Chickpea (Cicer arietinum) is an important grain legume in semi-arid regions and water-stress is a major constraint to its productivity. Area under chickpea cultivation is growing but climate change toward greater aridity results in higher precipitation instability and risks yields. The ability to assess water potential can support irrigation decisions. Thus, improved ability to spatially assess plants water status can promote more efficient irrigation. The current study aims to assess water potential, leaf area index (LAI) and grain yield by spaceborne, airborne and ground spectra sensors. During the growing season of 2018-2019, field experiments were conducted in two locations, representing different climatic conditions in Israel. Five irrigation regimes were applied: 50%, 75%, 100%, 120% and 140% of Penman-Monteith evapotranspiration were implemented at the Gilat research station and in a commercial field (Kibbutz Or-HaNer). Plants were characterized weekly for water potential and LAI, and grain yield data was obtained at the final harvest. Canopy reflectance was acquired with a MicroSatellite VENµS (11 spectral bands, 420-910 nm), a drone mounted Rededge MicaSense camera (5 spectral bands, 470-860 nm; only in Gilat) as well as hyper-spectral (ASD, 350-2500 nm) ground level, dual-field of view. The multispectral images as well as hyperspectral data were pre-processed to the level of reflectance. The VENµS and ground level hyperspectral data were divided to calibration and validation data sets while the multispectral 5 bands imagery was analyzed only for calibration. Leaf water potential (LWP), LAI and grain yield values were showing differences between most of the irrigation treatments. The VENµS data based on partial least squares regression (PLS-R) analysis for water potential, LAI and grain yield resulted in R2 and RMSEV of 0.8 and 0.217 MPa; 0.63 and 0.74 m2m-2; and 0.82 and 0.44 t ha-1, respectively. The PLS-R analysis for hyperspectral data for LWP, LAI and grain yield resulted in R2 and RMSEV of 0.74 and 0.18 MPa; 0.80 and 0.89 m2m-2; and 0.84 and 0.45 t ha-1, respectively. The multispectral 5 bands imagery was used to calculate vegetation indices. The best vegetation index for LWP, LAI and grain yield were OSAVI, NDVI and TCARI, respectively, R2 and RMSE values of, 0.56 and 0.168 Mpa, 0.69 and 0.78 m2m-2, 0.874 and 0.38 t ha-1, respectively. It is suggested that VENµS, 5 bands airborne imagery and hyperspectral ground level data are useful for evaluation of morpho-physiological traits and grain yield in chickpea. The VENµS and the 5 bands imagery are showing smaller RMSE than the hyperspectral data. In the next growing season the repeatability of the predictions and correlations will be validate and explored in field experiment.
African Conference on Precision Agriculture (AfCPA) Presentation
Spectral assessment of chickpea morpho-physiological traits from space, air and ground
The Hebrew University of Jerusalem