Release of APSIM SWIM3 model

[ApsimBot]

Legacy Bug ID: 1246 Author: Neil Neil Huth Date: 2011-02-09 14:32:16 +0000 UTC

STEP 1: PROPOSAL

Short description of the concept or proposed improvements: 

Swim version 2 has been in APSIM for many years, however various elements of it's complex parameterisation have precluded it from incorporation into the GUI.  Swim version 3 has been developed from V2. The approach taken in V3 is such as to facilitate a range of methods for specifying soil hydraulic properties.  The first approach implemented enables SWIM3 to run using properties as provided in the APSoil database.  Further developments should enable hydraulic properties to be specified from the ASRIS database or the user's own field measurements.  At present, only the first option is available

Rationale and Justification: 
A range of APSIM users are currently using, or exploring the use of SWIMv3.  This includes work in Australia, Uzbekistan, Finland, Bangladesh and the Philippines. Incorporation of SWIM3 into the release will enable these users to use a standard APSIM release for all of their work in the future.

Details of the proposed improvements: 
The approaches used for deriving soil hydraulic properties from SoilWat/APSoil parameter sets has been tested on a range of soil types (clays, sands, loams) on datasets from Australia and overseas.  Swim3 has the SCS Runoff calculations now included, thus allowing this familiar method for modelling runoff losses.

The implementation and user interface also provides access to the convection-dispersion equation for solute movement.  These have been tested using leaching studies under horticultural crops in the lockyer valley.  I have also included functionality for specifying sub-surface drains and ground water system dynamics in the GUI.  These elements of the model have been tested against results from other ground water models.

Implications for users: 

SWIMv3 can basically be used in place of soilwat in simulations with very minimal changes by the users.  Further developments, in conjunction with other interested parties, will enable a greater variety of soil property specification methods.  The first step will enable the users currently using SWIM3 in it's prerelease form to use a standard APSIM release.

These changes will have NO impact on modellers using soilwat.

Reviewer: 
Val Snow

Review Panel Contact: 
Val Snow

Supporting documentation (optional):

[ApsimBot]

Author: murphy Date: 2011-03-25 13:50:00 +0000 UTC

Email sent to Val Snow (Cc to Neil Huth) 25/03/11:

Dear Val,

The APSIM Initiative Reference Panel (RP) has received a proposal to modify APSIM as follows:

APSIM Task ID 1246 - Release of APSIM SWIM3 model http://www.apsim.info/BugTracker/sv-major_tasks.aspx?id=1246

This task has been implemented by Neil Huth and has been accepted by the RP for review (see Part D of the attached Reviewers Report - Process for Incorporating Improvements into APSIM ).

Given your field of scientific expertise, the Reference Panel believes that you would be an appropriate technical reviewer of this modification. 

If you can confirm that you are able to undertake this review within the time frame indicated in the attached Reviewers Report it would be greatly appreciated. 

Regards

Chris Murphy 
Principal Project Officer (APSIM)

[ApsimBot]

Author: sno036 Date: 2011-06-01 13:43:00 +0000 UTC

SWIM Review.pdf Size: 71909

[ApsimBot]

Author: sno036 Date: 2011-06-01 13:44:00 +0000 UTC

Hi Chris,

I have now had the opportunity to review the SWIM3 implementation and to test drive it with several NZ soils as well as see the results from the testing on the Australian soils that Neil had already done.

The module worked well and I was impressed with the ease of use.  I recommend that it be included in the next release.  More information on the testing etc is below.

 

Cheers, Val

 

 

The new SWIM module is very simple to use - simply remove the SoilWat component in the Soil node and drop in SWIM and a Thickness component.  This is substantially easier to use than the con-par version and also the temporary version that we have been using in AgResearch.  There were no numerical instabilities in the simulations that I ran.

I have tested the SWIM module against SoilWat using a range of soils and climates.  Some examples are shown below with using two contrasting soils in a Canterbury climate growing a perennial pasture either irrigated or dryland.  Key results are:

·         Total harvested pasture (not shown) is nearly unaffected by the choice of water module

·         There are minor and unimportant differences in standing drymatter in the dry simulations

·         Differences in drainage are very small

·         Differences in total water evaporated are small but in dryland simulations SWIM predicts a more evaporation than does SoilWat and there is a compensating increase in transpiration in the SWIM simulations (see below).  This is probably because the evaporation routine in SoilWat proceeds with the square root of time regardless of whether there are other competing demands for water removal from the surface layers where as SWIM negotiates between the demands.

·         In the NZ soils there was a pattern with SWIM predicting higher water storage as the soil dried but this effect was not seen in the example Australian soils.

From my perspective the tests were successful.  Although there were some differences between the two modules they were small and probably explained by differing assumptions (e.g. surface evaporation).  

 

The plots are attached in the PDF below - couldn't figure out how to do that there.

[ApsimBot]

Author: zur003 Date: 2011-06-01 14:12:00 +0000 UTC

I don't think SWIM3 can be used in AusFarm simulations in its current state, as it does not appear to register its drivers and readable properties with the protocol. This could be easily fixed. There may be other issues in the AusFarm context, but I haven't tested it yet.