1.  An excerpt of the user interface display for the configuration somponent of the prototype VAL

 

Once the HTML form requesting specific display options is selected and submitted, a CGI script is initiated that runs GEMPAK programs which create a GIF image and/or Postscript file for viewing or printing.  A variety of horizontal plan views of the model output were generated through the VAL interface.  Additional analysis was done using GEMPAK outside of the VAL.  Cross‑sectional views of a variety of parameters were created without the use of the VAL interface, and were found to be of great value.  Such options are clearly needed within the VAL.  The students' task was to evaluate the experimental results in an attempt to isolate differences between the control run and their own experiment.  This part of the experiment requires background preparation by the student since the students have to know what to look for and what parameters are best suited to diagnose the variations of the sea breeze circulation.

3.3 Evaluation of University VAL Performance

This section describes the experiments that were actually run by the students in the Synoptic Meteorology II and Dynamic Meteorology II classes at SLU and what we learned from this experience.

3.3.1 Description of the Experiments

In the Synoptic Meteorology II class a set of sea breeze experiments were run using the ideal or "synthetic" data set.  The following factors were investigated through simulation experiments: (1) ocean temperature, (2) shape of the coastline, (3) direction of the synoptic wind, and (4) strength of low level inversion.  The control simulation was initialized at 1200 UTC (0700 EST) on July 15th at latitude 28.0°N and longitude 81.5°W.  The terrain was flat and the coastline was straight with the grid being evenly divided with water to the east and land to the west.  The land use type was set to agricultural.  The mid-latitude summer sounding was used and the water temperature was set to 18.3°C.  There was a straight wind at 5 m/s from the east at all levels.  No inversion was imposed upon the base sounding.   Experiments were run by altering this base state in a variety of ways.  For example, a warm water case used a water temperature of 23°C while a cold water experiment used a water temperature of 13°C.  In the case of the "west wind" experiment a wind of 270° at 5 m/s was used in place of the 90° wind used in the control experiment.

There are several key factors that affect the strength and evolution of the sea breeze.  The factors considered the most significant are (1) the large-scale gradient wind which can overwhelm the mesoscale land/sea breeze circulation; (2) atmospheric stability which affects both the time of onset and intensity of the sea breeze; (3) moisture availability, which can help generate convective latent heating which can feedback to alter the sea breeze circulation; (4) topography and orography which can greatly influence the sea breeze evolution and the type of weather it produces; and (5)  the water temperature which will modify the land/water temperature contrast, and impact the thermal gradient thus influencing the sea breeze.  Time constraints did not permit investigation of all of these.  The experiments selected were designed to investigate the influence of the ocean temperature, the influence of the synoptic-scale wind, stability and coastline shape.

3.3.2 Experiment Results

There were two parts to evaluating the experimental results.  First, did the experiments reinforce the concept being taught and second, did the experiments provide any new insights to the sea breeze that could not be illustrated or determined with traditional teaching methods?  The answer to both questions are yes.

The basic structure and evolution of the direct thermal circulation was clearly depicted in each experiment.  A more classic picture emerged in the control case (Figure 3.2), but in each case the sea breeze circulation was evident. helped reinforce to the students how a direct thermal This circulation forms and evolves.

 

 

Figure 3.2.  A vertical cross-section of the ageostrophic component of the wind field and potential temperature (theta) at 9 hours into the control simulation of the sea breeze experiment.  This display was generated from the VAL interface by accessing the GEMPAK graphics package through a CGI script.

 

The sensitivity experiments clearly highlighted some aspects of the sea breeze formation and intensity which are not intuitively obvious.  First it was found that the strength and structure of the sea breeze is not influenced greatly by the exact temperature of the water so long as a minimal thermal contrast can develop between the land and the water of at least 5°C.  Greater thermal contrast has little influence.  This intuitively did not seem right to the students.  However, a rigorous reexamination of the theory revealed that this indeed should be the case. Theoretical studies of sea breeze circulations indicate that the exact temperature of the warmer (land) surface is critical to the structure and evolution of the sea breeze.  Even a degree or two can make significant differences.  However, the exact temperature or the colder (water) surface has only a minor impact on the circulation.  The reason for this is that it is the warmer air that is acted upon and lifted by the colder air, so the temperature (thermal stability) of the warmer air is very critical to how the direct thermal circulation evolves and what type of weather results in conjunction with it.  The colder air is not lifted, so its thermal stability has little impact.  This actually led to a reinforcement of the larger principle that applies to all lifting mechanisms, that the stability of the lifted air mass is much more important than the stability of the lifting air mass.  

The impact of wind direction, wind shear and thermal stability were also investigated.  Synoptic-scale wind direction was found to have the most significant impact on the sea breeze.  Even at a wind speed of 5 knots, the change in direction from east to west caused significant changes to the structure and evolution of the sea breeze.  In the case where the ocean is placed on the eastern half of the domain and land on the west, an easterly wind produced similar results to the control case. However, the west wind case produced a significantly deeper but narrower circulation that is more conducive to upright convection.

Ideal experiments examining geostrophic adjustment of the jet streak were also performed by the Phase I investigators. The hypothesis tested was the influence of curvature on the classic four‑cell vertical motion pattern traditionally taught and used as a forecasting guide.  The control experiment involved a straight jet streak that produced the classic vertical motion pattern.  However the experiment with a cyclonically-curved jet streak produced a very different pattern.   This demonstrated the possibility of gaining additional insight into the real nature of specific atmospheric phenomena.

Several real cases of the sea breeze and jet streak were also examined by the Phase I investigators using techniques that could be easily employed in the VAL.  The results were presented at the VAL workshop that included SLU students.  The selection of the real sea breezes cases was aided by information provided by meteorologists at the Kennedy Space Center's Applied Meteorology Unit who are currently involved in sea breeze modeling research.  Suggestions, data and analysis from yet unpublished cases were provided to MESO (Manaobianco and Nutter 1998).  Several things became apparent when examining the real cases:  (1) results of the real cases were much more complicated than the ideal cases;  (2) ideal cases were better to use to teach and reinforce basic meteorological concepts; and (3) after examining several ideal cases the real cases were easier to interpret.

3.3.3.  Technical Performance of the VAL

The VAL performed generally well at Saint Louis University.  The input interfaces allowed the students to set up and run the model quite easily.  This part of the VAL was judged by both students and instructors to be a complete success.  The only significant limitation within the VAL, which was mitigated to a large extent through analysis outside the VAL, was in evaluation of the model output and results.  It became clear that ageostrophic winds presented in cross‑sectional view are required to diagnose the sea breeze or land breeze circulation.  The system was somewhat slow in responding with a graphic once the parameters were chosen, and unfortunately only one person at a time could generate an image, otherwise the jobs got "crossed" and the user would see what the other person requested. Another limitation was the inability to generate soundings over the ocean or land to see how the model atmosphere was changing.  These limitations can be easily fixed in Phase II.   To provide a short-term solution, MESO analyzed the output using GEMPAK independent of the VAL and presented results that clarified many aspects of the experiments.  The analysis and display capability will need the most enhancements in Phase II. 

There were several advantages in building a web‑based system for the university VAL.  However, the web interface, although convenient for the user, required a significant amount of web server system configuration and administration.  Actions had to be taken to establish user accounts and directories.  Because the server used was Apache and the graphics package GEMPAK were required, no other web server or meteorological display package was supported within the VAL.  Users had to be on a system with these applications present for it to work.  This is not a major problem in the university setting, since most meteorology departments use UNIX-based systems and can obtain both Apache and GEMPAK at no cost. 

Currently PREPSYN is very capable of generating initial states with a wide range of surface characteristics and vertical atmospheric structure.  However there is a significant limitation when investigating phenomena such as fronts, baroclinic waves, and jet streaks that require initial states with meteorologically realistic horizontal variations. The major technical issue involved in the mesoscale modeling part of the VAL was that the synthetic initial conditions created by PREPSYN did not have time‑dependent lateral boundary conditions for the model to use.  For the simplest kind of idealized sea breeze cases used in the Phase I work, the lack of boundary conditions seemed to have little effect over the center of the grid.  However, artificial gradients developed near the lateral boundaries of the grid that would be eliminated with time-dependent lateral boundary conditions.  The lack of such time-dependent boundary conditions did not impact the educational utility of the VAL for the sea breeze case, but it would be a serious hindrance in attempting more complex simulations

3.3.4.  Meteorological Performance of the VAL

Meteorologically the VAL performed very well producing very realistic results that allowed the students to view aspects of the sea breeze development not easily understood.  The students were able to see how the temperature and pressure gradient strengthened while the sea breeze developed.  The vertical circulation was also clearly evident when viewing cross‑sections of ageostrophic winds on potential temperature surfaces.  As indicated in the technical evaluation, there were some problems with the boundary conditions; however they did not influence the meteorological meaningfulness of the results.  Clearly, the analysis of certain parameters (such as potential temperature) and certain perspectives (such as cross sections) demonstrated that the VAL could provide insight into the real nature of atmospheric phenomena that would otherwise be much more difficult to grasp.  This was made evident during the VAL workshop when many experienced teachers and professors came away saying that they learned unique aspects of the sea breeze that they had not understood before.  For example, the relative importance of the land‑water temperature contrast as compared to the synoptic-scale wind direction was not widely understood until running the VAL experiments.

3.4.  Conclusions on University Evaluation

Overall, the prototype performance of the VAL met the educational objectives.  The VAL interfaces worked well for the set up and execution.  However, there is a need to expand the number of parameters that are possible in the analyzed output.  Improving the appearance of the output is absolutely critical.  Today's students are used to point-and-click interfaces that work quickly, and they demand instant gratification for their efforts.  Features need to be added to the visualization capability of the VAL to allow more thorough analysis of the output.  Users need to have the capability to look for and examine those features characteristic of the sea breeze or whatever phenomena being examined as described in the theoretical and physical discussion from the lectures.  This became evident in the student's responses through the questionnaires, which are discussed later in section 4.2.2.  Another critical lesson was learned during the evaluation: the instructor needs to spend time working with the VAL in order to "spin‑up" before using it in the classroom setting.  With this experience they can help the students choose the most useful parameters and time periods to see the development of the phenomena being investigated, in this case the sea breeze direct thermal circulation.

Many things were learned from the university experiment.  This included: (1) users should have much more flexibility in the types of images they can display; (2) improved on‑line help needs to be integrated into the VAL;  (3) feedback informing the user as to how much time remains before the model run is completed is needed;  (4) although the web server interface was quick to develop and showed the potential utility of the VAL, it is probably much more sound to build an executable program (that can run on any UNIX operating system) that is adaptable to any operating system using an X‑Windows/Motif user interface;  (5) work needs to be done to create initial states with meteorologically‑realistic horizontal variations.  For instance, baroclinic waves, fronts, and jet streaks should be built into the initial state;  (6) there is a need to develop a method of generating synthetic time‑dependent lateral boundary conditions to go with the synthetic initial conditions.  Writing a new program that would be a companion to PREPSYN could accomplish this.  An alternate method would be to use PREPSYN to generate several initial states, then write a separate program which would extract the boundary values from a MASS initial state and write a MASS model‑ready boundary condition file.  Such a program would likely have applications beyond the VAL project.

We envisioned the VAL as fulfilling an important need in the atmospheric sciences, namely offering a "laboratory" within which students could test and explore theoretical concepts.  The laboratory in this case is a primitive equation, mesoscale numerical model capable of simulating physical processes in the atmosphere.  Adjusting critical parameters in this numerical model brings a capability to the atmospheric sciences that currently only exists in physics and chemistry laboratories, where students can make a hypothesis and then test it in a controlled environment.  In this sense, the VAL can be used as part of a "capstone" course in which students integrate their knowledge from other meteorology courses to study atmospheric processes associated with such diverse array of weather systems such as a sea breeze circulation or an extra‑tropical cyclone.

4.  Evaluation of VAL Educational Objectives

The development of the overall educational objectives, requirements and functionality of the commercial VAL was an extension of the process that established the university objectives. A number of sources were contacted to compile a list of potential educational objectives and associated VAL functional requirements for each educational level.  One source was the EAS faculty at SLU and the State University of New York (SUNY) Oswego, especially those who have experience with the instruction of introductory level general meteorology courses.  Faculty at other universities, four-year colleges, community colleges, high schools, middle schools, training centers and other private meteorological firms were also contacted and interviewed.  Twenty-three teachers involved in the American Meteorological Society (AMS) sponsored Atmospheric Education Resource Agent (AERA) (Smith et al. 1996) and DataStreme (Bastiaans et al. 1998) programs as well as 17 other science teachers were contacted and interviewed and provided questionnaires. We also consulted heavily with YCI on what is appropriate for the middle school from their experience of developing meteorological educational software which is currently used by over 830,000 classroom teachers.  The researchers also reviewed and participated in several software and web-based weather related programs such as Kids as Global Scientists (Songer et al. 1998) and Weatherschoool ®.  The fact we reached such a broad based group of educators during our research ensured that we have a good understanding of what the commercial VAL educational objectives appropriate for each educational level should be.  On May 21-22 1998, near the end of Phase I, a two-day hands-on evaluation of the prototype by representatives of each educational level took place at Saint Louis University. The following organizations participated: SLU, SUNY, Saint Louis Area Schools Grades 5-12, Omaha Nebraska Schools grades 5-12, Rome, New York Schools grades 7-12, Air Force Weather Training and the National Weather Service Training Center. 

4.1 Middle and High School Evaluation

The research into the currently published national, state and local standards was done by reviewing such documents as the National Science Education Standards (National Research Council 1996), National Science Teachers Pathways to Science Standards (Rakow 1998). Benchmarks for Science Literacy (1993), and New York and Missouri state science-teaching standards (1998).  The clear trend from reviewing these documents is that science education is shifting from a fact-oriented approach to one of viewing science learning as an active inquiry-based approach focusing on understanding concepts and processes. The 100% consensus in the interviews of all the grade 5-12 science teachers was that the primary reason for teaching meteorology and atmospheric science for this grade level is to teach the concepts and principles of the atmosphere using an inquiry-based systematic scientific approach.  The consensus was that this group should be further divided into cognitive skill levels of grade 5-8, and 9-12.  The Benchmarks for Science Literacy (1993) publication indicates that the topics appropriate to the VAL should begin to be introduced at grade 5, with increasing level of understanding expected in the processes of the earth and atmospheric sciences through g