 |
|
|
|
|
Early Detection of Exotic Pests and Diseases
in Asian Vegetables by Imaging Spectroscopyby B. Datt, A. Apan, and R. Kelly
August 2006
RIRDC Publication No 05/170 RIRDC Project No CEO-1A
The main aims of the project were to: (1) establish the links between pest and disease infestation and the changes in visible and infrared reflectance in selected vegetable crop canopies. (2) compile a reference dataset of healthy and disease/pest affected reflectance signatures for selected crops. (3) develop data analysis methods and tools for remote sensing of selected disease/ pest symptoms in crops. (4) assess the potential of aircraft and satellite-based detection of disease and pest activity in crops.
Field data were collected over selected vegetable crops in the Darling Downs region in Queensland during the summer growing season from December 2004-February 2005. A portable field spectrometer was used to measure the sunlight reflected by the leaves and canopies of several vegetable crops growing in the area. The field sampling procedure involved the measurement of leaf and canopy reflectance from healthy plants and plants with varying degrees of disease and pest damage. The measurements were done in a non-invasive way and no plant materials were removed or came in contact with the spectrometer. Data collection was done during sunlit, clear sky conditions.
The method allowed rapid collection of hundreds of reflectance spectra per day.
The dataset was processed and archived as a series of spectral libraries for the crops studied. Each crop spectral library consisted of reflectance spectra of healthy and diseased plants. An individual reflectance spectrum consisted of the fraction (0-1) of sunlight reflected at each wavelength of the solar spectrum. A single reflectance spectrum consisted of 2151 data points representing the reflectance values at 1-nanometer intervals across the 350 to 2500 nanometer wavelength region of the solar spectrum. Such high-resolution data capture the subtle changes in reflectance across the wavelengths caused by disease or other stress symptoms in the plant leaves.
Analysis of the data indicated that, for the crops and diseases studied, there was a clear separation between the diseased and healthy samples. These differences were further enhanced by mathematical transformations of the reflectance data such as the first derivative transform. A number of spectral indices were developed for specific crops, using those wavelengths of the spectrum which were most sensitive to the disease or pest infestation. These indices provide a simple method to quantify the level of disease activity within the range from no impact (healthy) to severe infestation.
The results from this pilot
study show that imaging spectroscopy has considerable potential for detecting
and monitoring the influence of diseases and pests in vegetable crops.
There is a need for more research to test this technology on a wider range
of crops and diseases than the ones studied in this project. Reference
data and algorithms/indices need to be developed for more crops and diseases
through field measurements. The methods need to be tested and validated
by repeating the field experiments over a number of growing seasons. Aircraft
and satellite imaging spectrometry data also need to be investigated for
disease/pest monitoring and mapping at scales ranging from individual paddocks
to whole farms and regions.
|