Effects of view and sun directions and reproductive organs on crop reflectance

The radiance reflected by a surface may significantly vary as a function of the geometry of observation and canopy structure. These bidirectional effects have received regular attention when interpreting remote sensing images, particularly regarding natural surfaces including the vegetation. After the flowering stage, a number of crops present reproductive organs (ears (wheat), male panicle (maize), flowers (sunflower)) at the top of the canopy that have particular structure and optical properties. This impacts the radiation regime in a significant manner which may induce artefacts in the temporal course of the quantities derived from reflectance and bias the estimates of canopy characteristics. The objective of this study is to quantify the influence of these reproductive organs at the top of the canopy on the estimation of some canopy characteristics. Specific attention will be paid on the directional variation of this effect. A dedicated field experiment was conducted in 2016 and 2017 over wheat, maize and sunflower, few times after flowering when green leaves still represent the main contribution of the green area index. An approximately 10m×10m square homogeneous area was selected in these fields. In the center of this area, a reference panel of 0.6*0.6 m² was positioned. An hexacopter drone equipped with a 6 bands Airphen-3 multispectral camera was flown three times during the day to get a range of sun zenith angles (SZA= [30°, 45° , 60°]. Their field of view is approximately ±15°. For each of the three elevations, the UAV flew in concentric circles such that the UAV is always at the same distance from the center of each circle. The diameter and altitude are computed to get a range of view zenith angles (VZA=[0°, 15°, 30°, 45° and 60°]). The experimental setup allows to take photos on the reference panel in each image in order to get accurate calibration of the bidirectional reflectance factor. The images taken from UAV were processed using the phenoscript pipeline that aligns the images and automatically extracts the reference panel and the surrounding wheat crop. The results show strong BRDF effects over the canopy, with a very pronounced hot-spot. These BRDF effects could be well depicted by kernel directional empirical models. Additionally, in an area where reproductive organs were intentionally removed, the effect of these organs on the BRDF is demonstrated to be also very significant. Conclusions on the methodology and on the results are drawn, with due attention on the consequences on the interpretation of remote sensing observations.