DairyNZ and Beef + Lamb New Zealand have backed a project that’s resulted in the tool developed by Hamilton firm Rezare Systems. Farm modelling and advisory company Farmax has also been involved in its development and testing.
The new tool gives farmers a prediction of pasture growth rate based on a raft of information drawn from weather stations and soil maps. Excitingly the computer models will also be able to learn from historical growth rate data and data put in on an ongoing basis. That will allow it to alter some of its assumptions and algorithms so that it is effectively learning over time, making it even more accurate.
Rezare managing director Andrew Cooke said it had been a bit of a convoluted road to get to the final farmer usable tool with the journey beginning back with an equally ambitious project, pastures from space. That project had aimed to find a way to calculate pasture covers using pictures taken by satellites passing over NZ. But our frequently cloudy skies meant not every pass captured every hectare so a model was developed to fill in the gaps.
Known as PGSUS the model involves using data from NIWA’s virtual climate station network. Rainfall and temperature data is collected from the climate stations along with a number of other weather-related indices and climate information for the gaps is predicted through a set of algorithms.
Accurate detail
That allows the model to provide climate information right down to a 5km square grid pattern across the country and that’s accurate enough to get right down to farm level.
But it’s not just recording climate information as it happens. Weather mapping and satellite information helps to formulate a forecast that the computer model can use to provide more detailed seven to 10 day weather forecasts than we’re used to at a regional level.
Cooke said the weather forecasts can be scaled down to about a 10km square grid. But to get enough information to predict pasture growth at a farm scale and even at a regional level the model needs more information.
Soils data is essential so the model integrates soils information provided by Landcare Research. The soil mapping is accurate enough to give a reasonable idea of the soil types at a farm level and provide the growth rate prediction model with essential information such as water-holding capacity and soil drainage.
Soil fertility information is also added, along with the predictions the climate information provides on factors such as potential evapotranspiration rates. When all of that’s combined with the model’s ability to learn and adjust its calculations the tool can quite accurately predict how much a pasture will grow in a given period.
Successful tests
To test it the model’s been run in eight areas across the country and compared with real growth rate data measured on farms in those eight areas. Pleasingly for the developers the forecasts and actuals have been very close.
For beef farmers, Beef + Lamb NZ has made the regional information free on its website but to get farm-specific information farmers register with Farmax and pay $9.95 a month. For that they get short forecasts, current season and long term pasture growth rate information.
If they’re using Farmax for monitoring and modelling on their farm already their data is automatically included in the modelling information for the pasture growth rate forecast. That allows the model to learn about the site’s specific fertility and productivity potential so that growth rate forecasts can be more accurate.
DairyNZ is also offering the regional information free but has left the delivery of the farm-specific information to a range of private-sector providers such as Farmax.
The regional information will be based on about seven focus farms across the country which will also be part of an ongoing validation project.
DairyNZ developer Steve Lee said the pasture growth forecaster had been compared with growth rate information gained by using a platemeter and by doing back calculations using cow production information such as liveweight and milk production data. The results had stacked up very favourably for the pasture growth forecaster with a strong correlation with the other methods.
As with all models the more information the model has the more accurate its outputs will be, so don’t throw away the platemeter just yet. What the forecaster will do is allow farmers to perhaps cut back on walking the farm as often as every week.
DairyNZ plans to launch the pasture growth forecaster regional data this month on its website www.dairynz.co.nz
