Research in the field of earthquake prediction has a long history, but the inadequacies of traditional approaches to the study of seismic threats have become increasingly evident. Remote sensing and Earth observation technology, an emerging method that can rapidly capture information concerning anomalies associated with seismic activity across a wide geographic area, has for some time been believed to be the key to overcoming the bottleneck in earthquake prediction studies. However, a multi-parametric method appears to be the most promising approach for increasing the reliability and precision of short-term seismic hazard forecasting, and thermal infrared (TIR) anomalies are important earthquake precursors. While several studies have investigated the correlation among TIR anomalies identified by the robust satellite techniques (RSTs) methodology and single earthquakes, few studies have extracted TIR anomalies over a long period within a large study area. Moreover, statistical analyses are required to determine whether TIR anomalies are precursors to earthquakes. In this paper, RST data analysis and the Robust Estimator of TIR Anomalies (RETIRA) index were used to extract the TIR anomalies from 2002 to 2018 in the Sichuan region using Moderate Resolution Imaging Spectroradiometer (MODIS) land surface temperature (LST) data, while the earthquake catalog was used to ascertain the correlation between TIR anomalies and earthquake occurrences. Most TIR anomalies corresponded to earthquakes, and statistical methods were used to verify the correlation between the extracted TIR anomalies and earthquakes. This is the first time that the ability to predict earthquakes has been evaluated based on the positive predictive value (PPV), false discovery rate (FDR), true-positive rate (TPR), and false-negative rate (FNR). The statistical results indicate that the prediction potential of RSTs with use of MODIS is limited with regard to the Sichuan region.

Object detection is a fundamental task in computer vision that involves accurately locating and classifying objects within images or video frames. In remote sensing, this task is particularly challenging due to the high resolution, multi-scale features, and diverse ground object characteristics inherent in satellite and UAV imagery. These challenges necessitate more advanced approaches for effective object detection in such environments. While deep learning methods have achieved remarkable success in remote sensing object detection, they typically rely on large amounts of labeled data. Acquiring sufficient labeled data, particularly for novel or rare objects, is both challenging and time-consuming in remote sensing scenarios, limiting the generalization capabilities of existing models. To address these challenges, few-shot learning (FSL) has emerged as a promising approach, aiming to enable models to learn new classes from limited labeled examples. Building on this concept, few-shot object detection (FSOD) specifically targets object detection challenges in data-limited conditions. However, the generalization capability of FSOD models, particularly in remote sensing, is often constrained by the complex and diverse characteristics of the objects present in such environments. In this paper, we propose the Generalization-Enhanced Few-Shot Object Detection (GE-FSOD) model to improve the generalization capability in remote sensing FSOD tasks. Our model introduces three key innovations: the Cross-Level Fusion Pyramid Attention Network (CFPAN) for enhanced multi-scale feature representation, the Multi-Stage Refinement Region Proposal Network (MRRPN) for more accurate region proposals, and the Generalized Classification Loss (GCL) for improved classification performance in few-shot scenarios. GE-FSOD demonstrates superior robustness and accuracy in remote sensing FSOD tasks through these enhancements. Extensive experiments on the DIOR and NWPU VHR-10 datasets show that our model achieves state-of-the-art performance, significantly advancing the field of few-shot object detection in remote sensing. The source code is available at (https://github.com/leenamx/GE-FSOD).

On 3 August 2014, an Mw 6.2 earthquake occurred in Ludian, Yunnan Province, China (27.245° N 103.427° E). This damaging earthquake caused approximately 400 fatalities, 1800 injuries, and the destruction of at least 12,000 houses. Using air temperature data of the National Center for Environmental Prediction (NCEP) and the tidal force fluctuant analysis (TFFA) method, we derive the temperature variations in multiple air layers between before and after the Ludian earthquake. In the spatial range of 30° × 30° (12°–42° N, 88°–118° E) of China, a thermal anomaly appeared only on or near the epicenter before earthquake, and air was heated from the land, then uplifted by a heat flux, and then cooled and dissipated upon rising. With the approaching earthquake, the duration and range of the thermal anomaly during each tidal cycle was found to increase, and the amplitude of the thermal anomaly varied with the tidal force potential: air temperature was found to rise during the negative phase of the tidal force potential, to reach peak at its trough, and to attenuate when the tidal force potential was rising again. A significance test supports the hypothesis that the thermal anomalies are physically related to Ludian earthquakes rather than being coincidences. Based on these results, we argue that the change of air temperature could reflect the stress changes modulated under the tidal force. Moreover, unlike the thermal infrared remote sensing data, the air temperature data provided by NCEP are not affected by clouds, so it has a clear advantage for monitoring the pre-earthquake temperature variation in cloudy areas.

SUMMARY: We use the spatial map of the logarithm of past estimated released earthquake energies as input of fully convolutional networks (FCN) to forecast future earthquakes. This model is applied to California and compared with an elaborated version of the epidemic type aftershock sequence (ETAS) model. Our long-term earthquake forecast simulations show that the FCN model is close to the ETAS model in forecasting earthquakes with $M \ge 3.0,\,\,4.0,\,\,{\rm{and\,\,}}5.0$ according to the Molchan diagram. Moreover, training and implementing the FCN model is 2000–4000 times faster than calibrating the ETAS model and generating its probabilistic forecasts. The FCN model is straightforward in terms of its neural network structure and feature engineering. It does not require extensive knowledge of statistical seismology or the analysis of earthquake catalogue completeness. Using the earthquake catalogue with $M \ge 0$ as FCN input can enhance the model's performance in some time–magnitude forecasting windows.

Using the model of the additive tectonic stress from celestial tide-generating force, we studied the relationship between the seismogenic structure and celestial tide-generating stress in the M7.8 Turkey earthquake on 6 February 2023. We analyzed the daily continuous variation characteristics of OLR before and after the Turkey earthquake and discussed the correlation characteristics of tidal stress, OLR, and the earthquake. The results showed that the observed OLR anomaly according to the tidal stress variation cycle “C” (1–15 February) presented a phase change in time, which was synchronized with a continuous trough-to-peak change in the additional tectonic main pressure stress. The spatial distribution of OLR anomalies was mainly concentrated in the southwest section of the East Anatolian Fault Zone, which indicates that seismic tectonic movements were the main causes of OLR anomaly variation during this earthquake. An OLR anomaly change was related to this M7.8 “Swarm Type” of earthquake in Turkey. Impending earthquake OLR anomalies represent that the stress of the seismogenic structure in the seismogenic region has entered a critical state, which can provide stress monitoring and a seismogenic region indication for earthquakes induced by tidal force. The change cycle of the celestial tide-generating force provides a time indication for the identification of seismic thermal anomalies, and it indicates that the combination of the additional tectonic stress of the tidal force and the change of OLR anomaly has value for the research on the short-impending earthquake precursor.