Wildfires have a significant impact on the atmosphere, terrestrial ecosystems, and society. Real-time monitoring of wildfire locations is crucial in fighting wildfires and reducing human casualties and property damage. Geostationary satellites offer the advantage of high temporal resolution and are gradually being used for real-time fire detection. In this study, we constructed a fire label dataset using the stable VNP14IMG fire product and used the random forest (RF) model for fire detection based on Himawari-8 multiband data. The band calculation features related brightness temperature, spatial features, and auxiliary data as input used in this framework for model training. We also used a recursive feature elimination method to evaluate the impact of these features on model accuracy and to exclude redundant features. The daytime and nighttime RF models (RF-D/RF-N) are separately constructed to analyze their applicability. Finally, we extensively evaluated the model performance by comparing them with the Japan Aerospace Exploration Agency (JAXA) wildfire product. The RF models exhibited higher accuracy, with recall and precision rates of 95.62% and 59%, respectively, and the recall rate for small fires was 19.44% higher than that of the JAXA wildfire product. Adding band calculation features and spatial features, as well as feature selection, effectively reduced the overfitting and improved the model’s generalization ability. The RF-D model had higher fire detection accuracy than the RF-N model. Omission errors and commission errors were mainly concentrated in the adjacent pixels of the fire clusters. In conclusion, our VIIRS fire product and Himawari-8 data-based fire detection model can monitor the fire location in real time and has excellent detection capability for small fires, making it highly significant for fire detection.

Long-term satellite observations of the water levels of lakes are crucial to our understanding of lake hydrological basin systems. The Ice, Cloud, and Land Elevation satellite (ICESat) and ICESat-2 were employed to monitor the water level of Qinghai Lake in the hydrological basin. The median of absolute deviation (MAD) method was exploited to remove the outliers. The results confirmed that the MAD range of ICESat was from 0.0525 to 0.2470 m, and the range of σ was from 0.0778 to 0.3662 m; the MAD range of ICESat-2 was from 0.0291 to 0.0490 m, and the range of σ was from 0.0431 to 0.0726 m; ICESat-2 was less than that of ICESat. The reference ellipsoid and geoid transfer equations were applied to convert the water level to the World Geodetic System (WGS84) and Earth Gravitational Model 2008 (EGM2008) geoid. The water level, as derived from laser altimeters, was validated by the Xiashe Hydrological Station; with ICESat, the coefficient of association (R) was 0.8419, the root mean square error (RMSE) was 0.1449 m, and the mean absolute error (MAE) was 0.1144 m; with ICESat-2, the R was 0.6917, the RMSE was 0.0531 m, and the MAE was 0.0647 m. The water levels from ICESat-2 are much more accurate than those from ICESat. The two combined laser altimeters showed that the R was 0.9931, the RMSE was 0.1309 m, and the MAE was 0.1035 m. The water level rise was 3.6584 m from 2004 to 2020. The rising rate was 0.2287 m/a. The collaborative use of the ICESat-2 and ICESat satellites made it easier to obtain the lake water levels.

Recently, time series remote sensing image (TSRSI) has been reported to be an effective resource for mapping fine land use/land cover (LULC), and deep learning, in particular, has been gaining growing attention in this field. However, existing deep learning methods often only learn features from either the temporal or spatial domain, neglecting the intercorrelation between temporal features, which may provide more information for classification, are not fully considered. To make full use of the relations between temporal features and to explore more objective features for improving classification accuracy, we proposed a feature relationship-based classification method. The method leverages the angles between features on the temporal curve to establish relationships between every pair and triplet of features, resulting in the creation of feature relationship matrices (FRMs) and feature relationship tensors (FRTs). Afterward, a 2-D–3-D multiscale convolutional neural network (2-D–3-D MSCNN) was designed to learn deep features from FRM and FRT, achieving the classification improvement of TSRSI. Our experiment was conducted on TSRSIs located in two counties, Sutter and Kings in California, USA. The experimental results indicate that compared to both deep learning and nondeep learning methods, the proposed approach achieves significant improvements in accuracy and LULC mapping, validating the effectiveness and feasibility of enhancing TSRSI classification accuracy through feature relationship learning.

Lake phenology is essential for understanding the lake freeze-thaw cycle effects on terrestrial hydrological processes. The Qinghai-Tibetan Plateau (QTP) has the most extensive ice reserve outside of the Arctic and Antarctic poles and is a sensitive indicator of global climate changes. Qinghai Lake, the largest lake in the QTP, plays a critical role in climate change. The freeze-thaw cycles of lakes were studied using daily Moderate Resolution Imaging Spectroradiometer (MODIS) data ranging from 2000–2018 in the Google Earth Engine (GEE) platform. Surface water/ice area, coverage, critical dates, surface water, and ice cover duration were extracted. Random forest (RF) was applied with a classifier accuracy of 0.9965 and a validation accuracy of 0.8072. Compared with six common water indexes (tasseled cap wetness (TCW), normalized difference water index (NDWI), modified normalized difference water index (MNDWI), automated water extraction index (AWEI), water index 2015 (WI2015) and multiband water index (MBWI)) and ice threshold value methods, the critical freeze-up start (FUS), freeze-up end (FUE), break-up start (BUS), and break-up end (BUE) dates were extracted by RF and validated by visual interpretation. The results showed an R² of 0.99, RMSE of 3.81 days, FUS and BUS overestimations of 2.50 days, and FUE and BUE underestimations of 0.85 days. RF performed well for lake freeze-thaw cycles. From 2000 to 2018, the FUS and FUE dates were delayed by 11.21 and 8.21 days, respectively, and the BUS and BUE dates were 8.59 and 1.26 days early, respectively. Two novel key indicators, namely date of the first negative land surface temperature (DFNLST) and date of the first positive land surface temperature (DFPLST), were proposed to comprehensively delineate lake phenology: DFNLST was approximately 37 days before FUS, and DFPLST was approximately 20 days before BUS, revealing that the first negative and first positive land surface temperatures occur increasingly earlier.