Multi-temporal Landsat-8 satellite images with fine spatial resolution (i.e., 30 m) are crucial for modern digital soil mapping (DSM). Generally, cloud-free images covering bare topsoil are common choices for DSM. However, the number of effective Landsat-8 data is greatly limited due to cloud contamination coupled with the coarse temporal resolution, and interference of material covering topsoil in most of the months, hindering the development of accurate DSM. To address this issue, temporally dense Landsat images were predicted using a spatio-temporal fusion method to improve DSM. Specifically, the recently developed virtual image pair-based spatio-temporal fusion (VIPSTF) method was adopted to produce simulated Landsat-8 time-series, by fusing with 500 m Moderate Resolution Imaging Spectroradiometer (MODIS) time-series with frequent observations. Subsequently, the simulated Landsat-8 data were used for distinguishing different soil classes via a random forest (RF) model. Training and validation samples of soil classes were collected from legacy soil data. Our results indicate that the simulated data were beneficial for improving DSM owing to the increase in class separability. More precisely, after combining the observed and simulated data, the overall accuracy (OA) and kappa coefficient (Kappa) were increased by 3.099% and 0.047, respectively. This research explored the potential of the spatio-temporal fusion method for DSM, providing a new solution for remote sensing-based DSM.

Fast-growing crops have been evolved in North China, accompanied by intense paddy expansion, leading to dramatic impacts on the agricultural environment. Among these environmental issues, the impact of paddy expansion on land surface temperature is still unclear. In the present study, based on Landsat images and MODIS land surface temperature (LST) products, the crop pattern and monthly LST in the northern Sanjiang Plain are obtained. A 1 km scale grid unit is built to investigate the relationship between LST and paddy expansion. The results obtained from the study are as follows. Firstly, for crop patterns, cropland planting is given priority to paddy fields, accompanied by an aggregated pattern, while upland crops present a discrete pattern. Secondly, for LST changes during the growing season, the maximum LST occurs in June, and the lowest values occur in October across the whole region. In addition, the LST of paddy fields is lower compared with that of upland crops for the whole growing season. Thirdly, at the 1 km grid scale, the relationship between monthly LST and paddy field ratio is significantly negative, and better represented by a cubic function rather than a linear fit. Finally, LST decreases with the increased fraction of the rice paddy area more rapidly when rice paddy is aggregated and accounted for by more than 80% of each study grid. The findings of this study are important to guide agricultural production and to better understand the environmental effects of paddy expansion in cold regions.

Remote sensing has been used as an important tool for disaster monitoring and disaster scope extraction, especially for the analysis of spatial and temporal disaster patterns of large-scale and long-duration series. Google Earth Engine provides the possibility of quickly extracting the disaster range over a large area. Based on the Google Earth Engine cloud platform, this study used MODIS vegetation index products with 250-m spatial resolution synthesized over 16 days from the period 2005–2019 to develop a rapid and effective method for monitoring disasters across a wide spatiotemporal range. Three types of disaster monitoring and scope extraction models are proposed: the normalized difference vegetation index (NDVI) median time standardization model (R_{NDVI_TM(i)}), the NDVI median phenology standardization model (R_{NDVI_AM(i)(j)}), and the NDVI median spatiotemporal standardization model (R_{NDVI_ZM(i)(j)}). The optimal disaster extraction threshold for each model in different time phases was determined using Otsu’s method, and the extraction results were verified by medium-resolution images and ground-measured data of the same or quasi-same period. Finally, the disaster scope of cultivated land in Heilongjiang Province from 2010–2019 was extracted, and the spatial and temporal patterns of the disasters were analyzed based on meteorological data. This analysis revealed that the three aforementioned models exhibited high disaster monitoring and range extraction capabilities, with verification accuracies of 97.46%, 96.90%, and 96.67% for R_{NDVI_TM(i)}, R_{NDVI_AM(i)}, and _(j)R_{NDVI_ZM(i)(j)}, respectively. The spatial and temporal disaster distributions were found to be consistent with the disasters of the insured plots and the meteorological data across the entire province. Moreover, different monitoring and extraction methods were used for different disasters, among which wind hazard and insect disasters often required a delay of 16 days prior to observation. Each model also displayed various sensitivities and was applicable to different disasters. Compared with other techniques, the proposed method is fast and easy to implement. This new approach can be applied to numerous types of disaster monitoring as well as large-scale agricultural disaster monitoring and can easily be applied to other research areas. This study presents a novel method for large-scale agricultural disaster monitoring.

Surface soil moisture (SSM) is a critical variable for understanding the terrestrial hydrologic cycle, and it influences ecosystem dynamics, agriculture productivity, and water resource management. Although SSM information is widely estimated through satellite-derived and model-assimilated methods, datasets with fine spatio-temporal resolutions remain unavailable at the continental scale and yet are essential for improving weather forecasting, optimizing precision irrigation, and enhancing fire risk assessment. In this study, we developed a new 3 h, 1 km spatially seamless SSM dataset spanning 2015 to 2023, covering the entire contiguous United States (CONUS), using a spatio-temporal fusion model. This approach effectively combines the distinct advantages of two long-term SSM datasets, namely, the Soil Moisture Active Passive (SMAP) L4 SSM product and the Crop Condition and Soil Moisture Analytics (Crop-CASMA) dataset. The SMAP product provides spatially seamless SSM observations with a 3 h temporal resolution but at a 9 km spatial resolution, while the Crop-CASMA SSM dataset offers a finer spatial resolution of 1 km but has a daily temporal resolution and contains spatial gaps. To overcome the spatio-temporal mismatch between the two products, we developed a time series data mining approach known as the highly comparative time series analysis (HCTSA) method to extract multiple spatially seamless characteristics (e.g., maximum and mean) from the two inter-annual SSM datasets (i.e., SMAP and Crop-CASMA). Then, the fusion model was constructed using the extracted 9 and 1 km characteristics and each scene of the SMAP in turn. Finally, the 3 h, 1 km SSM data (named STF_SSM) were predicted from 2015 to 2023. The comparison with in situ data from multiple SSM observation networks showed that the performance of our STF_SSM dataset is better than the Crop-CASMA and is close to the SMAP L4 product, with a mean correlation coefficient (CC) of 0.716 at the daily scale and 0.689 at the 3 h scale. The STF_SSM dataset in this study is the first long-time-series, spatially seamless SSM dataset to realize continuous intra-day 1 km SSM observations every 3 h across the CONUS, which provides a new insight into the fast changes in soil moisture along with drought and wet spell occurrences and ecosystem responses. Additionally, this dataset provides a valuable data source for the calibration and validation of land surface models. The STF_SSM dataset is available at https://doi.org/10.5066/P13CCN69 (Yang et al., 2025a).