1Central Science Laboratory, Sand Hutton, York YO4 1LZ, UK; 2ADAS Pwllpeiran, St Agnes Road, Cardiff CF4 4YH, UK; 3University of Hertfordshire, College Lane, Hatfield AL10 9AB, UK

Backgound and objectives
Late blight, caused by Phytophthora infestans (Mont.) de Bary. is the major fungal disease of potatoes in the UK. Control is usually by the use of a routine programme of fungicide sprays starting shortly after emergence and continuing until haulm destruction; the average number of fungicide applications made onto UK crops in recent years is between 7 and 9 [1]. Forecasting schemes for late blight interrogate recent weather observations in an attempt to quantify the actual risk of infection, so as to advise growers when to begin and, in some cases, schedule their fungicide applications. On-farm weather stations enable the collection of farm-specific readings but the responsibility for their accuracy rests with the grower. This paper describes the effect of inaccurate sensors on the performance of four forecasting schemes using data from an in-crop weather station.

Materials and methods
Using in-crop hourly weather data from a potato field the forecasting schemes of Smith [2], Schrödter & Ullrich [3], Blitecast [4] and NEGFRY [5] were evaluated with respect to the date of first fungicide application and the total number of sprays advised. The readings for hourly temperature and relative humidity were altered by 0.5ºC and 2% respectively, these figures correspond to the claimed accuracy of the sensors. The schemes were then run with a negative distortion, first with the temperature altered, second with the humidity altered and finally with both parameters altered. All three runs were then repeated with the alteration imposed in a positive direction.

Results and conclusions
A negative distortion of the temperature and/or humidity parameters resulted in a delay in the recommendation of the first fungicide application. With the NEGFRY scheme, sprays were delayed by 4 days for both temperature+relative humidity (r.h.) and r.h. alone, Blitecast was delayed by 4 days for temperature and r.h. and by 3 days for r.h. alone. The Schrödter & Ullrich scheme was delayed by 7 days for temperature+r.h. and r.h. alone, while the Smith Period was delayed by 3 days for temperature+r.h. and r.h. alone. A positive distortion of temperature and/or humidity brought forward the date of first application. This affected Blitecast by 14 days for temperature+r.h. and r.h. alone and 13 days for r.h. and temperature separately. NEGFRY scheduled sprays beginning 10 days earlier for temperature+r.h. and 9 days earlier for r.h. alone. Schrödter & Ullrich advised that the first application be advanced by 2 days for temperature+r.h. and r.h. alone, but the Smith Period guidance was unaffected.

Spray numbers were increased by a positive distortion, Blitecast and NEGFRY responded identically with two additional sprays for temperature+r.h. and one extra spray for temperature and r.h. alone, while Schrödter & Ullrich was unaffected. A negative distortion decreased spray numbers; Blitecast by two for temperature+r.h. and one for r.h. alone. NEGFRY and Schrödter & Ullrich lost one spray for both temperature+r.h. and r.h. alone but the Smith Period was unaffected by any sensor distortion, neither positive nor negative. These results suggest that a basic scheme such as Smith would be more robust than those which rely on intricate calculations when sensor accuracy could not be guaranteed.

1. Hardwick NV, Turner JA, 1996. Abstracts 13th Triennial Conference, European Association of Potato Research 1996 Veldhoven: EAPR, pp. 631-632.
2. Smith LP, 1956. Plant Pathology 5, 83-87.
3. Schrödter H, Ullrich J, 1967. Agricultural Meteorology 4, 119-135.
4. Krause RA, Massie, LB, Hyre RA, 1975. Plant Disease Reporter 59, 95-98.
5. Hansen JG, Andersson B, Hermansen A, 1995. Proceedings of Phytophthora 150, 1995. Dublin: EAPR, pp. 220-225.