Search

Improvements in hyperspectral image technology, diversification methods, and cost reductions have increased the convenience of hyperspectral data acquisitions. However, because of their multiband and multiredundant characteristics, hyperspectral data processing is still complex. Two feature extraction algorithms, the autoencoder (AE) and restricted Boltzmann machine (RBM), were used to optimize the classification model parameters. The optimal classification model was obtained by comparing a stacked autoencoder (SAE) and a deep belief network (DBN). Finally, the SAE was further optimized by adding sparse representation constraints and GPU parallel computation to improve classification accuracy and speed. The research results show that the SAE enhanced by deep learning is superior to the traditional feature extraction algorithm. The optimal classification model based on deep learning, namely, the stacked sparse autoencoder, achieved 93.41% and 94.92% classification accuracy using two experimental datasets. The use of parallel computing increased the model’s training speed by more than seven times, solving the model’s lengthy training time limitation.

Glacial lakes are important freshwater resources in southern Tibet. However, glacial lake outburst floods have significantly jeopardized the safety of local residents. To better understand the changes in glacial lakes in response to climate change, it is necessary to conduct a long-term evaluation on the areal dynamics of glacial lakes, assisted with local observations. Here, we propose an innovative method of classification and stacking extraction to accurately delineate glacial lakes in southwestern Tibet from 1990 to 2020. Based on Landsat images and meteorological data, we used geographic detectors to detect correlation factors. Multiple regression models were used to analyze the driving factors of the changes in glacier lake area. We combined bathymetric data of the glacial lakes with the changes in climatic variables and utilized HEC-RAS to determine critical circumstances for glacial lake outbursts. The results show that the area of glacial lakes in Nyalam County increased from 27.95 km2 in 1990 to 52.85 km2 in 2020, and eight more glacial lakes were observed in the study area. The glacial lake area expanded by 89.09%, where we found significant growth from 2015 to 2020. The correlation analysis between the glacial lake area and climate change throughout the period shows that temperature and precipitation dominate the expansion of these lakes from 1990 to 2020. We also discover that the progressive increase in water volume of glacial lakes can be attributed to the constant rise in temperature and freeze–thaw of surrounding glaciers. Finally, the critical conditions for the glacial lake’s outburst were predicted by using HEC-RAS combined with the changes in the water volume and climatic factors. It is concluded that GangxiCo endures a maximum water flow of 4.3 × 108 m3, and the glacial lake is in a stable changing stage. This conclusion is consistent with the field investigation and can inform the prediction of glacial lake outbursts in southwestern Tibet in the future.

Panzhihua City is a typical agricultural-forestry-pastoral and ecologically sensitive city in China. It is also an important ecological defense in the upper Yangtze River. It has abundant mineral resources, including vanadium, titanium, and water supplies. However, ecological and environmental problems emerge due to the excessive development of mining, agriculture, animal husbandry, and other non-natural urban economies. Therefore, a scientific understanding of the spatio-temporal changes of the eco-environment of Panzhihua is critical for environmental protection, city planning, and construction. To objectively evaluate the eco-environmental status of Panzhihua, the remote sensing-based ecological index (RSEI) was first applied to Panzhihua, a typical resource-based city, and its ecological environmental quality (EEQ) was quantitatively assessed from 1990 to 2020. This study explored the effects of mining activities and policies on EEQ and used change detection to reveal the spatial-temporal changes of EEQ in Panzhihua City over the past three decades. In addition, this study also verified the suitability of RSEI for evaluating EEQ in resource-based city using spatial autocorrelation, revealed the spatial heterogeneity of EEQ in Panzhihua City using optimized hot spot analysis, and showed different ecological clustering by hot spot analysis at two scales of urban and mining areas. According to the results: (1) From 1990 to 2020, the general eco-environmental condition of Panzhihua is improving, but there are still regional differences. (2) The Moran’s I value ranges from 0.436 (1990) to 0.700 (2020), indicating that there is autocorrelation in the distribution of eco-environmental quality. (3) At the mine, the mean value of RSEI dropped by 20–40%, and the EEQ decreased significantly due to mining activities. (4) A series of ecological restoration policies can buffer the negative impact of mining activities on the ecosystem, resulting in a slight improvement in the quality of the ecological environment. This study evaluates the EEQ of resource-based city and its spatial-temporal changes using RSEI constructed by the Google Earth Engine (GEE) platform, which can provide theoretical support for ecological and environmental conditions monitoring, development planning, and environmental protection policy-making of a resource-based city.