Estimation of Biophysical Variables in Alfalfa (Medicago sativa L.) using Spectral Information and Simple Linear Regression Models

Abstract

The fraction of vegetation cover (FVC) and the Leaf Area Index (LAI) are biophysical variables closely related to the evapotranspiration rate of crops and their biomass production. Despite their importance, they are usually not measured directly due to their time-consuming and costly nature; however, they can be estimated on a large scale and near-real-time using spectral information captured in satellite images, although this requires prior validation at the field level. Simple linear regression models were developed and validated using 13 vegetation indices (VI) related to LAI and FCV in alfalfa (Medicago sativa). Three spectral information sources were used: (i) reflectance to derive the VI; (ii) digital photographs processed with the Canopeo application to estimate FVC; and (iii) direct LAI measurements using a ceptometer. Six field samples were taken between March and June, 2020 in four alfalfa plots located in Palmar de Bravo, Puebla, Mexico. To evaluate each VI, the determination coef ficient (R2) and the root mean square error (RMSE) were used. We found that the most suitable VI for estimating FVC was the VIgreen (Green Vegetation Index, R2 =0.987 and RMSE = 0.093). For the LAI, the VI that performed best was the NDVI (Normalized Dif ference Vegetation Index, R2 = 0.935 and RMSE = 0.746). The results showed the suitability and practical utility of spectral information for estimating biophysical variables in alfalfa cultivation and their monitoring.