Predicting bacterial wilt disease of tomato plants using remotely sensed thermal imagery

• Published in: Journal of Agricultural Meteorology Vol. 61; no. 3; pp. 153 - 164
• Main Authors: Chiwaki, K.(Okayama-ken. General Agriculture Center, Sanyo (Japan). Agricultural Experiment Station), Nagamori, S, Inoue, Y
• Format: Journal Article
• Language: English
• Published: Tokyo The Society of Agricultural Meteorology of Japan 2005; Japan Science and Technology Agency
• Subjects: ANALISIS DE IMAGENES; ANALYSE D'IMAGE; Bacteria; Bacterial wilt disease; Disease prediction; FEUILLE; FLETRISSEMENT; FORECASTING; HOJAS; IMAGE ANALYSIS; Infrared thermal imagery; LEAVES; LYCOPERSICON ESCULENTUM; MARCHITEZ; PSEUDOMONAS; REMOTE SENSING; SINTOMAS; SYMPTOME; SYMPTOMS; TECHNIQUE DE PREVISION; TECNICAS DE PREDICCION; TELEDETECCION; TELEDETECTION; TEMPERATURA; TEMPERATURE; TERMOGRAFIA; THERMOGRAPHIE; THERMOGRAPHY; Tomato; WILTS
• ISSN: 0021-8588; 1881-0136
• DOI: 10.2480/agrmet.61.153

The objective of this study was to investigate basic relationships for early diagnosis of tomato disease (bacterial wilt disease) by use of infrared thermal imagery. We obtained infrared thermal images of un-grafted (diseased) and grafted (non-diseased) tomato plants over the period extending before the appearance of external symptoms to until the diseased conditions. The temperature differentials between diseased and non-diseased tomato plants were analyzed by two methods: First, visual diagnosis of the pseudo-color thermal images; and second, histogram analysis of leaf temperatures from the images. Relative temperature increases of the diseased plants could be recognized visually on the thermal images as early as 5 days before the appearance of external symptoms. Analyses of the temperature histograms for diseased and non-diseased plants in these images showed that the mean leaf temperature in diseased plants was no less than 0.8 deg C higher than in non-diseased plants, and the minimum leaf temperature in diseased plants was no less than 0 deg C higher than in non-diseased plants. No significant difference was found in the maximum and the mode leaf temperatures. Applying these two criteria, i.e., no less than 0.8 deg C higher in the mean leaf temperature, and no less than 0 deg C higher in the minimum leaf temperature, to the other images over the entire period, we found that some relative temperature increase could be detected in the diseased plants as early as 12 days, in the earliest case, before the appearance of external symptoms. Results suggested that remotely sensed thermal imagery would be applied to early detection of the bacterial wilt disease of tomatoes.