Multi-angle remote sensing can either be regarded as an added source of uncertainty for variable retrieval, or as a source of additional information, which enhances variable retrieval compared to traditional single-angle observation. However, the magnitude of these angular and band effects for forest structure parameters is difficult to quantify. We used the Discrete Anisotropic Radiative Transfer (DART) model and the Zelig model to simulate the forest canopy Bidirectional Reflectance Distribution Factor (BRDF) in order to build a look-up table, and eight vegetation indices were used to assess the relationship between BRDF and forest biomass in order to find the sensitive angles and bands. Further, the European Space Agency (ESA) mission, Compact High Resolution Imaging Spectrometer onboard the Project for On-board Autonomy (CHRIS-PROBA) and field sample measurements, were selected to test the angular and band effects on forest biomass retrieval. The results showed that the off-nadir vegetation indices could predict the forest biomass more accurately than the nadir. Additionally, we found that the viewing angle effect is more important, but the band effect could not be ignored, and the sensitive angles for extracting forest biomass are greater viewing angles, especially around the hot and dark spot directions. This work highlighted the combination of angles and bands, and found a new index based on the traditional vegetation index, Atmospherically Resistant Vegetation Index (ARVI), which is calculated by combining sensitive angles and sensitive bands, such as blue band 490 nm/−55°, green band 530 nm/55°, and the red band 697 nm/55°, and the new index was tested to improve the accuracy of forest biomass retrieval. This is a step forward in multi-angle remote sensing applications for mining the hidden relationship between BRDF and forest structure information, in order to increase the utilization efficiency of remote sensing data.

Forest fires pose severe ecological, climatic, and socio-economic threats, destroying habitats and emitting greenhouse gases. Early and timely warning is particularly challenging because fires often originate from small-scale, low-temperature ignition sources. Traditional monitoring approaches primarily rely on single-source satellite imagery and empirical threshold algorithms, and most forest fire monitoring tasks remain human-driven. Existing frameworks have yet to effectively integrate multiple data sources and detection algorithms, lacking the capability to provide continuous, automated, and generalizable fire monitoring across diverse fire scenarios. To address these challenges, this study first improves multiple monitoring algorithms for forest fire detection, including a statistically enhanced automatic thresholding method; data augmentation to expand the U-Net deep learning dataset; and the application of a freeze–unfreeze transfer learning strategy to the U-Net transfer model. Multiple algorithms are systematically evaluated across varying fire scales, showing that the improved automatic threshold method achieves the best performance on GF-4 imagery with an F-score of 0.915 (95% CI: 0.8725–0.9524), while the U-Net deep learning algorithm yields the highest F-score of 0.921 (95% CI: 0.8537–0.9739) on Landsat 8 imagery. All methods demonstrate robust performance and generalizability across diverse scenarios. Second, data-driven scheduling technology is developed to automatically initiate preprocessing and fire detection tasks, significantly reducing fire discovery time. Finally, an integrated framework of multi-source remote sensing data, advanced detection algorithms, and a user-friendly visualization interface is proposed. This framework enables all-weather, fully automated forest fire monitoring and early warning, facilitating dynamic tracking of fire evolution and precise fire line localization through the cross-application of heterogeneous data sources. The framework’s effectiveness and practicality are validated through wildfire cases in two regions of Yunnan Province, offering scalable technical support for improving early detection of and rapid response to forest fires.

Compared with traditional nadir observations, multi-angle hyperspectral remote sensing can obtain more spatial and spectral information and improve the inversion precision of structure information on the Earth’s surface. However, processing multi-angle remote-sensing images presents new challenges. Owing to the multi-angle sensors used to obtain multi-angle images, there are major differences in the spatial and spectral information between each angle in these images. Data from the Compact High-resolution Imaging Spectrometer (CHRIS) on Project for On-Board Autonomy (PROBA) should be pre-processed to extract the BRDF (Bidirectional Reflectance Distribution Functions). Given the limitations of the pre-processing software currently available for CHRIS/PROBA images, and the lack of metadata and auxiliary origin schedules, some CHRIS multi-angle images cannot be pre-processed correctly. In the study, to promote the application of multi-angle data, a formula for calculating key parameters according to in-orbit geometric imaging relationships is derived to design a multi-angle image process flow including image rollovers, bad-line repairs, orthorectification and atmospheric corrections accounting for terrain effects. Test results indicate that the pre-processing method can quickly and effectively recover multi-angle hyperspectral information and obtain spectral characteristics of multi-angle observations.

Snow depth is a critical parameter for characterizing snow dynamics and water resources, and its accurate inversion is essential for hydrological processes, climate studies, and disaster prevention in cold regions. Based on long-term daily ground meteorological observation data from the hydrological years 1961 to 2015 at two meteorological stations in Mohe and Mishan, Heilongjiang Province, China, this study integrates physical parameters of snow density and snow albedo from the ERA5-Land reanalysis data to systematically compare the performance of three machine learning and three deep learning models in retrieving daily snow depth. Four feature combination schemes were designed to evaluate the contributions of meteorological factors, lagged snow depth terms, and snow physical parameters. The results indicate that, for both machine learning and deep learning models, the first-order lagged value of snow depth is the most important variable determining prediction accuracy. In terms of model performance, machine learning methods excelled, with XGBoost performing particularly outstandingly, achieving optimal prediction accuracy and stability under the best feature combination (coefficient of determination, R², reaching 0.989; root mean square error, RMSE, of 1.19 cm). Among deep learning methods, 1D CNN demonstrated strong local feature extraction capabilities, achieving prediction accuracy comparable to the best-performing machine learning model (R² of 0.9878, RMSE of 1.26 cm). Notably, models specifically designed for time-series data, such as LSTM (R² of 0.9848, RMSE of 1.41 cm), and the more complex 1D CNN-LSTM hybrid model (R² of 0.9803, RMSE of 1.60 cm) did not show significant advantages in this study. This indicates that model complexity and predictive performance are not simply positively correlated. Through comprehensive analysis of data from both stations, this study demonstrates that a prediction framework centered on historical snow depth as the core driving factor, combined with key meteorological elements, is highly robust. Although the inclusion of ERA5-Land snow physical parameters did not significantly improve model accuracy, it provides important insights for the future development of hybrid models that integrate physical mechanisms with data-driven approaches. The findings offer an effective solution for reconstructing long-term snow depth time series and hold significant application value for simulating cryospheric hydrological processes and studying climate change.

The utilization of photography imagery captured using cameras mounted on unmanned aerial vehicles (UAVs) for aboveground biomass (AGB) inventory has seen rapid growth in recent years. Existing research has predominantly focused on utilizing spectral and textural features for biomass inversion. However, estimating the AGB of trees remains a great challenge using stereoscopic imagery without the help of a digital terrain model (DTM). This study introduces five DTM-independent crown metrics using a digital surface model (DSM) and a canopy height model (CHM) derived from UAV stereoscopic imagery. The accuracy of the five metrics was evaluated against field measurements. The results indicate that the relationship between the crown cross-sectional area (CCSA) and AGB is stronger than that between tree height (TH) and AGB, with R² = 0.62 and RMSE = 69.22 (kg/tree) for Larix gmelinii and R² = 0.93 and RMSE = 142.06 (kg/tree) for Pinus sylvestris. Moreover, these DTM-independent crown metrics could be used to estimate the AGB of forests in the Greater and Lesser Khingan Mountain, with R² = 0.77 and RMSE = 77.10 (kg/tree) for coniferous trees and R² = 0.78 and RMSE = 72.46 (kg/tree) for all other trees. The results of this study demonstrate that UAV stereoscopic imagery can capture forest canopy information, and DTM-independent crown metrics can be used for AGB inversion where information on terrain under forest is unavailable.