Remote Sensing-based Synchronized Seismic Precursor Detection using SARIMAX model Forecasting: A Case Study of the Mw 7.4 Taiwan Earthquake of April 2024

Syed Faizan Haider¹ Munawar Shah² Rasim Shahzad³

¹ Global Navigation Satellite System Research Lab, National Center of GIS and Space Applications, Institute of Space Technology, Islamabad 44000, Pakistan; Faizanhaider92110@gmail.com

Abstract

Earthquakes are among the most destructive natural phenomena, demanding innovative approaches to prediction and monitoring. The Mw 7.4 Taiwan earthquake exemplifies the need for robust methodologies to detect and analyze seismic precursors in vulnerable regions. This study utilizes Remote Sensing (RS) and Global Navigation Satellite System (GNSS) technologies to investigate possible seismic precursors, including Total Electron Content (TEC), Sea Surface Temperature (SST), Land Surface Temperature (LST), Air Pressure (AP), Relative Humidity (RH), Outgoing Longwave Radiation (OLR), and Air Temperature (AT). Employing statistical methods such as Standard Deviation (STDEV) and the Seasonal AutoRegressive Integrated Moving Average with Exogenous Variables (SARIMAX) model, this study identified synchronized anomalies 5–6 days before the earthquake. Additionally, geomagnetic disturbances were detected 9 days prior, coinciding with an active geomagnetic storm (Kp > 5; Dst < -75 nT; ap > 70 nT). A confutation analysis of atmospheric parameters from the same region and time period over the previous five years further validated the findings, enhancing confidence in the results. By combining statistical, spatial, and forecasting methodologies, this research advances the understanding of seismic precursors and underscores the potential of integrating multi-parameter analysis for improved earthquake prediction and disaster management strategies.

Keywords: earthquake precursors; geomagnetic storm; GNSS TEC; remote sensing; SARIMAX

Coastal heritage assets are essential components of a society’s history and must be transferred to the next generation sustainably. However, they are significantly prone to natural and anthropogenic hazards. Monitoring coastal assets is a complex phenomenon as the attributes are interdependent. Expert observations and on-site measurements have been practised to monitor these assets, but it takes time to cover a larger area over a longer time. Remote sensing observation could be used to assess a large area with regular observation. This study examines the effects of climate change on coastal heritage assets in terms of temperature and moisture variations. Such long-term variations could exacerbate the condition of coastal heritage assets and pose a risk. The study employs a multidisciplinary approach that combines satellite observations, publicly available in situ measurements and sustainable development practices. Using remote sensing, we examine satellite images to extract land surface temperature, monitor changes in coastal environments and assess their effects on coastal historic sites. We identify areas at risk based on temperature and moisture anomalies that need attention from asset owners and decision-makers. These findings will contribute to actions for the protection of coastal heritage and provide input for a framework for sustainable heritage conservation.

Urban areas, which accommodate more than half of the global population, are the major contributors to climate change and are significantly affected by it. They are the growth engines responsible for generating a large share of global emissions. However, making these cities and human settlements inclusive, safe, resilient, and sustainable is a critical goal of the Sustainable Development Goals (SDGs) adopted by the United Nations in 2015. Moreover, rapid and unplanned urban expansion exacerbates various environmental challenges, such as heat stress, habitat loss, air pollution, water scarcity, and reduced green cover in urban areas. To address these issues, city planners and policymakers focus on sustainable urban development and the effective management of urban spaces. Vegetation is a crucial component of the urban ecosystem and plays a vital role in mitigating climate change impacts, reducing urban heat island effects, improving air and water quality, fostering urban biodiversity, and promoting human well-being. Green Infrastructure has proven to be one of the most successful techniques for carbon sequestration, and to ensure its continued success and optimal performance, regular monitoring is essential. Here, we have used multiple satellite-derived products for urban vegetation mapping, health monitoring, and trend analysis. These indicators were generated at 10/20 m spatial resolution from Sentinel-2 data using Google Earth Engine from 2017 to 2024 for Ealing, a borough of London in the UK. We created several band combinations and covariates for vegetation health and anomaly detection at the borough level. The results show that approximately the entire borough has experienced an ascending trend in tree cover over the selected time duration. Our study underscores that the indices derived from medium-to-high resolution satellite data can be utilised in urban green infrastructure monitoring with reasonable accuracy and calls for evidence-based strategies to achieve economic growth, social inclusion, and urban resilience.

Green infrastructure in urban settings is an essential component of city sustainability. New developments in the built environment pose a huge threat to the reduction in and health of vegetation. Monitoring urban green spaces is important to understand the extent of the change in the urban micro-climate and its impact on public health and well-being. Traditional methods, such as in situ measurements and expert observations, are often constrained by spatial and temporal limitations. The dynamic changes in urban settings need efficient planning and maintenance of green spaces. Satellite observations have become a fundamental tool to provide city-scale coverage with sound temporal coverage. Leveraging the large volume of publicly available data, advanced machine learning models could enhance our understanding and analysis of the urban environment. We explore the potential of Sentinel-2 vegetation indices such as the Normalized Difference Vegetation Index (NDVI) or the Normalized Difference Water Index (NDWI) to classify and extract useful features from urban landscapes. By utilising state-of-the-art machine learning techniques, we aim to develop a robust and scalable framework for urban environment classification. The proposed models will facilitate monitoring changes in green spaces across diverse urban contexts, enabling timely and informed decision-making to support sustainable urban development. In addition, the integration of vegetation indices contributes to actionable insights for promoting eco-friendly and sustainable urban planning while supporting the development of resilient urban ecosystems, making it a valuable tool for decision-makers and policy developers.

Quantifying and monitoring carbon sequestration is an important tool for developing global policies, helping with the emerging carbon credit market, and understanding climate change. Above-ground biomass (AGB) is a commonly used indicator that describes the amount of carbon that is stored above ground. The estimation of AGB can be done using direct or indirect methods. Direct methods involve the destruction of the trees unlike indirect methods. Allometric equations constitute an indirect method that is widely used and does not involve the destruction of trees to estimate AGB. However, it involves collecting data from forest inventories, which is time consuming and expensive. A cheaper and faster alternative provided by technological development is remote sensing estimation. In this alternative, data collected using the satellite (remote sensing data) are used together with field data, which can lead to more accurate AGB estimates.

In recent years, several machine learning models have been used to predict AGB, mainly the Random Forest (RF) algorithm. However, RF models present limitations in their performance when dealing with spatial data, as is often the case in AGB, by ignoring the spatial autocorrelation, leading to one of the limitations in their predictive performance. Thus, within this context, we use a hybrid method to estimate the target AGB in Sierra de la Culebra, Spain, that combines the RF algorithm and a Bayesian geostatistical model and uses as features variables from remote sensing data from the GEOSAT-2 satellite, such as reflectance bands, vegetation indices, and texture variables. In addition, in this work, we compare the predictive performance of this hybrid model with the predictive results obtained by using solely the RF model and the Bayesian geostatistical model.

Hyperspectral imaging applied to fruits and vegetables has experienced significant growth in recent years. This technique involves collecting reflectance spectra containing valuable information for various applications, such as monitoring the nutritional and taste qualities of fruits and vegetables, detecting rotten samples, and classifying varieties. The potential impact of this technique is high; in fact, monitoring the spectral signatures of objects in fields or on trees allows farmers to assess pre-harvest gustatory and nutritional qualities by monitoring the supply of water, nutrients, and treatments.

Many physics-based models used in the literature have demonstrated potential for studying such objects, including Prospect, Farrell, MARMIT, and MPBOM. Among these, we are interested in two models: Prospect for the characterization of leaves and Farrell for the characterization of fruits and vegetables. In this article, we propose an initial theoretical approach based on the identification of input parameters for the spectra derived from these two models, complemented by an experimental validation using intermediate objects such as leeks and apple skins.

Parcel delineation represents a fundamental challenge in agricultural management. It plays a pivotal role in multiple domains, including agricultural policy implementation, crop yield estimation, and environmental monitoring. Modern Remote Sensing Images (RSIs) offer a revolutionary approach to parcel boundary identification, providing high-resolution, comprehensive spatial data that can capture intricate landscape details across vast agricultural regions. Advanced image processing techniques, particularly deep learning and machine learning algorithms, have dramatically enhanced the capability to automatically extract and define field boundaries with remarkable accuracy. In addition, state-of-the-art (SOTA) methods leverage sophisticated classification and segmentation techniques to precisely define agricultural boundaries from satellite imagery, addressing the complex challenge of accurately delineating Cultivated Land Parcels (CLPs). Traditional segmentation approaches, including edge-based, region-based, and hybrid methods, have long been fundamental in agricultural field mapping, each offering distinct methodological strategies for identifying and demarcating land parcels. The rapid advancement of deep learning techniques has revolutionized RSI analysis, particularly in complex agricultural domain applications. In this study, we introduce an innovative two-stage hybrid method for agricultural parcel delineation, leveraging a novel FracTAL UNet Attention Residual Refinement Module (FracTAL UNet ARRM). Our proposed method represents a significant leap forward in the SOTA. Comprehensive performance evaluation was conducted using multiple metrics, including accuracy, F1-score, mean intersection over union (mIoU), and Boundary Displacement Error (BDE). The FracTAL UNet ARRM method we developed outperforms SOTA methods, achieving the best performances in terms of accuracy, F1-score, mIoU, and BDE, respectively.

Remote sensing has been instrumental in advancing forest conservation and guiding efforts to sustain ecosystem services. Imaging spectroscopy—with its ability to capture canopy traits, accurately classify vegetation species, and estimate biodiversity—has enhanced these efforts by providing detailed insights into forest composition and health. This study presents a case study of imaging spectroscopy applications in native forest conservation.

Hawaiʻi has long served as a model system for ecological studies and a testing ground for advanced remote sensing technologies. With the emergence of Rapid ʻŌhiʻa Death (ROD), a novel disease complex first observed around 2010, Hawaiʻi offers a crucial opportunity to integrate remote sensing into the management of native forest ecosystems. ROD has caused millions of ʻōhiʻa lehua (Metrosideros polymorpha) mortalities across Hawaiʻi Island and Kauaʻi. Accounting for approximately 80% of native Hawaiian forests, ʻōhiʻa forests are the ecological foundation for diverse endemic avifauna and understory communities, and provide critical ecosystem services such as watershed protection.

We used Global Airborne Observatory airborne imaging spectroscopy and coaligned LiDAR data to classify ʻōhiʻa in 28,000 km² of area at a 2m x 2m spatial resolution. The ability to map species distributions has far-reaching conservation implications, yet few studies have used imaging spectroscopy, which yields higher classification accuracies than multispectral datasets, to classify tree species across ecosystems. As ʻōhiʻa is characterized by a high degree of intraspecific variation, we explored how its spectral variation challenged species classifications, especially across biomes, and whether this species follows the leaf economic spectrum.

Building on these results, we are collaborating with local management agencies to integrate this dataset into ROD and native forest management strategies. These data will inform efforts to control non-native biological agents spreading ROD and support scalable conservation solutions to safeguard Hawaiʻi’s forests and their vital ecosystem services.

Fuel and fire behavior modeling is essential for wildfire prevention and control. Currently, advanced 3D physics-based fire behavior models, such as Wildland Fire Dynamics Simulator (WFDS), are able to represent heterogeneous fuels and simulate fire behavior processes with much greater detail than conventional semi-empirical models. However, they require accurate information about the locations and dimensions of individual trees, species compositions, spatial distributions of understory fuels, as well as3D distributions of fuel mass and bulk density at the voxel level and finer spatial scales. Point clouds derived from airborne, terrestrial, and mobile laser scanners are uniquely suited for quantifying the three-dimensional structure of canopy and understory vegetation, but UAV-based digital aerial photogrammetric (DAP) point clouds have the advantage of allowing for a higher frequency of data acquisition and the integration of structural and spectral data.

Working in a Mediterranean ecosystem study area with four dominant shrub species and Pinus halepensis trees, we developed a methodology based on the use of geometric and spectral features from UAV-DAP point clouds for (i) species segmentation and classification using machine learning algorithms, (ii) the generation of biomass prediction models and estimation of bulk density at the individual plant level, and (iii) the creation of 3D fuel scenarios andwildfire behavior modeling with WFDS. Field measurements and allometric equations were used for the evaluation of classification and prediction models. Fire behavior variables, such as rate of spread, heat release rate, and mass loss rate, were monitored and assessed as outputs from WFDS. The overall species classification accuracy was 80.3%, and the biomass regression R² values obtained by cross-validation were 0.77 for Pinus halepensis, 0.83 for Anthyllis cytisoides, 0.69 for Quercus coccifera, 0.60 for Genista scorpius, and 0.54 for Salvia rosmarinus. These results are encouraging for further improvement based on the integration of multi- and hyper-spectral sensors onboard UAVs, and the characterization of fuels for fire behavior modeling.

Urban areas are increasingly impacted by climate change, including air quality degradation and elevated temperatures. Building on previous research that assessed the effectiveness of Nature-Based Solutions (NBSs) in mitigating surface temperatures, this study expands the analysis to investigate the impact of NBSs on urban air quality. By integrating remote sensing data from the Sentinel-5P (TROPOMI) satellite, with daily revisits, and in situ measurements of atmospheric pollutants (NO₂, PM₁₀, and PM₂.₅), this research analyzes spatiotemporal trends in air quality and their relationship with Land Surface Temperature (LST) and the implementation of NBSs, having as a case study the city of Guimarães, Portugal.

Using machine learning models and multitemporal datasets, this study evaluates changes in air quality and LST between 2017 and 2023, with projections for 2028. The methodology combines satellite data, such as nitrogen dioxide (NO₂) concentrations, with daily and annual in situ measurements from local monitoring stations. To enhance both temporal and spatial analysis, LST data from two sources will be utilized: Landsat 8, offering high spatial resolution (30 meters) with a 16-day revisit cycle, and Sentinel-3, providing frequent revisits (1 to 2 days) but lower spatial resolution (1 km x 1 km). This integration reconciles Landsat 8’s spatial precision with Sentinel-3’s temporal frequency, enabling a more comprehensive analysis.

Preliminary findings highlight the role of NBSs, such as green roofs and urban gardens, in reducing surface temperatures. This study aims to deepen these observations by quantifying the potential of NBSs to mitigate urban air pollution and identifying critical hotspots where targeted interventions are most needed.

This research provides a replicable framework for evaluating the impacts of NBSs on air quality and urban temperature, offering practical findings for sustainable urban planning. Future directions include testing the methodology in diverse urban contexts and exploring the scalability and adaptability of NBS interventions.