       KEY POINTS IN THE REVISION HISTORY OF SIBERIA V8
     ====================================================

V8.01
-----

- Introduction of soils model. Simple exponential insitu soil production 
(ala Heimsmath) with a spatial soil wetness feedback on soil production 
rate soils model. Preliminary testing.

V8.10
-----

- Last version fully compatible with Fortran77. Future versions to incorporate
Fortran 95 features.

V8.23
-----

- Release version of model with soils module allowing output of various aspects 
of the soil model for Patricia Saco project. 
- Some introduction of dynamic memory

- FINALISED (9/03)

V8.24
-----

- Bug fixes on soils module. 
- Further introduction of dynamic memory.
- Existing LAYERS module removed and work commenced on a new more general LAYERS
and sediment tracking module.

- FINALISED (23/11/03)

V8.25
-----

- First reliable release with full dynamic memory. In Windows 2000/XP  
the limitation on the maximum grid size of 1200x1200 is removed. Performance 
is limited by installed memory with speed dropping dramatically once installed 
memory is exceeded (2 Gigabytes on Windows approx equivalent to 1200x1200).
- No longer runs in parallel under openMP V1.0 (due to dynamic memory features
which require use of MODULES for global data rather than COMMON).
Parallel version under openMP V2.0 planned for future version.
- Work continuing on the new LAYERS module.
- bug fix for Dinfinity

- RELEASE VERSION: dynamic memory (8/4/04)

V8.26
-----

- work continuing on the LAYERS module and sediment tracking
- improved error checking on RST2 file input
- modeMC=1 completely removed ... request for ModeMC=1 sets ModeMC=0
- adding binary file output as an alternative to .RSU file ... for
  use by IDL and other visualisation suites.
- preliminary testing of Dinfinity fix ... still doesn't work reliably

- FINALISED (25/10/04): Sent to LANL

V8.27
-----

- work continuing on the LAYERS module and sediment tracking.
- files for user defined models (e.g. test.erode) are now called model files
(e.g. test.model) and are now completely portable between the various user 
defined models (e.g. erosion, runoff, tectonics).
- improved modelling of detachment-limitation in ModeErode=3. 
ModeSolver=8 triggers the new capabilities. For ModeErode=0 (or where regions
all have the same erodibility) the results will be the same (+/- round off error)
as for ModeSolver=4. Performance penalty of ModeSolver=8 relative to ModeSolver=4
about 20%.

- FINALISED (25/11/04): Sent to LANL

