Spatiotemporal fusion is considered a feasible and cost-effective way to solve the trade-off between the spatial and temporal resolution of satellite sensors. Recently proposed learning-based spatiotemporal fusion methods can address the prediction of both phenological and land-cover change. In this paper, we propose a novel deep learning-based spatiotemporal data fusion method that uses a two-stream convolutional neural network. The method combines both forward and backward prediction to generate a target fine image, where temporal change-based and a spatial information-based mapping are simultaneously formed, addressing the prediction of both phenological and land-cover changes with better generalization ability and robustness. Comparative experimental results for the test datasets with phenological and land-cover changes verified the effectiveness of our method. Compared to existing learning-based spatiotemporal fusion methods, our method is more effective in predicting phenological change and directly reconstructing the prediction with complete spatial details without the need for auxiliary modulation.
Seismically active fault zones are complex natural systems and they exhibit multifractal correlation between earthquakes in space and time. In this paper, the seismicity of the Taiwan seismic region was studied through the multifractal characteristics of the spatial-temporal distribution of earthquakes from 1st January 1995 to 1st January 2019. We quantified the multifractal characteristics of Taiwan at different scales and defined them as ΔD values. Furthermore, we studied the relationship between the ΔD and b values, which signifies the average size distribution of those earthquakes. The results are as follows. (1) The temporal multifractal curve changes substantially before and after the strong earthquakes. (2) The maximum ΔD value of the seismic region in Taiwan occurs at depths of 0~9 km, indicating that geological structures and focal mechanisms is the most complex at these depths compared with other depths. (3) ΔD values for different regions range from 0.2~1.5, and b values range from 0.65~1.3, with a significant positive correlation between them (ΔD = 1.5 × b − 0.68). For this purpose, a statistical relationship is developed between b and ΔD values, and regional and temporal changes of these parameters are analyzed in order to reveal the potential of future earthquakes in the study region.
Changes in the stress state of faults and their surroundings is a highly plausible mechanism explaining earthquake interaction. These stress changes can impact the seismicity rate and the size distribution of earthquakes. However, the effect of large earthquakes on the earthquake size distribution along the Longmenshan fault has not been quantified. We evaluated the levels of the b value for the stable state before and after the large earthquakes on 12 May 2008 (Wenchuan, MS 8.0) and 20 April 2013 (Lushan, MS 7.0) along the Longmenshan fault. We found that after the mainshocks, the size distribution of the subsequent earthquakes shifted toward relatively larger events in the Wenchuan aftershock zone (b value decreased from 1.21 to 0.84), and generally remained invariable in the Lushan aftershock zone (b value remained at 0.76). The time required for the b value to return to stable states after both mainshocks was entirely consistent with the time needed by the aftershock depth images to stop visibly changing. The result of the temporal variation of b values shows decreasing trends for the b value before both large earthquakes. Our results are available for assessing the potential seismic risk of the Longmenshan fault as a reference.
The geodynamic mechanism is the research focus and core issue of plate motions and plate tectonics. Analyzing the time series of earthquakes may help us understand the relationship between two plate boundaries and further explore movement mechanisms. Therefore, this paper uses earthquake event data and the Granger causality test method to quantitatively analyze the interaction and energy transfer relationship of plate boundaries from the viewpoint of statistics. The paper aims to explore the relationship between the pull effect and the push effect of plate motion and to provide knowledge to explore seismic energy transfer relationships, and even to predict earthquakes: (1) The directions of the global plate motion field are opposite to the directions of Granger causality between plate boundaries of the Pacific, Nazca, African, Australian, Eurasian, and Philippine plates. (2) The slab-pull force (not limited to the subduction force of the ocean plates) provides a main driving force for plate motions in the Pacific plate, Nazca plate, African plate, Australian plate, Eurasian plate, and Philippine sea plate. (3) The causality relationship and optimal lag length of energy release between plate boundaries may provide another view to forecasting earthquakes.