Dual isotopes of sulfate (δ³⁴SSO₄ and δ¹⁸OSO₄), along with isotopes in water and trace elements of geothermal waters, are systematically investigated to quantitatively elucidate sulfate sources and oxygen and sulfur isotopic behaviors during deep groundwater circulation and to constrain reservoir temperatures in the Jimo nonvolcanic geothermal system on the eastern coast of China. The results show that δ³⁴SSO₄ and δ¹⁸OSO₄ values in geothermal waters ranged from −21.0 to 5.7‰ and from 1.1 to 8.8‰, respectively. An increase in SO₄ concentrations (140–796 mg/L) with a systematic decrease in δ³⁴SSO₄ and δ¹⁸OSO₄ values was observed along the flow path from the central to eastern and western parts. The sulfate in the Middle Group was predominantly from atmospheric deposition, with sulfide oxidation contributions of <27%. In contrast, 80–85% of SO₄ in the Eastern Group is derived from pyrite oxidation. In the Western Group, the oxidation of multiple metal sulfides contributed 43–66% of SO₄. Sulfate oxidation and mixing of shallow groundwater caused reservoir temperatures to be underestimated by 9 ± 6–14 ± 16% using silica and K-Mg geothermometers but overestimated by up to 52–62% using sulfate–water oxygen isotope geothermometers. The estimated average target reservoir temperature was 144 ± 8 °C, with geothermal waters circulating to depths of 3.6–4.6 km. This study offers new insights into the significant impact of sulfate-related processes on geothermometric estimates, a factor often overlooked when using aqueous geothermometers. It also provides valuable guidance for accurately estimating target geothermal reservoir temperatures and advancing exploration in nonvolcanic geothermal systems.

The application of hyperspectral imagery coupled with deep learning shows vast promise in plant species discrimination. Reshaping one-dimensional (1D) leaf-level reflectance data (LLRD) into two-dimensional (2D) grayscale images as convolutional neural network (CNN) model input demonstrated marked effectiveness in plant species distinction. However, the impact of the image shape on CNN model performance remained unexplored. This study addressed this by reshaping data into fifteen distinct rectangular formats and creating nine CNN models to examine the effect of image structure. Results demonstrated that irrespective of CNN model structure, elongated narrow images yielded superior species identification results. The ‘l’-shaped images at 225 × 9 pixels outperformed other configurations based on 93.95% accuracy, 94.55% precision, and 0.94 F1 score. Furthermore, ‘l’-shaped hyperspectral images consistently produced high classification precision across species. The results suggest this image shape boosts robust predictive performance, paving the way for enhancing leaf trait estimation and proposing a practical solution for pixel-level categorization within hyperspectral imagery (HSIs).