ZWZhiwei Wang

Zhiwei Wang

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Zhiwei Wang has a total of 9 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Coupled multi-factor urban subsidence prediction based on small baseline subset interferometric SAR and transformer model: a case study of the Jinan Start-up Area, China

    by Zhiwei Wang, Fangfang Li, Xiang Wang, Ying Cui, Haolin Fu, Wenzhuang Ding, Wanqiu Li, Aihui Jiang

    Urban land subsidence has emerged as a critical issue in China, driven by rapid urbanization and extensive infrastructure development, posing serious threats to public safety and economic sustainability. However, most existing subsidence prediction models rely primarily on one-dimensional time-series deformation data, neglecting the combined influence of multiple environmental factors. To address this limitation, we propose an optimized transformer-based prediction model capable of capturing long-range temporal dependencies and complex nonlinear relationships among multi-source variables. The Jinan Start-up Area, a region facing high subsidence risk due to accelerated urban expansion, was selected as the study area. High-resolution time-series deformation data were obtained using a small baseline subset interferometric synthetic aperture radar technique, and auxiliary datasets including land cover classification, road network density, and precipitation were integrated to construct a comprehensive multi-feature dataset. Experimental results indicate that subsidence exhibits distinct spatial heterogeneity across land cover types, with dense road networks amplifying local instability and precipitation-driven recharge alleviating deformation. The proposed model achieves superior predictive accuracy, outperforming recurrent neural network, gated recurrent unit, and long short-term memory by 20.6%, 1.4%, and 7.9%, respectively. These results demonstrate the advantages of the model for addressing urban land subsidence forecasting.

  • Open Access

    BTFC-SBAS: A Boltzmann Time Function Constrained SBAS-InSAR Method for Nonlinear Deformation Estimation in Mining Areas

    by Zhiwei Wang, Wenzhuang Ding, Haolin Fu, Tengfei Gao, Ying Cui, Fangfang Li, Wanqiu Li

    Subsidence monitoring in mining areas is crucial for environmental protection in mining regions. The small baseline subset interferometric synthetic aperture radar (SBAS-InSAR) technique has proven highly effective for monitoring ground subsidence. However, solutions based on the SBAS-InSAR deformation model typically assume linear deformation, which limits their ability to capture the nonlinear, S-shaped subsidence patterns commonly observed in mining areas. A method referred to as Boltzmann time function-constrained SBAS-InSAR (BTFC-SBAS) is proposed to effectively address this limitation by introducing the Boltzmann function as a constraint into the SBAS-InSAR deformation model. This method incorporates a nonlinear functional constraint that aligns with the deformation characteristics of mining areas. Based on the small-baseline combination strategy, the BTFC-SBAS method establishes a mapping relationship between the interferometric phase of individual ground points and nonlinear subsidence by incorporating the Boltzmann function in place of the traditional linear deformation assumption. The model parameters are estimated on a per-pixel basis using a genetic algorithm, enabling accurate inversion of the nonlinear subsidence process in mining areas. Both simulated and real-data experiments based on the Mengkeqing Mine in Ordos City are conducted to verify the feasibility of the proposed BTFC-SBAS method. In the simulation experiment, six subsidence curves with different shapes are designed, and their inversion results obtained using the BTFC-SBAS, least squares (LS), and singular value decomposition (SVD) methods are compared. The BTFC-SBAS method outperforms both the LS and SVD methods under all scenarios, achieving an average RMSE of 9.833 mm, lower than those of the SVD method (17.324 mm) and the LS method (44.158 mm). In the real-data experiment, the monitoring results derived from the BTFC-SBAS and traditional SBAS methods show highly consistent spatial patterns. With levelling data as a reference, comparative analysis at four representative validation points reveals that the RMSE of the BTFC-SBAS method decreases by 25.6%, 19.9%, 30.4%, and 29.7% relative to that of the SBAS method. These findings confirm that the BTFC-SBAS method provides higher inversion accuracy and produces time-series subsidence curves that more closely match the actual measurements. Overall, the proposed method offers an effective approach for dynamic monitoring and prediction of nonlinear surface subsidence in mining areas.