Search

The carbon dioxide (CO2) fertilization effect has captured worldwide attention, owing to its tremendous potential to challenge existing predictions of future climate. However, quantifying the CO2 fertilization effect (CFE) has proven to be challenging, given that it is closely entangled with other ecological and environmental processes. Recent years have witnessed significant advances with breakthroughs using theoretical methods to infer the CFE from eddy covariance tower measurements. Building on earlier findings, this study presents an innovative approach that utilizes explainable machine-learning techniques—describing the partial dependence of the response variable to each explanatory variable—to quantify the global CFE from remote-sensing platforms with an averaged $R ^{2}$ of 0.85. This study provides the first data-driven evidence of the global CFE and confirms the potential for extrapolation to the globe. The findings suggest that: 1) the employment of satellite vegetation proxies contributed to more than 50% of the fitting of gross primary productivity (GPP) and 2) the manifestation of the CO2 fertilization impact demonstrated heterogeneity among various types of ecosystems, and in some cases, an adverse effect was detected in broadleaf forests. Our results have significant implications for the preservation and protection of terrestrial ecosystems, particularly for a carbon-neutral future. This study, therefore, provides a valuable contribution to the growing body of knowledge in this area and highlights the potential of innovative analytical techniques to address complex ecological challenges. [ABSTRACT FROM AUTHOR]

To meet the demand of standard spectral information for the subtropical forest, a three-level spectral library including leaf spectra of 67 typical subtropical tree species was built using two spectrometers. Towards the spectra measured by different spectrometers, the spectra consistency was tested using one-way Analysis of Variance (ANOVA) and Jeffries-Matusita distance (J-M distance). The optimal band range and spectral dimension reducing method for subtropical tree species discrimination were assessed using J-M distance based on original spectra and derivative spectra on visible-near infrared band (VNIR, 350–1000 nm), shortwave infrared band (SWIR, 1000–2500 nm) and full band (350–2500 nm). It was found that: (1) the spectra measured by the two spectrometers were significantly different, but the derivative spectra were not; (2) the SWIR spectral band was optimal for separating these subtropical tree species; (3) the selection method outperformed transform method for reducing spectra dimension and separating subtropical tree species [ABSTRACT FROM AUTHOR]

This study aimed to analyze the spectral features of alpine wetland grasses and evaluate three types of spectral feature computational methods. In situ spectra of alpine wetland grasses were collected in the northeastern Qinghai–Tibetan Plateau (Niaodao Wetland of International Importance and Ruoergai Wetland of International Importance), and spectral features were extracted using the revised vegetation spectral feature extraction (VSFE) model, wavelet transform, and the characteristic band selection method based on trivariate mutual information. These methods were then compared for their ability to amplify spectral separability and reflect grass properties, using the Jeffries–Matusita distance and relative vegetation parameters, respectively. The results show distinct spectral features of alpine wetland grasses within the band from 717.80 nm to 1050 nm. These differences are chiefly caused by leaf area index and chlorophyll content, but not plant taxonomy. Among the three spectral feature analysis methods, the revised VSFE model provides the most information for separating different alpine wetland grasses, as well as indices reflecting chlorophyll, leaf area index, and nitrogen. However, only 70 alpine wetland grass species pairs among 138 can be clearly classified within the band from 350 nm to 1050 nm. Further study on wetland grass species classification over wider band ranges should be implemented. This study also provides suggestions for selecting a spectral feature computational method in hyperspectral data processing. [ABSTRACT FROM PUBLISHER]