Mike Grundy Leeper Lecture 2019
A National Soil Outl tlook GW Leeper Mem emori rial Lec Lecture - - PowerPoint PPT Presentation
A National Soil Outl tlook GW Leeper Mem emori rial Lec Lecture - - PowerPoint PPT Presentation
A National Soil Outl tlook GW Leeper Mem emori rial Lec Lecture re Mike Grundy 22 November 2019 Mike Grundy Leeper Lecture 2019 Mike Grundy Leeper Lecture 2019 Sustainable prosperity is possible, but not
Mike Grundy Leeper Lecture 2019
Mike Grundy Leeper Lecture 2019
Sustainable prosperity is possible, but not predestined. Australia is free to choose.
A business-led forum to shape Australia’s future
Mike Grundy Leeper Lecture 2019
Mike Grundy Leeper Lecture 2019
Mike Grundy Leeper Lecture 2019
Mike Grundy Leeper Lecture 2019
Mike Grundy Leeper Lecture 2019
Mike Grundy Leeper Lecture 2019
Mike Grundy Leeper Lecture 2019
Mike Grundy Leeper Lecture 2019
Mike Grundy Leeper Lecture 2019
$ returns Desired Outputs (e.g. Yield)
- r quality)
Inputs (nutrients, water, labour, agro-chemicals, energy etc.)
Biological Optimum
Risk- adjusted Optimum
Economic Optimum
Mike Grundy Leeper Lecture 2019
The AgData Challenge
Dan Gladish Dan Pagendam Ross Searle Chris Sharman Ashley Sommer Matt Stenson Cameron Taylor Peter Taylor Jamie Vleeshouwer
The Team
Return Per Farm($ per Ha) $ Machine Usage (Ha)
Ha
Number of Herbicide Applications per Paddock per Year
Total Granular Fertiliser App Rates Per Aggregation
Return Per Aggregation ($ per Ha) $
Mike Grundy Leeper Lecture 2019
Amount of Atrazine Applied (Ha) Proportion of Time Employee Spends on a Machine
From the management database we can also extract information about human resource management
Ha Sprayed
Mike Grundy Leeper Lecture 2019
Mark Branson - SA
Fertiliser Replacement Maps
https://view.knowledgevision.com/presentation/6bfd29c449bb4c82b564edf4204b738a
Mike Grundy Leeper Lecture 2019
Return on Nitrogen Aggregating Data
- Benchmarking – district, region, state, soil type
- Improved soil attribute maps
- Variety trials
- Improved research targeting
Mike Grundy Leeper Lecture 2019
EM Shallow EM Deep Radiometrics Th Radiometrics Total Slopes Ruggedness Flatness Elevation
Using statistical data mining techniques we can combine the topographic indices with the geophysics collected by LG and the soil test data, to generate soil property maps. Topographic Indices LG Geophysics Soil Test Data
Mike Grundy Leeper Lecture 2019
Organic Carbon Electrical Conductivity pH PBI
Soil property maps generated from digital soil modelling (DSM)
As we collect more soil test data we can potentially use this approach to make comparisons of changes
- ver time.
Mike Grundy Leeper Lecture 2019
Actual Yields
(machine data)
Statistically Predicted Yields
Cond Model Variable 97 84 rain 72 81 elevation 30 50 N 22 71 DualEM.Shallow 21 58 Radiometrics.Potassium 15 84 Radiometrics.Total.Count 13 76 Radiometrics.Thorium 10 68 DualEM.Deep 7 18 gradient 6 55 TerrainSurfaceTexture 3 56 Radiometrics.Uranium
So using the models from the previous slide we can generate maps of predicted yields. The models can also potentially give us some insights into what is driving yield
- utcomes.
The table below shows us the relative importance of the various input predictors in the overall predictions of yield – no surprises in the top predictors here!
Modelled Vs Predicted Yields
- 20. Data Analysis – Yield Modelling
Mike Grundy Leeper Lecture 2019
Modelled Canola Yield 400mm Rain Modelled Canola Yield 450mm Rain Difference between Modelled Yields 450 - 400mm Rain
Canola Yield Vs Rainfall (Statistical Modelling)
Evaluating options . . .
Yield (T/Ha) Scenario Yield Differences (T/Ha)
Mike Grundy Leeper Lecture 2019
Mike Grundy Leeper Lecture 2019
R2 0.91 LCCC 0.95 RMSE 2.61 ME 0.158 Proportion 0.99 Observed Average 35.69 Modelled Average 35.85
Modelled Vs Observed at Bolac
Mud Maps
Global Model Fit Stats
Mike Grundy Leeper Lecture 2019
Mike Grundy Leeper Lecture 2019
Mike Grundy Leeper Lecture 2019
Homi Kharas, Brookings Institute, 2017
Mike Grundy Leeper Lecture 2019