No abstract

No abstract

Understanding the effects of land cover changes on ecosystem carbon stocks is essential for ecosystem management and environmental protection, particularly in the transboundary region that has undergone marked changes. This study aimed to examine the impacts of land cover changes on ecosystem carbon stocks in the transboundary Tumen River Basin (TTRB). We extracted the spatial information from Landsat Thematic Imager (TM) and Operational Land Imager (OLI) images for the years 1990 and 2015 and obtained convincing estimates of terrestrial biomass and soil carbon stocks with the InVEST model. The results showed that forestland, cropland and built-up land increased by 57.5, 429.7 and 128.9 km 2 , respectively, while grassland, wetland and barren land declined by 24.9, 548.0 and 43.0 km 2 , respectively in the TTRB from 1990 to 2015. The total carbon stocks encompassing aboveground, belowground, soil and litter layer carbon storage pools have declined from 831.48 Tg C in 1990 to 831.42 Tg C in 2015 due to land cover changes. In detail, the carbon stocks decreased by 3.13 Tg C and 0.44 Tg C in Democratic People's Republic of Korea (North Korea) and Russia, respectively, while increased by 3.51 Tg C in China. Furthermore, economic development, and national policy accounted for most land cover changes in the TTRB. Our results imply that effective wetland and forestland protection policies among China, North Korea, and Russia are much needed for protecting the natural resources, promoting local ecosystem services and regional sustainable development in the transnational area.

The accurate characterization of tree species distribution in forest areas can help significantly reduce uncertainties in the estimation of ecosystem parameters and forest resources. Deep learning algorithms have become a hot topic in recent years, but they have so far not been applied to tree species classification. In this study, one-dimensional convolutional neural network (Conv1D), a popular deep learning algorithm, was proposed to automatically identify tree species using OHS-1 hyperspectral images. Additionally, the random forest (RF) classifier was applied to compare to the algorithm of deep learning. Based on our experiments, we drew three main conclusions: First, the OHS-1 hyperspectral images used in this study have high spatial resolution (10 m), which reduces the influence of mixed pixel effect and greatly improves the classification accuracy. Second, limited by the amount of sample data, Conv1D-based classifier does not need too many layers to achieve high classification accuracy. In addition, the size of the convolution kernel has a great influence on the classification accuracy. Finally, the accuracy of Conv1D (85.04%) is higher than that of RF model (80.61%). Especially for broadleaf species with similar spectral characteristics, such as Manchurian walnut and aspen, the accuracy of Conv1D-based classifier is significantly higher than RF classifier (87.15% and 71.77%, respectively). Thus, the Conv1D-based deep learning framework combined with hyperspectral imagery can efficiently improve the accuracy of tree species classification and has great application prospects in the future.

No abstract

No abstract

No abstract

In terms of evident climate change and human activities, investigating changes in lakes and reservoirs is critical for sustainable protection of water resources and ecosystem management over the Nenjiang watershed (NJW), an eco-sensitive semi-arid region and the third-largest inland waterbody cluster in China. In this study, we established a multi-temporal dataset documenting lake and reservoir (area ≥ 1 km2) changes in this region using an object-oriented image classification method and Landsat series images from 1980 to 2015. Using the structural equation model (SEM), we analyzed the diverse impacts of climatic and anthropogenic variables on lake changes. Results indicated that lakes experienced significant changes with fluctuations over the past 35 years including obvious declines in the total area (by 42%) and number (by 51%) from 1980 to 2010 and a slight increase in the total lake area and number from 2010 to 2015. More than 235 lakes in the size class of 1–10 km2 decreased to small lakes (area < 1 km2), while 59 lakes covering 243.75 km2 disappeared. Total reservoir area and number had continuous increases during the investigated 35 years, with an areal expansion of 54.9% from 919 km2 to 1422 km2, and a number increase by 65.3% from 78 to 129. The SEM revealed that the lake area in the NJW had a significant correlation with the mean annual precipitation (MAP), suggesting that the MAP decline clarified most of the lake shrinkage in the NJW. Furthermore, agricultural consumption of water had potential impacts on lake changes, suggested by the significant relationship between cropland area and lake area.