Sammanfattning

The study aims to estimate forage yield and quality parameters by fusing field spectroscopy data and crop height with regression-based mathematical models. Field experiments were carried out to obtain canopy spectral reflectance (CSR) of grass and clover mixtures. Additionally, grass height (H_grass) and clover height (H_clover) were used as auxiliary explanatory variables with CSR to estimate forage yield and quality. Variable importance in projection (VIP) was utilized for sensitive wavelength selection. Two chemometric methods, namely partial least squares regression (PLSR) and support vector machine (SVM), were implemented to build models using full spectra and sensitive wavelengths for estimating dry matter yield (DMY), in vitro true digestibility (IVTD), neutral detergent fiber (NDF), neutral detergent fiber digestibility (NDFD), acid detergent fiber (ADF), acid detergent lignin (ADL), crude protein (CP), crude protein yield (CPY), and botanical composition (BC). Of the total 235 samples, 157 samples were randomly selected for model calibration while the remaining 78 samples were used for model validation. Results showed that both PLSR and SVM could reasonably estimate forage yield and quality variables, although performances of PLSR were more stable in terms of R² and relative root mean square error (RRMSE) for both calibration and validation. Prediction performances of models using only full spectra data (PLSR_spec) and models also using crop height information (PLSR_spec+H) as model inputs were compared in this study. PLSR_spec+H presented higher R² and lower RRMSE than PLSR_spec models (e.g. R² improved from 0.83 to 0.90 for NDF and from 0.56 to 0.73 for IVTD, and RRMSE decreased from 8.14% to 6.58% for NDF and from 2.55% to 2.02% for IVTD). In addition, PLSR that used sensitive wavelengths and crop height (PLSR_wave+H) as model inputs also had good performance, although slightly worse than PLSR_spec+H. The results suggest that there is good potential to predict forage biomass and nutritive value by combining spectral and height variables with chemometric methods.

Nyckelord

forage crop; field spectroscopy; crop height; partial least squares regression; support vector machine; yield and quality

Publicerad i

Computers and Electronics in Agriculture
2022, Volym: 192, artikelnummer: 106571

SLU författare

• Morel, Julien

  • Institutionen för norrländsk jordbruksvetenskap, Sveriges lantbruksuniversitet
    

• Parsons, David

  • Institutionen för norrländsk jordbruksvetenskap, Sveriges lantbruksuniversitet
    

• Peng, Junxiang

  • Institutionen för norrländsk jordbruksvetenskap, Sveriges lantbruksuniversitet
    

UKÄ forskningsämne

Jordbruksvetenskap


Publikationens identifierare

DOI: https://doi.org/10.1016/j.compag.2021.106571

Permanent länk till denna sida (URI)

https://res.slu.se/id/publ/114998