Recently, ensemble multiple deep learning (DL) classifiers has been reported to be an effective method for improving remote sensing classification accuracy. Although these approaches still follow the conventional pattern of inputting instance features and outputting corresponding classes, they often overlook the intrinsic relationships between pixels beyond their spatial features. As a result, the diversity in the ensemble classification results primarily relies on different DL models. However, training the DL models consumes a significant amount of time, and training multiple networks not only incurs additional time costs but also affects the overall efficiency. To address this, a new approach has been proposed in this paper, which takes advantage of the relationships between pixels and their combinations to generate diverse classification results. It’s a novel ensemble classification framework, termed as the Doublet-Based Ensemble Classification Framework (DBECF), which eliminates the need for multiple classifiers. The DBECF starts by utilizing the training set to combine different samples to generate doublets. Then, features are assigned to these doublets through an exponentiation operation, resulting in a doublet training set. Using both the original training set and the derived doublet datasets, the DBECF is trained. For each input pixel, the DBECF produces multiple classification results, which are then integrated to obtain a more accurate output. To validate the proposed approach, experiments were conducted on three datasets, including multispectral images, hyperspectral images, and time series images. The maximum accuracies achieved by DBECF on the three datasets are 87.80 %, 97.71 %, and 83.51 %, respectively. In comparison to the contrastive methods, the incremental improvements in accuracy are 3.73 %, 7.66 %, and 9.16 %, respectively. The experimental results indicate that no matter using DL or non-deep learning for training, our proposed framework achieves progress on accuracy improvement outperforming classifications using comparative approach that based on single instance. This research provides a new perspective on the combination of DL and ensemble learning, highlighting its important implications and practical value in enhancing classification accuracy and efficiency.

Robust streamflow simulation at glacial basins is essential for the improvement of water sustainability assessment, water security evaluation, and water resource management under the rapidly changing climate. Therefore, we proposed a hybrid modelling framework to link the SWAT+ model considering glacial hydrological processes (GSWAT+) with Gated Recurrent Unit (GRU) neural networks to improve the model simulations and to establish a framework for the robust simulation and forecast of high and low flows in glacial river basins, which could be further used for the explorations of extreme hydrological events under a warming climate. The performance of different models (GSWAT+, GRU, and GRU-GSWAT+, respectively) were thoroughly investigated based on numerical experiments for two data-scarce glacial watersheds in Northwest China. The results suggested that the hybrid model (GRU-GSWAT+) outperformed both the individual deep learning (DL) model (GRU) and the conventional hydrological model (GSWAT+) in terms of simulation and prediction accuracy. Notably, the proposed hybrid model considerably enhanced the simulations of low and high flows that the conventional GSWAT+ failed to capture. Furthermore, utilizing suitable data integration (DI) schemes on feature and target sequences can substantially help to strengthen model stability and representativeness for monthly and annual streamflow sequences. Specifically, introducing one order differential method and decomposition approach, such as ensemble empirical signal decomposition (EEMD) and complete EEMD with adaptive noise (CEEMDAN), into feature and target sequences enriched the learnable ancillary information, which consequently strengthened the predictive performance of the proposed model. Overall, the proposed hybrid model with the suitable DI scheme has the potential to significantly enhance the accuracy of streamflow simulation in data-scarce glacial river basins. This hybrid model not only upheld the fundamental physical principles from the GSWAT+ model, but also considerably mitigated the accumulated bias errors, which caused by the shortage of climate data and inadequate hydrological principles, by using DL based model and DI schemes.

Vegetation dynamics in Qinghai-Tibet Plateau (QTP) have been debated in recent decades. Most studies suggest that wetter and warmer climatic conditions would release low temperature constraints and stimulate alpine vegetation growth. Other studies suggest that climate warming might inhibit vegetation growth by increasing soil moisture depletion in the southern QTP. Most of previous studies have relied on vegetation indices derived from satellite observations to retrieve large-scale vegetation changes, and the uncertainty of vegetation indices makes it difficult to accurately characterize the vegetation trends on the QTP. Here, we developed a deep learning algorithm in the Google Earth Engine (GEE) platform to accurately map the land use/cover change (LUCC) on the QTP, and then infer vegetation gain and loss and their drivers during the period 1988-2018. The vegetation on the QTP experienced rapid greening, which was dominated by transitions from bareland to alpine grassland (27.45 x 104 km2) and from alpine grassland to alpine meadow (17.43 x 104 km2) during 1988-2018. Furthermore, although human activities influence vegetation succession at the local scale, the dominant influ-encing factors affecting vegetation greening on the QTP are precipitation (q -statistic = 23.87 %) and temperature (q-statistic = 11.01 %). A 30-year time series analysis clarified the specific dynamics of vegetation on the QTP, thus contributing to the understanding of the response mechanisms of alpine vegetation under climate change and providing a reference for the formulating of reasonable ecological protection policies and human develop-ment strategies.

The growth of vegetation on the Qinghai Tibet Plateau (QTP) is experiencing significant changes due to climate change. There is still a lack of high -precision simulation methods for alpine grassland cover (AGC), and the climate feedback mechanisms of AGC remain unclear, which poses challenges for the production of highprecision AGC products and the formulation of ecological conservation policies. In this study, a transferable stacking deep learning (Stacking -DL) model is proposed based on a CNN, a DNN, and a GRU for AGC time series simulation. The applicability of deep learning models for AGC simulation is evaluated based on long time series of measured data, MODIS data, and environmental factors. Finally, the AGC spatiotemporal changes and controlling environmental factors in the alpine region were analyzed based on Sen 's slope and structural equation modeling (SEM). The results showed that feature selection and parameter optimization improved the applicability of the deep learning models in AGC simulations, and the DNN (R 2 = 0.899, RMSE = 0.078) model performed best among the base deep learning models. The Stacking -DL model combines the advantages of multiple models and achieves high transfer accuracy. In the YRSR, the AGC increase area (20.34 %) is greater than the AGC decrease area (3.34 %), the increase area is mainly located in the northeast, and the decrease area is mainly located in the southwest. AGC changes in the YRSR are mainly controlled by permafrost and climate. This study provides a high -precision and transferable vegetation monitoring model for alpine mountain regions based on advanced deep learning models and clarifies the response mechanism of AGC under climate change.

An accurate estimation of thaw depth is critical to understanding permafrost changes due to climate warming on the Qinghai-Tibetan Plateau (QTP). However, previous studies mainly focused on the interannual changes of active layer thickness (ALT) across the QTP, and little is known about the changes in the seasonal thaw depth. Machine learning (ML) is a critical tool to accurately estimate the ALT of permafrost, but a direct comparison of ML with deep learning (DL) in ALT projection regarding the model performance is still lacking. Here, ML, namely randomforest (RF), and DL algorithms like convolutional neural networks (CNN) and long short-term memory (LSTM) neural networks were compared to estimate the interannual changes of ALT and seasonal thaw depth on the QTP. Meteorological series, in-situ collected ALT observations, and geospatial information were used as predictors. The results show that both ML and DL methods are capable of estimating ALT and seasonal thaw depth in permafrost areas. The CNN and LSTM models developed using longer lagging times exhibit better performance in thaw depth prediction while the RF models are either mediocre or sometimes even worse as the lagging time increases. The results showthat the ALT from 2003 to 2011 on the QTP exhibits an increasing trend, especially in the northern region. In addition, 68.8%, 88.7%, 52.5%, and 47.5% of the permafrost regions on the QTP have deepened seasonal thaw depth in spring, summer, autumn, and winter, respectively. The correlation between air temperature and permafrost thaw depth ranges from 0.65 to 1 with the time lag ranging from 1 to 32 days. This study shows that ML and DL can be effectively used in retrieving ALT and seasonal thaw depth of permafrost, and could present an efficient way to figure out the interannual and seasonal variations of permafrost conditions under climate warming.

Vegetation dynamics in Qinghai-Tibet Plateau (QTP) have been debated in recent decades. Most studies suggest that wetter and warmer climatic conditions would release low temperature constraints and stimulate alpine vegetation growth. Other studies suggest that climate warming might inhibit vegetation growth by increasing soil moisture depletion in the southern QTP. Most of previous studies have relied on vegetation indices derived from satellite observations to retrieve large-scale vegetation changes, and the uncertainty of vegetation indices makes it difficult to accurately characterize the vegetation trends on the QTP. Here, we developed a deep learning algorithm in the Google Earth Engine (GEE) platform to accurately map the land use/cover change (LUCC) on the QTP, and then infer vegetation gain and loss and their drivers during the period 1988-2018. The vegetation on the QTP experienced rapid greening, which was dominated by transitions from bareland to alpine grassland (27.45 x 104 km2) and from alpine grassland to alpine meadow (17.43 x 104 km2) during 1988-2018. Furthermore, although human activities influence vegetation succession at the local scale, the dominant influ-encing factors affecting vegetation greening on the QTP are precipitation (q -statistic = 23.87 %) and temperature (q-statistic = 11.01 %). A 30-year time series analysis clarified the specific dynamics of vegetation on the QTP, thus contributing to the understanding of the response mechanisms of alpine vegetation under climate change and providing a reference for the formulating of reasonable ecological protection policies and human develop-ment strategies.

Retrogressive thaw slumps (RTSs) are among the most dynamic landforms resulting from the thawing of ice-rich permafrost. However, RTS distribution and evolution are poorly quantified because most of them occur in remote and inaccessible areas. In this study, we propose a method that integrates deep learning, change detection, and medial axis transform, aiming to automatically quantify the RTS development on multi-temporal images in the Beiluhe region on the Tibetan Plateau from 2017 to 2019. The images are taken by the Planet CubeSat constellation with high spatial and temporal resolution. The experiments show that automatic delineation based on deep learning can produce similar results to manual delineation, providing the potential of using these results to quantify the changes of RTS boundaries in different years. Our method reveals that among manuallydelineated 342 RTSs in the Beiluhe region, 83% and 76% of them expanded from 2017 to 2018 and 2018 to 2019, respectively. For the expansion from 2017 to 2018, the average and maximum expanding areas are 0.20 ha and 1.47 ha, while the average and maximum retreat distances are 21.3 m and 91 m, respectively. For 2018 to 2019 the average and maximum expansion areas and retreat distances are 0.22 ha, 2.53 ha, 25.0 m, and 212 m, respectively. The results show that the method can quantify RTS development automatically on multi-temporal images but may miss some small and subtle RTSs. Moreover, this study provides the very first quantitative report on RTS development on the Tibetan Plateau, which helps to advance the understanding of permafrost degradation.

Mass absorption cross- of black carbon (MAC(BC)) describes the absorptive cross- per unit mass of black carbon, and is, thus, an essential parameter to estimate the radiative forcing of black carbon. Many studies have sought to estimate MAC(BC) from a theoretical perspective, but these studies require the knowledge of a set of aerosol properties, which are difficult and/or labor-intensive to measure. We therefore investigate the ability of seven data analytical approaches (including different multivariate regressions, support vector machine, and neural networks) in predicting MAC(BC) for both ambient and biomass burning measurements. Our model utilizes multi-wavelength light absorption and scattering as well as the aerosol size distributions as input variables to predict MAC(BC) across different wavelengths. We assessed the applicability of the proposed approaches in estimating MAC(BC) using different statistical metrics (such as coefficient of determination (R-2), mean square error (MSE), fractional error, and fractional bias). Overall, the approaches used in this study can estimate MAC(BC) appropriately, but the prediction performance varies across approaches and atmospheric environments. Based on an uncertainty evaluation of our models and the empirical and theoretical approaches to predict MAC(BC), we preliminarily put forth support vector machine (SVM) as a recommended data analytical technique for use. We provide an operational tool built with the approaches presented in this paper to facilitate this procedure for future users.