Contribution of chlorophyll fluorescence to the apparent vegetation reflectance.

Current strategies for monitoring the physiologic status of terrestrial vegetation rely on remote sensing reflectance data, which provide estimates of vigor based primarily on chlorophyll content. Chlorophyll fluorescence (ChlF) measurements offer a non-destructive alternative and a more direct approach for diagnosis of vegetation stress before a significant reduction in chlorophyll content has occurred. Thus, technology based on ChlF may allow more accurate carbon sequestration estimates and earlier stress detection than is possible when using reflectance data alone. However, the observed apparent vegetation reflectance (Ra) in reality includes contributions from both the reflected and fluoresced radiation. The aim of this study is to determine the relative contributions of reflectance and ChlF fractions to Ra in the red to near-infrared region (650-800 nm) of the spectrum. The practical objectives of the study are to: 1) evaluate the relationship between ChlF and reflectance at the foliar level for corn, soybean and maple; and 2) for corn, determine if the relationship established for healthy vegetation changes under nitrogen (N) deficiency. To obtain generally applicable results, experimental measurements were conducted on unrelated crop and tree species (corn, soybean and maple) under controlled conditions and a gradient of inorganic N fertilization levels. Optical reflectance spectra and actively induced ChlF emissions were collected on the same foliar samples, in conjunction with measurements of photosynthetic function, pigment levels, and carbon (C) and N content. The spectral trends were examined for similarities. On average, 10-20% of Ra at 685 nm was actually due to ChlF. The spectral trends in steady state and maximum fluorescence varied significantly, with steady state fluorescence (especially red, 685 nm) showing higher ability for species and treatment separation. The relative contribution of ChlF to Ra varied significantly among species, with maple emitting much higher fluorescence amounts, as compared to corn and soybean. Steady state fluorescence from individual red and far-red emission bands (F685 and F740, respectively) and their ratio consistently enabled species separation. For corn, the relative ChlF fraction increased in concert with the nutrient stress levels from <2% for non-stressed foliage to >7% for severely N deficient plants. Steady state ChlF at 685 nm provided optimal N treatment separation. This study confirms the trends in the steady state red/far-red ratio (F685s/F740s) associated with N deficiency and vegetation stress, previously established using active single narrow band excitation.

Assessment of vegetation stress using reflectance or fluorescence measurements.

Current methods for large-scale vegetation monitoring rely on multispectral remote sensing, which has serious limitation for the detection of vegetation stress. To contribute to the establishment of a generalized spectral approach for vegetation stress detection, this study compares the ability of high-spectral-resolution reflectance (R) and fluorescence (F) foliar measurements to detect vegetation changes associated with common environmental factors affecting plant growth and productivity. To obtain a spectral dataset from a broad range of species and stress conditions, plant material from three experiments was examined, including (i) corn, nitrogen (N) deficiency/excess; (ii) soybean, elevated carbon dioxide, and ozone levels; and (iii) red maple, augmented ultraviolet irradiation. Fluorescence and R spectra (400-800 nm) were measured on the same foliar samples in conjunction with photosynthetic pigments, carbon, and N content. For separation of a wide range of treatment levels, hyperspectral (5-10 nm) R indices were superior compared with F or broadband R indices, with the derivative parameters providing optimal results. For the detection of changes in vegetation physiology, hyperspectral indices can provide a significant improvement over broadband indices. The relationship of treatment levels to R was linear, whereas that to F was curvilinear. Using reflectance measurements, it was not possible to identify the unstressed vegetation condition, which was accomplished in all three experiments using F indices. Large-scale monitoring of vegetation condition and the detection of vegetation stress could be improved by using hyperspectral R and F information, a possible strategy for future remote sensing missions.