Evaluating methods of using data from boom-mounted sensors to adjust PGR inputs and nitrogen application timing

Abstract

The aim of the project was to interpret and evaluate crop canopy information for winter wheat crops collected using boom-mounted sensors that could then be used in the planning of nitrogen fertiliser timing and plant growth regulator application strategies. These strategies would aim to minimise the risk of lodging whilst also maintaining crop yield and minimising plant growth regulator (PGR) use.

Experimental plots were established over three seasons on two sites with different soil types using seed rates in the range 70 to 400 seeds/m2 to give different crop canopy characteristics. Measurements of the crop canopy characteristics (crop height and a Normalised Vegetation Difference Index - NDVI) made at each site and on three occasions in each season using a boom-mounted sensor system showed that there was a consistent relationship between NDVI and Leaf Area Index (LAI)/Green Area Index (GAI) measured with a manually operated light attenuation instrument. There was also a relationship between the variability of the NDVI signal and the tiller/shoot number which, for a given field calibration, could provide information about plant populations on a spatially selective basis. It was therefore concluded that crop canopy sensing systems could provide input data for managing root and stem lodging risk based on existing decision rules. The sensor systems gave the steepest responses for both the NDVI/LAI (GAI) and the variability of NDVI/tiller (shoot) counts relationships at the early measurement timings and this corresponds to the times at which decisions relating to nitrogen timings and PGR inputs are most relevant.

In the first year of the project, plots were managed using variable treatments determined from a set of decision rules defined for each of three separate strategies: standard, variable nitrogen timings and variable PGR inputs. Results were found to be dominated by weather effects during the 2006/07 season and therefore for the final two years of the work a more conventional randomised plot design was used. Crop yields from plots treated with no early nitrogen and a low PGR input consistently gave lower yields than those treated with a standard nitrogen split and either a low PGR input or a robust PGR input. Crop lodging was only recorded on the heavy land site with some lodging for all strategies and seed rates in the first year of the work, for the highest seed rate plots only in the second year and no lodging in the third year. Lodging risk analyses were therefore based on previously established relationships.