Land surface temperature (LST) plays a critical role in the water cycle and energy balance at global and regional scales. Large-scale LST estimates can be obtained from satellite observations and reanalysis data. In this study, we first investigate the long-term changes of LST during 2003–2017 on a per-pixel basis using three different datasets derived from (i) the Atmospheric Infrared Sounder (AIRS) onboard Aqua satellite, (ii) the Moderate Resolution Imaging Spectroradiometer (MODIS) also aboard Aqua, and (iii) the recently released ERA5-Land reanalysis data. It was found that the spatio-temporal patterns of these data agree very well. All three products globally showed an uptrend in the annual average LST during 2003–2017 but with considerable spatial variations. The strongest increase was found over the region north of 45° N, particularly over Asian Russia, whereas a slight decrease was observed over Australia. The regression analysis indicated that precipitation (P), incoming surface solar radiation (SW↓), and incoming surface longwave radiation (LW↓) can together explain the inter-annual LST variations over most regions, except over tropical forests, where the inter-annual LST variation is low. Spatially, the LST changes during 2003–2017 over the region north of 45° N were mainly influenced by LW↓, while P and SW↓ played a more important role over other regions. A detailed look at Asian Russia and the Amazon rainforest at a monthly time scale showed that warming in Asian Russia is dominated by LST increases in February–April, which are closely related with the simultaneously increasing LW↓ and clouds. Over the southern Amazon, the most apparent LST increase is found in the dry season (August–September), primarily affected by decreasing P. In addition, increasing SW↓ associated with decreasing atmospheric aerosols was another factor found to cause LST increases. This study shows a high level of consistency among LST trends derived from satellite and reanalysis products, thus providing more robust characteristics of the spatio-temporal LST changes during 2003–2017. Furthermore, the major climatic drivers of LST changes during 2003–2017 were identified over different regions, which might help us predict the LST in response to changing climate in the future.

An integrated understanding of the biogeochemical consequences of climate extremes and land use changes is needed to constrain land-surface feedbacks to atmospheric CO₂ from associated climate change. Past assessments of the global carbon balance have shown particularly high uncertainty in Southeast Asia. Here, we use a combination of model ensembles to show that intensified land use change made Southeast Asia a strong source of CO₂ from the 1980s to 1990s, whereas the region was close to carbon neutral in the 2000s due to an enhanced CO₂ fertilization effect and absence of moderate-to-strong El Niño events. Our findings suggest that despite ongoing deforestation, CO₂ emissions were substantially decreased during the 2000s, largely owing to milder climate that restores photosynthetic capacity and suppresses peat and deforestation fire emissions. The occurrence of strong El Niño events after 2009 suggests that the region has returned to conditions of increased vulnerability of carbon stocks.

Soil moisture (SM) plays an important role in hydrological cycle and weather forecasting. Satellite provides the only viable approach to regularly observe large-scale SM dynamics. Conventionally, SM is estimated from satellite observations based on the radiative transfer theory. Recent studies have demonstrated that the neural network (NN) method can retrieve SM with comparable accuracy as conventional methods. Here, we are interested in whether the NN model with more complex structures, namely deep convolutional neural network (DCNN), can bring about further improvement in SM retrievals when compared with the NN model used in recent studies. To achieve this objective, the same input data are used for the DCNN and NN models, including L-band Soil Moisture and Ocean Salinity (SMOS) brightness temperature (TB), C-band Advanced Scatterometer (ASCAT) backscattering coefficients, Moderate Resolution Imaging Spectroradiometer (MODIS) Normalized Difference Vegetation Index (NDVI) and soil temperature. The target SM used to train the DCNN and NN models is the European Center for Medium-range Weather Forecasts Re-Analysis Interim (ERA-Interim) product. The experiment consists of two phases: the learning phase from 1 January to 31 December 2015 and the testing phase from 1 January to 31 December 2016. In the learning phase, we train the DCNN and NN models using the ERA-Interim SM. When evaluation between DCNN and NN against in situ measurements in the testing phase, we find that the temporal correlations between DCNN SM and in situ measurements are higher than those between NN SM and in situ measurements by $6.2\%$ and $2.5\%$ on ascending and descending orbits, respectively. In addition, from the perspective of temporal and spatial dynamics, the simulated SM values by DCNN/NN and the ERA-Interim SM agree relatively well at a global scale. Results suggest that both NN and DCNN models are effective in estimating SM from satellite observations, and DCNN can achieve slightly better performance than NN.

Satellite-derived geophysical variables provide valuable information about the earth's functioning, but there are errors that limit their direct uses. Linearly combining two or more data sources is a simple and efficient method to reduce the uncertainty between the truth and observations. However, calculating the optimal weight for such a linear combination generally needs a reference ``truth'' that is rarely available in practical applications. To address this limitation, a triple collocation (TC) technique is often used to estimate data error by using a data triplet without the truth. The TC-based error variances are then applied to calculate the optimal weight calculation with an assumption of independence between errors. However, ignoring the error cross-correlations (ECCs) can lead to inaccurate optimal weight for a linear combination and hence may degrade the performance of calculations using merged products. Recently, an extended double instrumental variable (EIVD) method has been proposed to estimate ECC using only three products like TC. In this study, we examined the performance of an EIVD-based linear combination approach through two applications using synthetic data and actual satellite-derived soil moisture (SM) products. For direct comparison, the TC-based linear combination was also implemented using the same data sets. The verification showed that the EIVD-based products are generally better than the TC-based products and result in statistically significant differences in correlation coefficients. While the results here are based on SM, the proposed method is potentially applicable for reconstruction of other data sets that seldom have suitable references in large-scale regional settings.

Accurate assessment of the extent of crop distribution and mapping different crop types are essential for monitoring and managing modern agriculture. Medium and high spatial resolution remote sensing (RS) for Earth observation and deep learning (DL) constitute one of the most major and effective tools for crop mapping. In this study, we used high-resolution Sentinel-2 imagery from Google Earth Engine (GEE) to map paddy rice and winter wheat in the Bengbu city of Anhui Province, China. We compared the performance of different popular DL backbone networks with the traditional machine learning (ML) methods, including HRNet, MobileNet, Xception, and Swin Transformer, within the improved DeepLabv3+ architecture, Segformer and random forest (RF). The results showed that the Segformer based on the combination of the Transformer architecture encoder and the lightweight multilayer perceptron (MLP) decoder achieved an overall accuracy (OA) value of 91.06%, a mean F1 Score (mF1) value of 89.26% and a mean Intersection over Union (mIoU) value of 80.70%. The Segformer outperformed other DL methods by combining the results of multiple evaluation metrics. Except for Swin Transformer, which was slightly lower than RF in OA, all DL methods significantly outperformed RF methods in accuracy for the main mapping objects, with mIoU improving by about 13.5~26%. The predicted images of paddy rice and winter wheat from the Segformer were characterized by high mapping accuracy, clear field edges, distinct detail features and a low false classification rate. Consequently, DL is an efficient option for fast and accurate mapping of paddy rice and winter wheat based on RS imagery.