• Updated the software to support newer versions of Windows.

Release Notes - SCL - Version 2.2

This release replaces the previous version 2.1b which expires on 2007-03-01. This is purely a maintenance release and there is no modification to the application, apart from a minor update to the literature references in the manual.

Release Notes - SCL - Version 2.1b

This release replaces the previous version 2.0b which expires 1/03/2006. There are two fixes but mostly this is a maintenance release with minor improvements. The product is still considered "beta" because there were some changes, albeit minor, to the existing models. However it is likely to be the last beta version if no further model is added.

Fixes

• Modifications of the daily climate model to prevent the occurrence of unrealistically high temperatures.
• With version 2.0b the computer memory may be reclaimed even when closing the form of a given model. This may have led to an increasing memory footprint and a need to restart the application to free up memory. We apologise for any inconvenience caused.

Changes

• The daily rainfall scaling is modified to remove the bias in the generated mean annual rainfall
• The climate models can generate mean maximum temperatures that are negative. While this is of little practical relevance for users in Australia, this is logical to allow for data in other parts of the world.
• In the AR(1) annual rainfall model, the normal random deviates are truncated at 3.75 to avoid generating very large values.
• All models now reproduce by default the same replicate time series, all other things being equal. That is, the random processes are seeded with a deterministic sequence of  seeds instead of pseudo-random seeds derived from the current time. An option is available to revert back to a fully randomised processed if required.
• The first year of the monthly rainfall and climate models used to have the same pattern for all replicates, due to the lack of antecedent values (december month) to select the the disaggregation pattern in the method of fragments. This has been improved by rolling over the last december values of the previously generated replicates, to prevent this artefact.

Improvement

• The default temporal characteristics of the replicates are now based on the temporal characteristics of the observed time series, and the user can customise the starting year of the replicates. The previous versions of SCL were starting at "year one" to prevent confusion the replicates with real data, but this was an issue for some users and required them to post-process the replicates to shift them in time.
• Most models (except DRIP) now refuse to load time series which have any missing values. The relatively complex statical analysis of input time series may not be done reliably on data with missing entries. The previous version was giving a warning message, but letting the user continue otherwise.

Changes:

• Addition of multi-site daily rainfall model. Daily rainfall can now be simulated based on multiple site concurrent rainfall data using the multisite two part model.
• Addition of single site sub-daily rainfall model. The DRIP subdaily rainfall model, developed jointly by the University of Adelaide and University of Newcastle, has been added to SCL. This model simulates 6-min rainfall intensities based on input parameter files. As the model calibration module has not been included in this release of SCL, parameter files have been provided for a number of pluviograph sites Australia wide. These parameter files along with corresponding pluviograph data (provided by the Australian Bureau of Meteorology) can be found under additional downloads with the SCL software download area. Sites chosen were based on relatively long length of digitised pluviograph data available at those sites, along with there being sufficient continuous daily records for reasonable verification of model performance.
• Improved interface and user options.
• Expiry date of this beta version set to 01/03/2006

Bug fixes:

• A coding error affecting the skew of daily temperature generated by the daily climate model has been corrected. This error would typically underestimate the skew in simulated temperature series relative to the observed record. Note this error does not have implications for generated rainfall or evaporation sequences.

Changes:

• Addition of 'View Distributions' option - under 'Analyse Results'. Provides a graphical display comparing the historical vs. generated empirical cumulative distributions at various aggregation levels. For the Daily rainfall model the annual maxima curve is also calculated - allowing validation of model performance using Annual maxima Depth-Frequency-Duration curves.
• Addition of user manual under 'Help'.
• Expiry date of this beta version set to 01/09/2005.

Bug fixes:

• Sequential running of 1. Load data, 2. Estimate parameters and 3. Generate replicates is now enforced. Previously selecting Estimate or Generate before data was loaded caused an unhandled error.
• Minor change to the random generation of the intial time-step of the AR(1) with parameter uncertainty (Annual rainfall) model.
• Minor changes to statistics calculations. Occassionally a 'Nan' (Not a number) will appear for generated or historical statistics. Usually this is due to that event not occurring in a particular time-series. Attempts have been made for the replicates statistics mean to ignore such a result in individual replicated series.

Changes:

• MFM uses AR1 with parameter uncertainty to generate annual values internally, instead of a simple AR1
• The fit of AR1 with parameter uncertainty uses the Shuffle Complex Evolution (SCE) algorithm, completed with a direct search method (Rosenbrock) for the final convergence.
• Extended allowed number of attempts to generate a given item for a year in AR1 with parameter uncertainty. There may still be some cases where data is produced with a NAN if this model is fitted using a few years of data only
• Expiry date of this beta version set to 01/03/2005
• The software includes an update notifier. Users connected to the internet will be notified at start up of their currently installed SCL of any newer download available
• Following a Toolkit Steering Group decision, the comma separated values file extension is now '.cdt'. The '.csv' extension was too generic and leading some users into thinking that any such file produced by Excel would be readable.

First public release of SCL.

Product Manager

Scientific development and testing

Code development

CRC Researchers

Other Researchers

Project Reviewers

Subdaily rainfall model development