MetApps
Discussion
Progress to Date
(illustrated with demos)
MetApps prototypes:
- Image Viewer (IV)
- Interactive Sounding Application
(ISA)
- Gridded Data Viewer (GDV)
- Earlier Surface Obs Viewer
components and lessons are embedded in the above Viewers.
Advances in underlying fundamentals:
- The above prototypes deal with
important, difficult problems, including:
- image navigation (IV)
- multiple projections
(all)
- thermodynamic coordinate
transforms (ISA, GDV)
- interactive selection
& manipulation of data points (ISA, GDV)
- direct access to remote
data servers (IV, ISA, soon GDV)
- interactive selection
of input data (all)
- interactive cross-sectioning
(GDV)
- color-table manipulation
(IV, GDV)
- direct value readout
(ISA, GDV, soon IV)
- 3D rendering (all)
- highly connected component
networks (ISA, GDV)
- zoom & pan (all)
- animation (IV, GDV)
- Prototypes embody key discipline-
& data-independent abstractions.
Communication
structure:
·
Requirements
·
Approaches
o
email
o
UMADA
§
Centralized discussion area
§
Historical record of transactions
§
Combines documentation and use cases
Collaborations:
·
Australian Bureau of Meteorology (wind components, ADDE data access, data abstractions)
·
SSEC (VisAD, ADDE access, data abstractions)
·
NCAR - ATD/RAP (radar displays), CGD (GDV)
Lessons Learned
Suitability of the chosen infrastructure
- Java is adequate & getting
better.
- Java (esp Java3D) is not yet
as portable as hoped.
- VisAD is viable & has powerful
data & display models.
- VisAD is difficult to learn
& needs improved documentation.
- Java Beans are useful largely
for internal development at this stage.
Effectiveness of the process
- User-centered design is suitable
& workable.
- Task force input is sporadic
& requires seeding.
- We understand better the kinds
of communication required.
- Extensive user involvement
is unrealistic until MetApps are really useful (e.g., satisfies a use case).
- The "Planning Game"
method (Extreme Programming) can be effective.
Technical issues
- Development is resource intensive
& many of the tasks are hard.
- Collaboration takes effort
and resources but can have benefits
- Other Unidata matters (LDM/IDD,
e.g.) often distract from MetApps.
- We have not agreed on a data
model, & the problem is very hard.
Future Directions
Near-term objectives
- DODS and enhanced ADDE access
for MetApps, VisAD
- Radar data viewer (in collaboration
with ATD/RAP/TF and ?)
- Initial design of integrated
application
- Framework for derived quantities
(specification, selection)
- Explore mechanisms for broad
community input on prototype and component development
Long-term objectives
- One or two integrated applications
with broad functionality similar to, but not replacing, GEMPAK & McIDAS
(Data-type extensibility more important than feature matching)
- Component architecture, as
practical
- A few applets for users to
embed in education materials
- Broad data-access & data-discovery
landscape, linked to DLESE
- Architecture & framework
for user authoring & assembly
- Data & metadata models
that support the above
