Assessing structural effects on PRI for stress detection in conifer forests

Abstract

The retrieval of indicators of vegetation stress from remote sensing imagery is an important issue for the accurate assessment of forest decline. The Photochemical Reflectance Index (PRI) has been demonstrated as a physiological index sensitive to the epoxidation state of the xanthophyll cycle pigments and to photosynthetic efficiency, serving as a proxy for short-term changes in photosynthetic activity, stress condition, and pigment absorption, but highly affected by illumination conditions, viewing geometry and canopy structure. In this study, a diurnal airborne campaign was conducted over Pinus sylvestris and Pinus nigra forest areas with the Airborne Hyperspectral Scanner (AHS) to evaluate the effects of canopy structure on PRI when used as an indicator of stress in a conifer forest. The AHS airborne sensor was flown at two times (8:00 GMT and 12:00 GMT) over forest areas under varying field-measured stress levels, acquiring 2 m spatial resolution imagery in 80 spectral bands in the 0.43–12.5 μm spectral range. Five formulations of PRI (based on R 531 as a xanthophyll-sensitive spectral band) were calculated using different reference wavelengths, such as PRI 570 (reference band R REF = R 570), and the PRI modifications PRI m1 (R REF = R 512), PRI m2 (R REF = R 600), PRI m3 (R REF = R 670), and PRI m4 (R REF = R 570, R 670), along with other structural indices such as NDVI, SR, OSAVI, MSAVI and MTVI2. In addition, thermal bands were used for the retrieval of the land surface temperature. A radiative transfer modeling method was conducted using the LIBERTY and INFORM models to assess the structural effects on the PRI formulations proposed, studying the sensitivity of PRI m indices to detect stress levels while minimizing the effects caused by the conifer architecture. The PRI indices were related to stomatal conductance, xanthophyll epoxidation state (EPS) and crown temperature. The modeling analysis showed that the coefficient of variation (CV) for PRI was 50%, whereas the CV for PRIm1 (band R512 as a reference) was only 20%. Simulation and experimental results demonstrated that PRI m1 (R REF = R 512) was less sensitive than PRI (R REF = R 570) to changes in Leaf Area Index (LAI) and tree densities. PRI 512 was demonstrated to be sensitive to EPS at both leaf (r 2 = 0.59) and canopy level (r 2 = 0.40), yielding superior performance than PRI 570 (r 2 = 0.21) at the canopy level. In addition, PRI 512 was significantly related to water stress indicators such as stomatal conductance (Gs; r 2 = 0.45) and water potential (Ψ; r 2 = 0.48), yielding better results than PRI 570 (Gs, r 2 = 0.21; Ψ, r 2 = 0.21) due to the structural effects found on the PRI 570 index at the canopy level.