Canopy transmittance inversion using a line quantum probe for a row crop

Abstract

An 80-photo-sensor line probe was used to measure canopy transmittance of direct solar radiation under clear skies and to estimate plant area index (PAI) by inversion in row crops. The 80 sensors were found to have significant differences in sensitivity and offset. Standard deviations of offsets and calibration intercepts were approximately 5 μmol m −2s −1 for two Decagon Sunlinks and greater for three CID CI150s. Measurement of ‘gap frequency’ with the line probe by counting sensors above an appropriate threshold gave almost identical results to measurement of transmittance of direct photosynthetically active radiation (PAR), for high transmittance measured at high solar angles, but at low solar angles the ‘gap frequency’ was lower than transmittance. This was attributed to the intensity of irradiance in small gaps not exceeding the threshold. Variations in non-beam PAR were mostly in the direction perpendicular to the rows whereas those parallel to the row were small. A sample of four point measurements of diffuse irradiance at each position across the row was found to give an adequate mean for correcting transmittance values. Linear averaging of transmittance perpendicular to the row at high solar angles caused overestimates of transmittance, because of row clumpiness, leading to errors in the estimated mean leaf angle. Averages taken parallel to the row were much closer to theoretical predictions for a homogeneous canopy, and are therefore recommended. When measuring transmittance across the row, the number of sample positions required for 5% resolution of inversion estimates depends on row shadow length perpendicular to the row. When shadow length is less than row width, five positions are recommended; when all the ground is shaded two positions are sufficient. To obtain a standard error of less than 0.4 in estimated PAI, we recommend eight repetitions along the row. Inversion estimates of PAI from measurements made according to the recommended protocol were highly correlated with destructive measurements of PAI ( r 2 = 0.93). However, at high PAI the results still underestimated PAI by approximately 20%, presumably because of clumping of plant parts in the row.