Page Banner

United States Department of Agriculture

Agricultural Research Service

Research Project: Hyperspectral and Multispectral Image Analyses of Potatoes under Different Nutrient Management with Center Pivot Irrigation

Location: Vegetable and Forage Crops Production Research

2012 Annual Report


1a.Objectives (from AD-416):
Non-destructive estimation of nutritional status of potato canopy using multispectral imaging and prediction of tuber yield and quality response to variable nutrient management under pivot irrigation.


1b.Approach (from AD-416):
Spectral characteristics of vegetation are a quantitative measure and can offer a non-destructive method to assess crop nutrition; biomass production; and, in turn, yield and quality of crop products. This type of sensing technology has been successfully developed for detecting nitrogen stress in agronomic crops, such as corn and rice. This technology can be modified for its application to potato production under center pivot irrigation. In this research, the following steps will be used to investigate the feasibility of developing a multispectral imaging based sensing system for estimation of nutritional status of potato canopy under variable nutrient management programs and, in turn, predict biomass production, tuber yields and tuber quality parameters. Furthermore, multispectral image sensing can be an efficient tool of non-destructive evaluation of potential non-uniformity in water distribution through sprinklers in center pivot irrigation system. 1. Collect multispectral images of potato canopy grown under different nutrient management programs; 2. Analyze the spectral characteristics of the canopy and search for a trend of such characteristic change with the corresponding nutrient management programs; 3. Analyze relationships between the spectral information carried in multispectral images and nutritional status of the plants monitored based on the petiole analyses and destructive plant sampling; and 4. Define a calibration equation for quantitatively estimating the level of nutritional statuses based on multispectral images. Documents SCA with WSU.


3.Progress Report:

In this study, hyprespectral imaging was used as a non-destructive method for detecting water stress in potato plants, and relates to sub-objective 2.B. of the in-house project, "Evaluate the effects of deficit irrigation practices on potato yield and tuber quality". The reflectance plots of plants under different water treatment levels showed consistent trend within the spectral range from 500 nm to 700 nm. Lowest reflectance level was observed in this spectral region for the treatment with lowest level of soil water content. The spectral reflectance showed an increasing trend with increasing soil water content. In the near infrared (NIR) region, plant reflectance at the lowest soil water content level was observed to be maximum in most cases. Various spectral indices were calculated in this research using reflectance data to study the correlation between those indices and plant water stress. The indices studied included Water Index, Normalized Difference Vegetation Index (NDVI), Modified NDVI, Red Edge NDVI, Simple Ratio Index (SRI), Modified Red Edge SRI, and Vogelmann Red Edge Indices. These indices were evaluated for all replications from images taken on 19th April, and 4th, 8th, 9th, 10th and 11th May. Correlation coefficients between these indices and soil moisture content were determined for each day of data collection. Modified Red Edge SRI showed the highest correlation (-0.81) with soil moisture. Modified NDVI had the second highest correlation of -0.79. These results showed a promise to predict water stress level in potato plants non-destructively. The spectral indices having high correlation with plant water stress can be used to develop remote sensing devices to monitor plant stress. The wavelength associated with a particular spectral index can be incorporated into a hand held device to predict plant water stress in real time.


Last Modified: 9/10/2014
Footer Content Back to Top of Page