Estimating fAPAR from Nine Vegetation Indices for Irrigated and Nonirrigated Faba Bean and Semileafless Pea Canopies

Abstract

Canopy reflectance measurements of semileafless pea and faba bean crops grown under irrigated and rain-fed conditions were taken with a field spectroradiometer. These legume crops have different canopy architectures. During crop growth nine vegetation indices (VIs) were calculated from reflectance measurements taken in TM2, TM3, TM4, and TM5 bands. Five VIs were ratio indices, (RVI, NDVI, SAVI2, TSAVI, and RDVI), two were orthogonal indices (PVI and GVI3), and the other two (STVI and CRVI) included the mid-infrared band. The fraction of absorbed photosynthetically active radiation (fAPAR) from a ceptometer and reflectance were measured simultaneously. Linear, exponential, and power relationships between VI and fAPAR were constructed to evaluate the capability of each VI to estimate fAPAR in these crops. The influence of solar angle on VI determination was also studied, using reflectance measurements at several hours during the day at two different crop growth stages. Results show that, in both species, different relationships are needed for pre-and postmaximum leaf area index (LAImax) phases of development. In pre-LAImax of both species, all VI correlated highly with fAPAR, but the coefficients of determination ( R 2) for ratio indices were generally higher than orthogonal indices (PVI and GVI3) and indices with the TM5 band (STVI and CRVI). Among ratio indices, in pre-LAImax, R 2 did not differ significantly between those that included the soil effect (SAVI2 and TSAVI) and those that did not (RVI, NDVI, and RDVI). Relationships with indices that include the TM5 band (STVI and CRVI) showed a great slope, increasing the sensitivity of the estimation of fAPAR to errors and frequency of the reflectance measurements. During pre-LAImax phase, curvilinear relationships are better for pea canopies, while linear ones were found more appropriate for faba beans. In post-LAImax, linear relations were in general better, especially in peas. Relationships did not differ between irrigated and rainfed treatments within crops, even in early stages of growth with poor soil cover, so that only one was proposed for each species. All VI values were greatly affected by the hour of measurement at solar angles greater than 45°. The indices most affected by solar angle changes were STVI and CRVI. We can conclude that simple indices, RVI and NDVI, based on reflectance measurements in TM3 and TM4 bands can be used to accurately asses canopy development in faba bean and semileafless peas, allowing good and fast estimations of fAPAR and LAI, both important parameters in crop physiology and modeling. More research in different type of soils is needed to check if ratio, SAVI2 and TSAVI, or orthogonal, PVI and GVI3, indices that include the effect of soils can improve the estimation of fAPAR in these two crops.