Previous wildfire risk assessments have problems such as subjectivity of weight allocation and the linearization of statistical models, resulting in generally low robustness and low generalization ability of fire risk assessment models. Therefore, in this paper, we explored the potential of integration machine learning algorithms to build wildfire risk assessment models. Based on analyzing fire data’s spatial and temporal distribution, we selected 10 triggering factors of topography, meteorology, vegetation, and human activities, using frequency ratio (FR) to provide uniform data representation of triggering factors. Next, we used the Bayesian optimization (BO) algorithm to perform hyperparametric optimization solutions for various machine learning models: support vector machine (SVM), random forest (RF), and extreme gradient boosting (XGBoost). Finally, we constructed an integration machine learning algorithm to acquire a fire risk grading map and the importance evaluation corresponding to each triggering factor. For validation purposes, we selected Liangshan Prefecture in Sichuan Province as the specific study area and obtained MCD64A1 burned area product to extract the extent of burned areas in Liangshan Prefecture from 2011 to 2020. The accuracy, kappa coefficient, and area under curve (AUC) were then applied to assess the predictive power and consistency of the fire risk classification maps. The experimental analysis showed that among the three models, FR-BO-XGBoost had the best performance in wildfire risk assessment in the Liangshan region (AUC = 0.887), followed by FR-BO-RF (AUC = 0.876) and FR-BO-SVM (AUC = 0.820). The feature importance result indicated that the study area’s most significant effects on wildfires were precipitation, NDVI, land cover, and maximum temperature. The proposed method avoided the subjective weighting and model linearization problems. Compared with the previous methods, it automatically acquired the importance of the triggering factors to the wildfire, which had certain advantages in wildfire risk assessment, and was worthy of further promotion.

Being built on the reclamation area, Shanghai Pudong International Airport (SPIA) has been undergoing uneven subsidence since the beginning of its operation in 1999. In order to explore the evolution characteristics of ground deformation in the SPIA reclamation area and further provide assurance for the airport’s safe operation, 141 Sentinel-1A images from October 2016 to September 2021 were selected to acquire time-series ground deformation observations by the StaMPS PSI processing procedure. We subsequently built a ground deformation prediction model using the Long Short Term Memory (LSTM) neural network for the short-term prediction of the SPIA deformation severity area. On this basis, the spatial-temporal evolution trends of SPIA ground deformation in the reclamation area were revealed concerning the influence and mode of action of geological conditions and environmental factors. Finally, we proposed targeted recommendations and strategies for the comprehensive ground deformation prevention and control needs of SPIA. The results indicated that the SPIA exhibits overall subsidence in the eastern part, with the maximum deformation rate reaching −57.29 mm/a. Meanwhile, the central and western part has a local uplift with the maximum deformation rate reaching 32.76 mm/a. The proposed LSTM ground deformation prediction model demonstrated excellent robustness in the region of uneven deformation, and the prediction results were in high agreement with the StaMPS PSI monitoring results. The time-series observations and prediction results are expected to provide references for the expansion project of SPIA and help the research of ground deformation and prevention in related fields.

There are two problems with using global navigation satellite system-interferometric reflectometry (GNSS-IR) to retrieve the soil moisture content (SMC) from single-satellite data: the difference between the reflection regions, and the difficulty in circumventing the impact of seasonal vegetation growth on reflected microwave signals. This study presents a multivariate adaptive regression spline (MARS) SMC retrieval model based on integrated multi-satellite data on the impact of the vegetation moisture content (VMC). The normalized microwave reflection index (NMRI) calculated with the multipath effect is mapped to the normalized difference vegetation index (NDVI) to estimate and eliminate the impact of VMC. A MARS model for retrieving the SMC from multi-satellite data is established based on the phase shift. To examine its reliability, the MARS model was compared with a multiple linear regression (MLR) model, a backpropagation neural network (BPNN) model, and a support vector regression (SVR) model in terms of the retrieval accuracy with time-series observation data collected at a typical station. The MARS model proposed in this study effectively retrieved the SMC, with a correlation coefficient (R²) of 0.916 and a root-mean-square error (RMSE) of 0.021 cm³/cm³. The elimination of the vegetation impact led to 3.7%, 13.9%, 11.7%, and 16.6% increases in R² and 31.3%, 79.7%, 49.0%, and 90.5% decreases in the RMSE for the SMC retrieved by the MLR, BPNN, SVR, and MARS model, respectively. The results demonstrated the feasibility of correcting the vegetation changes based on the multipath effect and the reliability of the MARS model in retrieving the SMC.

In recent years, many cities in the Chinese loess plateau (especially in Shanxi province) have encountered ground subsidence problems due to the construction of underground projects and the exploitation of underground resources. With the completion of the world’s largest geotechnical project, called “mountain excavation and city construction,” in a collapsible loess area, the Yan’an city also appeared to have uneven ground subsidence. To obtain the spatial distribution characteristics and the time-series evolution trend of the subsidence, we selected Yan’an New District (YAND) as the specific study area and presented an improved time-series InSAR (TS-InSAR) method for experimental research. Based on 89 Sentinel-1A images collected between December 2017 to December 2020, we conducted comprehensive research and analysis on the spatial and temporal evolution of surface subsidence in YAND. The monitoring results showed that the YAND is relatively stable in general, with deformation rates mainly in the range of −10 to 10 mm/yr. However, three significant subsidence funnels existed in the fill area, with a maximum subsidence rate of 100 mm/yr. From 2017 to 2020, the subsidence funnels enlarged, and their subsidence rates accelerated. Further analysis proved that the main factors induced the severe ground subsidence in the study area, including the compressibility and collapsibility of loess, rapid urban construction, geological environment change, traffic circulation load, and dynamic change of groundwater. The experimental results indicated that the improved TS-InSAR method is adaptive to monitoring uneven subsidence of deep loess area. Moreover, related data and information would provide reference to the large-scale ground deformation monitoring and in similar loess areas.

Compared with previous snow depth monitoring methods, global navigation satellite system-interferometric reflectometry (GNSS-IR) technology has the advantage of obtaining continuous daily observation data, and has great application potential. However, since GNSS satellites are in motion, their position in the sky is constantly varying, and the Fresnel reflection areas about the receiver in different periods alter accordingly. As a result, the retrieving results obtained from different GNSS satellites, and data sets collected in different periods, fluctuate considerably, making the traditional single-satellite-based GNSS-IR retrieving method have limitations in accuracy and reliability. Therefore, this paper proposed a novel GNSS-IR signal-to-noise ratio (SNR) retrieving snow depth method for fusing the available GNSS-IR observations to obtain an accurate and reliable result. We established the retrieval model based on the backpropagation algorithm, which makes full use of the back propagation (BP) neural network’s self-learning and self-adaptive capability to exploit the degree of contribution of different satellites to the final results. Then, the SNR observations of the global positioning system (GPS) L1 carrier from the Plate Boundary Observation (PBO) site P351 were collected to experiment for validation purposes. For all available GPS L1 carrier data, the snow depth values retrieved for each satellite were first obtained by the existing single-satellite-based GNSS-IR retrieval method. Then, four groups of comparison results were acquired, based on the multiple linear regression model, random forest model, mean fusion model, and the proposed BP neural network model, respectively. Taking the snow depth in-situ data provided by snow telemetry (SNOTEL) as a reference, the root mean squared error (RMSE) and mean absolute error (MAE) of the proposed solution are 0.0297 m and 0.0219 m, respectively. Furthermore, the retrieving results are highly consistent with the measured data, and the correlation coefficient is 0.9407.