Assessing water quality from farms how much detail is required to - PowerPoint PPT Presentation

Assessing water quality from farms – how much detail is required to model? �������������������������������� ���������������� ������������������������������������������� ������� � !�����"������#$�

Context and Overview Assess impact of management on water quality from range of land uses at the paddock scale (sediment, nutrients, pesticides) – Capture key datasets and insights from WQ studies (n= 112, 12 with detail) – Add value: best bet model parameters, and – “Publish” on a web site ( www.Howleaky.net )

������������������� ����������������� �������� ����������� ������������� ���������� ���������������� ������ ������ �������� ����������� !������������������"�����������������������������

A common question in synthesis and modelling of experimental studies: How much effort/accuracy is needed in describing a system (soils, vegetation, agronomy) to adequately simulate the real world?

Levels of experimental description vary Of 112 studies, 12 had detailed data

How much detail is needed to describe vegetation? Vegetation/crop Grazing at Mt Mort Averaged green cover Management practice Exclosed treatment Actual green cover Cover (%) 80 60 40 20

Are generic models able to predict runoff and soil loss? Runoff Semi-optimised model Optimised model 3 soils 3 veg series 1 soils 3 veg series Generic model 1 soils 3 veg annual Mt Mort Pasture study – Mark Silburn

Are generic models able to predict runoff and soil loss? Sediment loss Semi-optimised model Optimised model 1 soils 3 veg series 3 soils 3 veg series Generic model Mt Mort Pasture study – 1 soils simple annual veg Mark Silburn

For the statistically minded! Bare runoff } Three levels of vegetation description Bare soil loss Grazed runoff Exclosed runoff Grazed, exclosed soil loss Mt Mort Pasture study – Mark Silburn

System descriptions predict runoff and soil loss Sediment loss Runoff Level 4 data for Goomboorian Pineapple study per Cyril Ciesolka

In summary We can live with less detail in system description and still use conservative (mass balance) models with reasonable confidence This opens up access to a wider set of experimental datasets Empirical evidence remains king, and with pragmatic application of models we can stretch our data sets a long way