Collecting georeferenced data in farm surveys Philip Kokic

Collecting georeferenced data in farm surveys Philip Kokic, Kenton Lawson, Alistair Davidson and Lisa Elliston

Overview < < < Objectives ABARE farm surveys Georeferenced paddock data Data modelling Conclusions

Objectives Improve responsiveness < Improve timeliness < Improve policy relevance < 8 More appropriate analysis 8 More detailed estimation 8 Better modelling of data

Coverage § § Survey ~ 2000 farms annually Broadacre and dairy industries only Stratified balanced random sample Estimates produced at ABARE region level

Survey regions

Collection of Georeferenced paddock data

Study region

Data modelling

Data modelling using spatial covariates < Intensity of agricultural operations (AAGIS) 8 Arable hectares equivalent /ha operated < Pasture productivity index (AGO) 8 Biophysical: incorporates climate and soil type Vegetation density (AGO) < Land capability measure (NSW Dept Ag) < Distance to nearest town (ABS) < Stream frontage (Geoscience Australia) <

Land value reg. n=232, R 2=80% Dependent variable: log (land value per hectare) Log intensity Log PPI Veg. density (%) Log land capability index Log travel costs Stream buffer prop. Estimate 0. 42 1. 16 -0. 02 p-value (%) < 0. 01 -0. 24 < 0. 01 -0. 45 4. 46 < 0. 01

Probability of exceeding median wheat yields in 2003 Emerald # Roma # Dalby # Goondi#wi ndi Courtesy of QDPI

Remotely sensed crop classification 2003 season 2004 season Courtesy of QDPI

Benefits of geo-spatial data < < < Increase responsiveness Biophysical modelling of crop and pasture data Reduced response burden Continuous in season crop estimates Improved accuracy of Small Area Estimation Econometric modelling