研究目的
To develop a rapid and reliable method for determining the nitrogen (N) status of pear leaves during the growing season using visible and near-infrared reflectance (Vis/NIR) spectroscopy for timely fertilization to improve yields and fruit quality.
研究成果
The PLS regression model with preprocessing of full spectra using MAS is optimal for modeling N concentration in fresh pear leaves, providing a rapid and reliable method for N status determination. This method can be useful for fertilization management in pear orchards, with high correlation coefficients and low mean relative errors in validation.
研究不足
The study was limited to two cultivars of pear, and the model developed may not be applicable to other pear cultivars. Future research should include different varieties, modeling methods, ages of pear trees, and N fertilization doses for a more comprehensive evaluation.
1：Experimental Design and Method Selection:
The study used partial least squares (PLS) regression to evaluate Vis/NIR spectra of fresh pear leaves for determining N concentration. Different preprocessing techniques were studied to modify spectra.
2：Sample Selection and Data Sources:
450 leaf samples from two pear cultivars were collected, with 294 samples for calibration and 180 for validation after excluding outliers using principal component analysis.
3：List of Experimental Equipment and Materials:
A portable field spectrometer FieldSpec 3 (ASD Co., Ltd., Boulder, CO) was used for spectral measurements. A Teflon white reference panel was used for setting up maximum reflectance conditions.
4：Experimental Procedures and Operational Workflow:
Spectral measurements were taken in a laboratory setting on the ventral side of leaves. The scan number for each spectrum was set to 10 at the same position, with the average value of 20 spectra used as the final reflectance.
5：Data Analysis Methods:
Original spectra were modified by filtering out wavelength segments highly correlated with N concentration, using ViewSpec for data derivation and transformation, and Unscrambler 9.7 for preprocessing data by various techniques including MAS, SGS, MSC, normalization, baseline, noise, and SNV.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
