Regression forecast of UK harvest wheat quality from the preceding winter North Atlantic Oscillation

contributed by Mark D Atkinson¹, Peter S Kettlewell¹, Philip D Hollins¹ and David B Stephenson²

¹Crop and Environment Research Centre, Harper Adams University College, Newport TF10 8NB, UK,

²Department of Meteorology, University of Reading, PO Box 243, Reading RG6 6BB, UK

Crop growth is dependent on a complex array of climatic influences over a prolonged period of time and, as a consequence, measures of crop performance can integrate a number of climatic factors over time. Here, a measure of crop quality can be predicted with a considerable lead-time.

The breadmaking quality of wheat grown in the United Kingdom is strongly related to the preceding winter North Atlantic Oscillation (Kettlewell et al. 1999). The quality factors under study, which are of considerable importance in determining saleability and price, pertain to the grain at harvest (August or September). The lead time of the forecast presented here is therefore between seven and eight months. The forecast presented is for one quality factor, specific weight (also known as bushel- or test-weight), and is for a UK national average. The forecast is based on annual quality survey data which were first collected in 1974 (Atkinson et al. 2001).

Various NAO indices have been compared, but for operational forecasting purposes, the index must be available as soon after the end of January as possible. An NAO index was therefore constructed from NCEP-NCAR reanalysis monthly data http://iridl.ldeo.columbia.edu/SOURCES/.NOAA/.NCEP-NCAR/ }http://iridl.ldeo.columbia.edu/SOURCES/.NOAA/.NCEP-NCAR/). Mean sea-level pressures over a large region near Iceland were averaged (30°W to 0°W, 60°N to 70°N) as were those over a region near the Azores (50°W to 5°W, 30°N to 40°N). January MSL pressures were standardised over the period 1948-2000 and the northern pressure value subtracted from the southern for each year.

Fig. 1 shows time series of observed specific weight and the preceding January NAO index. The forecast models were developed using linear regression weighted with wheat quality survey sample numbers. A combination of the November and January NAO indices has been found to have the most forecasting skill, although the January NAO accounts for the majority of the variance. The relationship between the specific weight and the January NAO index (with November NAO accounted for) is shown in Fig. 2.

Retroactive real time validation (Unger 1996) has been used to assess the skill of the forecasting procedure. Starting with the period 1974-1988, a forecast for 1989 was made. Proceeding then to the period 1974-1989, a forecast was made for 1990 and so on up to 2001. The correlation between these forecasts and the observed values was 0.78, significant at P < 0.01. The correlation for the persistence forecast (assuming the next year's specific weight to be the same as the present) was 0.15 and not significant.

The first operational forecast was made in March 2001 for the harvest of 2001 and the result can be seen at http://www.harper-adams.ac.uk/wheatqualityforecasts/ }http://www.harper- adams.ac.uk/wheatqualityforecasts/. Our prediction for the summer of 2002, based on the November NAO index of 0.167 and the January NAO index of 0.08, is that specific weight will be 76.3 kg hl-1. The 75% prediction limits are 75.5 and 77.2

Ongoing research is addressing the physical mechanism of this forecasting method (Kettlewell et al. 2001). Summer weather is known to be an important determinant of specific weight (Smith and Gooding 1999). There is evidence for a relationship between the winter NAO index and summer precipitation (Kettlewell et al. 2002) and hydrological factors (Wedgebrow et al. 2002) over the UK. Modelling of the effect of winter and spring temperatures and evapotranspiration on wheat growth and development is starting to give some insights into how the winter climate may also have a long-range effect on development of the wheat crop in the harvest period.

References :

Atkinson, M.D., P.S. Kettlewell, P.D. Hollins, D.B. Stephenson, and S.T. Parsons, 2001: Forecasting quality and breadmaking premium of wheat from the winter North Atlantic Oscillation. Aspects of Applied Biology 64, Wheat Quality, published by The Association of Applied Biologists, c/o Horticulture Research International, Wellesbourne, Warwick CV35 9EF, UK, pp 127-128.

Kettlewell, P.S., R.B. Sothern and W.L. Koukkari, 1999: U.K. wheat quality and economic value are dependent on the North Atlantic Oscillation. Journal of Cereal Science, 29, 205-209.

Kettlewell, P.S., M.D. Atkinson, D.B. Stephenson and P.D. Hollins, 2001: Climatic factors mediating the effect of the winter North Atlantic Oscillation on wheat quality. Aspects of Applied Biology 64, Wheat Quality, published by The Association of Applied Biologists, c/o Horticulture Research International, Wellesbourne, Warwick CV35 9EF, UK, pp 129-130.

Kettlewell, P.S., D.B. Stephenson, M.D. Atkinson, and P.D. Hollins, 2002: UK Wheat harvest quality and summer rain in relation to the North Atlantic Oscillation. Weather, submitted.

Smith, G. P. and M.J. Gooding, 1999: Models of wheat grain quality considering climate, cultivar and nitrogen effects. Ag. For. Meteorol., 94, 159-170.

Unger, D.A., 1996: Skill assessment strategies for screening regression predictions based on a small sample size. In: 13th Conference on probability and statistics in the atmospheric sciences. Feb. 21-23, 1996, San Francisco. American Meteorological Society, pp. 260-267.

Wedgebrow, C., R.L. Wilby, H.R. Fox and G. O'Hare, 2002: Prospects for seasonal forecasting of Summer Drought and Low River Flow Anomalies in England and Wales. Int. J. Climatol., 22, 219-236.