Wheat yield resilience metrics for sample 10km x 10km grid cells in England, 2008-2017

The original yield data were collected by Defra using a survey of a stratified random sample of farm holdings, as part of the annual June census of the English agricultural industry. Data were available for ten years (2008-2017), comprising average winter wheat yield per holding and coordinates locating each to 1 km. UKCEH obtained these data under a confidentiality agreement with Defra. Data were cleaned to remove anomalous yield values (zero values, whole crop silage harvests and obvious outliers).

Because a new random sample of farm holdings is drawn each year, few had consecutive data across 10 years. To analyse yield variation over time and account for local spatial variation in farming practices data were therefore aggregated to mean annual yield per 10km x 10km Ordnance Survey grid cell ('hectad'). This also preserves survey respondent anonymity by ensuring that no yield value can be attributed to the location of a single holding (a condition of use for the original Defra survey data, in line with the Agricultural Statistics Act 1979). Resilience metrics were then calculated across the time series (see below). Data included here give these metrics for 315 hectads with at least one sampled holding in each year of the time-series (i.e. the minimum data required to calculate the resilience metrics), although further analyses used only 137 that were deemed to be sufficiently well-sampled to avoid biases relating to sample size. This was based on thresholds for the minimum permissible number of samples per year and the number of years permitted to have this value, to determine which combination retained the maximum number of samples whilst removing the significant correlations between sample size and mean yield and between mean sample size and mean yield over time. A threshold value of no more than one year with a single sample and no more than two years having less than three samples was found to meet achieve this.