Liaoning Meteorological Bureau

Ultraviolet absorption spectroscopy, recognized for its rapid, nondestructive, and noninvasive nature, has been widely employed in fields such as environmental monitoring and chemical analysis. However, the accuracy of quantitative spectral analysis remains challenged by data distortions arising from various factors in the instrumental transmission process, including particle scattering, spectral drift, excessive denoising, and fitting errors. To overcome these issues, we propose a specific regression approach that integrates variational mode decomposition (VMD) with a convolutional neural network (CNN) for absorption spectra, termed VMD-CNN. The method first decomposes the spectrum into a series of intrinsic mode functions, from which mode components associated with gas absorption characteristics are extracted. A one-dimensional CNN is then trained on a large dataset of spectra with known concentrations to establish a quantitative inversion model. Experiments conducted on NO and SO2 absorption spectra within the 200 to 226 nm range demonstrate the effectiveness of the proposed approach. Comparative analyses with conventional techniques—including nonlinear least squares, partial least squares, and fast Fourier transform amplitude methods—before and after noise reduction and background elimination show that traditional methods heavily depend on data quality and may produce distorted results under low-concentration conditions due to over-optimization. By contrast, the proposed VMD-CNN model, which requires no preprocessing, consistently achieves the lowest error metrics across broad concentration ranges, specifically yielding root mean square error as low as 12.0 for NO and 2.0 for SO2, with mean absolute percentage error maintained 1.8% and 2.35%, respectively.

Crustal movement and deformation monitoring are important methods for reflecting changes in crustal stress. Crustal deformation data can be used to accurately describe the movement and deformation characteristics of active blocks and their boundary zones, providing effective data constraints for earthquake prediction and scientific research. Crustal deformation monitoring mainly includes crustal movement monitoring (such as Global Navigation Satellite System (GNSS) observations), surface strain monitoring (such as borehole strain observations), and surface tilt monitoring (such as vertical pendulum tilt observations). The high-precision and high temporal resolution data generated are widely used in the study of slow earthquakes, volcanic activity, and earthquake precursors. Given the close relationship between geophysical instruments and observation environments, crustal deformation monitoring instruments can not only record structural signals, but also interference signals caused by changes in surrounding loads. This article is based on the analysis displacement solution caused by the point load model, and derives formulas for calculating the surrounding tilt field and strain field, providing a theoretical basis for the quantitative calculation of the influence of surrounding loads in crustal deformation monitoring. In addition, this article also proposes a method for calculating the strain effects of two-dimensional and three-dimensional irregular shaped loads. Finally, based on the four component borehole strain observation data from Guza borehole strainmeter and the observation data of the surrounding river water level, the applicability of this analytical solution in quantitatively calculating the degree of influence of irregular load models in the surrounding area was verified under the conditions of setting the surrounding medium parameters (elastic modulus and Poisson's ratio). The results indicate that: (a) for the two-dimensional irregular shaped load model problem, vector superposition calculation can be performed after load scattering; (b) For the problem of three-dimensional irregular shaped load models, different weights can be assigned to scattering points after load scattering, and the two-dimensional irregular shaped load method can be used for calculation. The convergence process during vector superposition proves the correctness and feasibility of this method. This study provides a research foundation for the quantitative analysis of the influence of surrounding load interference in crustal deformation monitoring; (c) There is a high possibility that the data disturbance information of the four component drilling strain observation data at Guzan Station in summer is affected by the disturbance of the water level data of nearby rivers. This research work can quantitatively explain the degree of influence of load type interference factors on high-precision geophysical observation data, providing a quantitative interpretation scheme for the extraction of earthquake precursor anomalies.Fund support: Ningxia Natrual Science Foundation Project (2024AAC03436); National Natural Science Foundation of China (41704062); National Key Research and Development Program of China (2021YFC3000705-06).Figure 1 Example of the influence of irregularly shaped loads in the near field. In the figure: (a) represents the positional relationship between the Guzan borehole strainmeter and the Dadu River; (b) Representing the comparison of the changes in actual observed data with the results of irregular load model calculations.

Cognitive impairment is a debilitating comorbidity in temporal lobe epilepsy (TLE), linked to suppressed hippocampal neurogenesis in the chronic phase. Building on evidence that links SDF-1 to epilepsy, emerging research, including our own preliminary data, suggests that its cleaved product, SDF-1(5-67), may play a critical role in this process. To investigate the therapeutic potential of SDF-1(5-67) neutralizing antibody (NAb) in a kainic acid (KA)-induced temporal lobe epilepsy (TLE) model. TLE was induced in male Wistar rats by intraventricular KA injection. Animals were divided into acute (8 days) and chronic (72 days) phase groups, including acute phase control (SHAM-A), acute phase epilepsy model (EP-A), SDF-1(5-67) NAb-treated acute phase epilepsy (S + EP-A), chronic phase control (SHAM-C), chronic phase epilepsy model (EP-C), and SDF-1(5-67) NAb-treated chronic phase epilepsy groups (S + EP-C). NAb treatment was administered via osmotic pumps. Neurogenesis (DCX + cells), neuronal loss (Nissl staining), MFS (Timm staining), cognition (Morris water maze), and SDF-1(5-67) expression (western blot) were assessed. Data were analyzed by one-way ANOVA with Tukey's post hoc test, or repeated-measures ANOVA for water maze latency, presented as mean ± SEM. During the acute phase, increased dentate gyrus (DG) neurogenesis in EP-A vs. SHAM-A (210.43 ± 12.16 vs. 94.25 ± 7.76 cells/mm², P < 0.05) was unaffected by NAb (195.84 ± 9.71, P > 0.05). In the chronic phase, EP-C showed suppressed neurogenesis (51.63 ± 7.36 vs. SHAM-C 85.12 ± 5.51, P < 0.01), which was rescued by NAb (77.56 ± 9.88, P < 0.01). CA3 neuronal loss was attenuated by NAb in EP-C (174.33 ± 13.06 vs. EP-C 127.47 ± 10.22, P < 0.05). MFS was increased in both phases and was unaffected by NAb. NAb improved cognitive deficits in EP-C (escape latency: 28.14 ± 1.6s vs. EP-C 35.80 ± 2.41s, P < 0.05). NAb suppressed SDF-1(5-67) expression in both phases (P < 0.05). SDF-1(5-67) NAb promoted neurogenesis and reduced neurodegeneration in chronic TLE, and improved cognitive function without affecting MFS or seizure duration. These findings highlight its potential for postepileptic cognitive rehabilitation.

Understanding both the spatial and temporal variations of atmospheric water vapor is crucial for forecasting regional weather and comprehending the global climate system. The Advanced Geostationary Radiation Imager (AGRI) on-board China's first next-generation geostationary meteorological satellite, FY-4A, provides operational high spatial-temporal resolution Total Precipitable Water (TPW) products. This study evaluates FY-4A TPW retrievals by comparing them with ground-based Global Navigation Satellite System (GNSS) measurements of TPW from over 1,100 stations across China, covering the period from June 2023 to March 2024. The results demonstrate strong agreement between the two datasets, yielding a correlation coefficient of 0.98 and a standard deviation of difference (STD) of 2.87 mm. FY-4A TPW is slightly lower than GNSS TPW, with a bias of -0.49 mm; this negative bias generally exhibits larger magnitudes for higher TPW values (usually observed in humid areas, during wet months, or in the afternoon hours). Furthermore, we investigate the dependence of evaluation results on surface type, clear sky fraction, NDVI, and satellite zenith and azimuth angles. Our findings indicate that 36% of the variance in the STD can be attributed to station attributes, including clear sky fraction, NDVI, and mean TPW. In contrast, the percentage of explained variance for the correlation coefficient and bias is lower, at 23% and 12%, respectively, with the satellite azimuth angle identified as an important factor. Further research is necessary to determine whether this pattern in TPW retrieval errors originates from background inputs (forecast temperature and water vapor profiles), or from the brightness temperature measurements of the FY-4A AGRI.