摘要：The Chinese Software Developer Network (CSDN) is one of the largest information technology communities and service platforms in China. This paper describes the user profiling for CSDN, an evaluation track of SMP Cup 2017. It contains three tasks: (1) user document keyphrase extraction, (2) user tagging and (3) user growth value prediction. In the first task, we treat keyphrase extraction as a classification problem and train a Gradient-Boosting-Decision-Tree model with comprehensive features. In the second task, to deal with class imbalance and capture the interdependency between classes, we propose a two-stage framework: (1) for each class, we train a binary classifier to model each class against all of the other classes independently; (2) we feed the output of the trained classifiers into a softmax classifier, tagging each user with multiple labels. In the third task, we propose a comprehensive architecture to predict user growth value. Our contributions in this paper are summarized as follows: (1) we extract various types of features to identify the key factors in user value growth; (2) we use the semi-supervised method and the stacking technique to extend labeled data sets and increase the generality of the trained model, resulting in an impressive performance in our experiments. In the competition, we achieved the first place out of 329 teams.
Hyperspectral remote sensing technology is widely used to detect element contents because of its multiple bands, high resolution, and abundant information. Although researchers have paid considerable attention to selecting the optimal bandwidth for the hyperspectral inversion of metal element contents in rocks, the influence of bandwidth on the inversion accuracy are ignored. In this study, we collected 258 rock samples in and near the Kalatage polymetallic ore concentration area in the southwestern part of Hami City, Xinjiang Uygur Autonomous Region, China and measured the ground spectra of these samples. The original spectra were resampled with different bandwidths. A Partial Least Squares Regression (PLSR) model was used to invert Cu contents of rock samples and then the influence of different bandwidths on Cu content inversion accuracy was explored. According to the results, the PLSR model obtains the highest Cu content inversion accuracy at a bandwidth of 35 nm, with the model determination coefficient (R2) of 0.5907. The PLSR inversion accuracy is relatively unaffected by the bandwidth within 5–80 nm, but the accuracy decreases significantly at 85 nm bandwidth (R2=0.5473), and the accuracy gradually decreased at bandwidths beyond 85 nm. Hence, bandwidth has a certain impact on the inversion accuracy of Cu content in rocks using the PLSR model. This study provides an indicator argument and theoretical basis for the future design of hyperspectral sensors for rock geochemistry.
Long-term monitoring of the ecological environment changes is helpful for the protection of the ecological environment. Based on the ecological environment of the Sahel region in Africa, we established a remote sensing ecological index (RSEI) model for this region by combining dryness, moisture, greenness, and desertification indicators. Using the Moderate-resolution Imaging Spectroradiometer (MODIS) data in Google Earth Engine (GEE) platform, this study analyzed the ecological environment quality of the Sahel region during the period of 2001–2020. We used liner regression and fluctuation analysis methods to study the trend and fluctuation of RSEI, and utilized the stepwise regression approach to analyze the contribution of each indicator to the RSEI. Further, the correlation analysis was used to analyze the correlation between RSEI and precipitation, and Hurst index was applied to evaluate the change trend of RSEI in the future. The results show that RSEI of the Sahel region exhibited spatial heterogeneity. Specifically, it exhibited a decrease in gradient from south to north of the Sahel region. Moreover, RSEI in parts of the Sahel region presented non-zonal features. Different land-cover types demonstrated different RSEI values and changing trends. We found that RSEI and precipitation were positively correlated, suggesting that precipitation is the controlling factor of RSEI. The areas where RSEI values presented an increasing trend were slightly less than the areas where RSEI values presented a decreasing trend. In the Sahel region, the areas with the ecological environment characterized by continuous deterioration and continuous improvement accounted for 44.02% and 28.29% of the total study area, respectively, and the areas in which the ecological environment was changing from improvement to deterioration and from deterioration to improvement accounted for 12.42% and 15.26% of the whole area, respectively. In the face of the current ecological environment and future change trends of RSEI in the Sahel region, the research results provide a reference for the construction of the ''Green Great Wall'' (GGW) ecological environment project in Africa.
摘要： Doped elements in alloys significantly impact their performance. Conventional methods usually sputter the surface material of the sample, or their performance is limited to the surface of alloys owing to their poor penetration ability. The X-ray K-edge subtraction (KES) method exhibits great potential for the nondestructive in situ detection of element contents in alloys. However, the signal of doped elements usually deteriorates because of the strong absorption of the principal component and scattering of crystal grains. This in turn prevents the extensive application of X-ray KES imaging to alloys. In this study, methods were developed to calibrate the linearity between the grayscale of the KES image and element content. The methods were aimed at the sensitive analysis of elements in alloys. Furthermore, experiments with phantoms and alloys demonstrated that, after elaborate calibration, X-ray KES imaging is capable of nondestructive and sensitive analysis of doped elements in alloys.
摘要： The advantages of once-through molten salt reactors include readily available fuel, low nuclear proliferation risk, and low technical difficulty. It is potentially the most easily commercialized fuel cycle mode for molten salt reactors. However, there are some problems in the parameter selection of once-through molten salt reactors, and the relevant burnup optimization work requires further analysis. This study examined a once-through graphite moderated molten salt reactor using enriched uranium and thorium. The fuel volume fraction (VF), initial heavy nuclei concentration (HN0), feeding uranium enrichment (EFU), volume of the reactor core, and fuel type were changed to obtain the optimal conditions for burnup. We found an optimal region for VF and HN0 in each scheme, and the location and size of the optimal region changed with the degree of EFU, core volume, and fuel type. The recommended core schemes provide a reference for the core design of a once-through molten salt reactor.
摘要：Information on the Fe content of bare rocks is needed for implementing geochemical processes and identifying mines. However, the influence of Fe content on the spectra of bare rocks has not been thoroughly analyzed in previous studies. The Saur Mountain region within the Hoboksar of the Russell Hill depression was selected as the study area. Specifically, we analyzed six hyperspectral indices related to rock Fe content based on laboratory measurements (Dataset I) and field measurements (Dataset II). In situ field measurements were acquired to verify the laboratory measurements. Fe content of the rock samples collected from different fresh and weathered rock surfaces were divided into six levels to reveal the spatial distributions of Fe content of these samples. In addition, we clearly displayed wavelengths with obvious characteristics by analyzing the spectra of these samples. The results of this work indicated that Fe content estimation models based on the fresh rock surface measurements in the laboratory can be applied to in situ field or satellite-based measurements of Fe content of the weathered rock surfaces. It is not the best way to use only the single wavelengths reflectance at all absorption wavelengths or the depth of these absorption features to estimate Fe content. Based on sample data analysis, the comparison with other indices revealed that the performance of the modified normalized difference index is the best indicator for estimating rock Fe content, with R2 values of 0.45 and 0.40 corresponding to datasets I and II, respectively. Hence, the modified normalized difference index (the wavelengths of 2220, 2290, and 2370 nm) identified in this study could contribute considerably to improve the identification accuracy of rock Fe content in the bare rock areas. The method proposed in this study can obviously provide an efficient solution for large-scale rock Fe content measurements in the field.
摘要：The Tarim River is the longest inland river in China and is considered as an important river to protect the oasis economy and environment of the Tarim Basin. However, excessive exploitation and over-utilization of natural resources, particularly water resources, have triggered a series of ecological and environmental problems, such as the reduction in the volume of water in the main river, deterioration of water quality, drying up of downstream rivers, degradation of vegetation, and land desertification. In this study, the land use/land cover change (LUCC) responses to ecological water conveyance in the lower reaches of the Tarim River were investigated using ENVI (Environment for Visualizing Images) and GIS (Geographic Information System) data analysis software for the period of 1990–2018. Multi-temporal remote sensing images and ecological water conveyance data from 1990 to 2018 were used. The results indicate that LUCC covered an area of 2644.34 km2 during this period, accounting for 15.79% of the total study area. From 1990 to 2018, wetland, farmland, forestland, and artificial surfaces increased by 533.42 km2 (216.77%), 446.68 km2 (123.66%), 284.55 km2 (5.67%), and 57.51 km2 (217.96%), respectively, whereas areas covered by grassland and other land use/land cover types, such as Gobi, bare soil, and deserts, decreased by 103.34 km2 (14.31%) and 1218.83 km2 (11.75%), respectively. Vegetation area decreased first and then increased, with the order of 2010<2000<1990<2018. LUCC in the overflow and stagnant areas in the lower reaches of the Tarim River was mainly characterized by fragmentation, irregularity, and complexity. By analyzing the LUCC responses to 19 rounds of ecological water conveyance in the lower reaches of the Tarim River from 2000 to the end of 2018, we proposed guidelines for the rational development and utilization of water and soil resources and formulation of strategies for the sustainable development of the lower reaches of the Tarim River. This study provides scientific guidance for optimal scheduling of water resources in the region.
摘要：Nitrogen (N) and phosphorus (P) are two essential nutrients that determine plant growth and many nutrient cycling processes. Increasing N and P deposition is an important driver of ecosystem changes. However, in contrast to numerous studies about the impacts of nutrient addition on forests and temperate grasslands, how plant foliar stoichiometry and nutrient resorption respond to N and P addition in alpine grasslands is poorly understood. Therefore, we conducted an N and P addition experiment (involving control, N addition, P addition, and N+P addition) in an alpine grassland on Kunlun Mountains (Xinjiang Uygur Autonomous Region, China) in 2016 and 2017 to investigate the changes in leaf nutrient concentrations (i.e., leaf N, Leaf P, and leaf N:P ratio) and nutrient resorption efficiency of Seriphidium rhodanthum and Stipa capillata, which are dominant species in this grassland. Results showed that N addition has significant effects on soil inorganic N (NO3–-N and NH4+-N) and leaf N of both species in the study periods. Compared with green leaves, leaf nutrient concentrations and nutrient resorption efficiency in senesced leaves of S. rhodanthum was more sensitive to N addition, whereas N addition influenced leaf N and leaf N:P ratio in green and senesced leaves of S. capillata. N addition did not influence N resorption efficiency of the two species. P addition and N+P addition significantly improved leaf P and had a negative effect on P resorption efficiency of the two species in the study period. These influences on plants can be explained by increasing P availability. The present results illustrated that the two species are more sensitive to P addition than N addition, which implies that P is the major limiting factor in the studied alpine grassland ecosystem. In addition, an interactive effect of N+P addition was only discernable with respect to soil availability, but did not affect plants. Therefore, exploring how nutrient characteristics and resorption response to N and P addition in the alpine grassland is important to understand nutrient use strategy of plants in terrestrial ecosystems.
摘要：Central Asia is located in the hinterland of Eurasia, comprising Kazakhstan, Uzbekistan, Kyrgyzstan, Turkmenistan, and Tajikistan; over 93.00% of the total area is dryland. Temperature rise and human activities have severe impacts on the fragile ecosystems. Since the 1970s, nearly half the great lakes in Central Asia have shrunk and rivers are drying rapidly owing to climate changes and human activities. Water shortage and ecological crisis have attracted extensive international attention. In general, ecosystem services in Central Asia are declining, particularly with respect to biodiversity, water, and soil conservation. Furthermore, the annual average temperature and annual precipitation in Central Asia increased by 0.30°C/decade and 6.9 mm/decade in recent decades, respectively. Temperature rise significantly affected glacier retreat in the Tianshan Mountains and Pamir Mountains, which may intensify water shortage in the 21st century. The increase in precipitation cannot counterbalance the aggravation of water shortage caused by the temperature rise and human activities in Central Asia. The population of Central Asia is growing gradually, and its economy is increasing steadily. Moreover, the agricultural land has not been expended in the last two decades. Thus, water and ecological crises, such as the Aral Sea shrinkage in the 21st century, cannot be attributed to agriculture extension any longer. Unbalanced regional development and water interception/transfer have led to the irrational exploitation of water resources in some watersheds, inducing downstream water shortage and ecological degradation. In addition, accelerated industrialization and urbanization have intensified this process. Therefore, all Central Asian countries must urgently reach a consensus and adopt common measures for water and ecological protection.
摘要：As the largest inland river basin of China, the Tarim River Basin (TRB), known for its various natural resources and fragile environment, has an increased risk of ecological crisis due to the intensive exploitation and utilization of water and land resources. Since the Ecological Water Diversion Project (EWDP), which was implemented in 2001 to save endangered desert vegetation, there has been growing evidence of ecological improvement in local regions, but few studies have performed a comprehensive ecological vulnerability assessment of the whole TRB. This study established an evaluation framework integrating the analytic hierarchy process (AHP) and entropy method to estimate the ecological vulnerability of the TRB covering climatic, ecological, and socioeconomic indicators during 2000–2017. Based on the geographical detector model, the importance of ten driving factors on the spatial-temporal variations of ecological vulnerability was explored. The results showed that the ecosystem of the TRB was fragile, with more than half of the area (57.27%) dominated by very heavy and heavy grades of ecological vulnerability, and 28.40% of the area had potential and light grades of ecological vulnerability. The light grade of ecological vulnerability was distributed in the northern regions (Aksu River and Weigan River catchments) and western regions (Kashgar River and Yarkant River catchments), while the heavy grade was located in the southern regions (Kunlun Mountains and Qarqan River catchments) and the Mainstream catchment. The ecosystems in the western and northern regions were less vulnerable than those in the southern and eastern regions. From 2000 to 2017, the overall improvement in ecological vulnerability in the whole TRB showed that the areas with great ecological improvement increased by 46.11%, while the areas with ecological degradation decreased by 9.64%. The vegetation cover and potential evapotranspiration (PET) were the obvious driving factors, explaining 57.56% and 21.55% of the changes in ecological vulnerability across the TRB, respectively. In terms of ecological vulnerability grade changes, obvious spatial differences were observed in the upper, middle, and lower reaches of the TRB due to the different vegetation and hydrothermal conditions. The alpine source region of the TRB showed obvious ecological improvement due to increased precipitation and temperature, but the alpine meadow of the Kaidu River catchment in the Middle Tianshan Mountains experienced degradation associated with overgrazing and local drought. The improved agricultural management technologies had positive effects on farmland ecological improvement, while the desert vegetation in oasis-desert ecotones showed a decreasing trend as a result of cropland reclamation and intensive drought. The desert riparian vegetation in the lower reaches of the Tarim River was greatly improved due to the implementation of the EWDP, which has been active for tens of years. These results provide comprehensive knowledge about ecological processes and mechanisms in the whole TRB and help to develop environmental restoration measures based on different ecological vulnerability grades in each sub-catchment.
摘要：Endophytic bacteria from halophytes have a wide range of application prospects in various fields, such as plant growth-promoting, biocontrol activity and stress resistance. The current study aimed to identify cultivable endophytic bacteria associated with halophytes grown in the salt-affected soil in Xinjiang Uygur Autonomous Region, China and to evaluate their plant beneficial traits and enzyme-producing activity. Endophytic bacteria were isolated from Reaumuria soongorica (PalL Maxim.), Artemisia carvifolia (Buch.-Ham. ex Roxb. Hort. Beng.), Peganum harmala L. and Suaeda dendroides (C. A. Mey. Moq.) by using the cultural-dependent method. Then we classified these bacteria based on the difference between their sequences of 16S rRNA (16S ribosomal RNA) gene. Results showed that the isolated bacteria from R. soongorica belonged to the genera Brucella, Bacillus and Variovorax. The bacteria from A. carvifolia belonged to the genera Micromonospora and Brucella. The bacteria from P. harmala belonged to the genera Paramesorhizobium, Bacillus and Peribacillus. The bacteria from S. dendroides belonged to the genus Bacillus. Notably, the genus Bacillus was detected in the three above plants, indicating that Bacillus is a common taxon of endophytic bacteria in halophytes. And, our results found that about 37.50% of the tested strains showed strong protease-producing activity, 6.25% of the tested strains showed strong cellulase-producing activity and 12.50% of the tested strains showed moderate lipase-producing activity. Besides, all isolated strains were positive for IAA (3-Indoleacetic acid) production, 31.25% of isolated strains exhibited a moderate phosphate solubilization activity and 50.00% of isolated strains exhibited a weak siderophore production activity. Our findings suggest that halophytes are valuable resources for identifying microbes with the ability to increase host plant growth and health in salt-affected soils.
摘要：There are numerous valley farmlands on the Chinese Loess Plateau (CLP), where suffers from low soil quality and high risk of soil salinization due to the shallow groundwater table and poor drainage system. Currently, research on the evolution processes and mechanisms of soil quality and salinization in these dammed-valley farmlands on the CLP is still inadequately understood. In this study, three kinds of dammed-valley farmlands in the hilly-gully areas of the northern CLP were selected, and the status of soil quality and the impact factors of soil salinization were examined. The dammed-valley farmlands include the new farmland created by the project of Gully Land Consolidation, the 60-a farmland created by sedimentation from check dam, and the 400-a farmland created by sedimentation from an ancient landslide-dammed lake. Results showed that (1) the newly created farmland had the lowest soil quality in terms of soil bulk density, porosity, soil organic carbon and total nitrogen among the three kinds of dammed-valley farmlands; (2) soil salinization occurred in the middle and upper reaches of the new and 60-a valley farmlands, whereas no soil salinization was found in the 400-a valley farmland; and (3) soil salinization and low soil nutrient were determined to be the two important factors that impacted the soil quality of the valley farmlands in the hilly-gully mountain areas of the CLP. We conclude that the dammed-valley farmlands on the CLP have a high risk of soil salinization due to the shallow groundwater table, alkalinity of the loessial soil and local landform feature, thus resulting in the low soil quality of the valley farmlands. Therefore, strengthening drainage and decreasing groundwater table are extremely important to improve the soil quality of the valley farmlands and guarantee the sustainable development of the valley agriculture on the CLP.
摘要： By re-analzying public metagenomic data from 101 patients infected with influenza A virus during the 2007-2012 H1N1 flu seasons in France, we identified 22 samples with SARS-CoV sequences. In 3 of them, the SARS genome sequences could be fully assembled out of each. These sequences are highly similar (99.99% and 99.7%) to the artificially constructed recombinant 5 SARS-CoV (SARSr-CoV) strains generated by the J. Craig Venter Institute in USA. Moreover, samples from different flu seasons have different SARS-CoV strains, and the divergence between these strains cannot be explained by natural evolution. Our study also shows that retrospective studies using public metagenomic data from past major epidemic outbreaks serves as a genomic strategy for the research of origins or spread of infectious diseases.
摘要：Nitrogen (N) addition has profound impacts on litter-mediated nutrient cycling. Numerous studies have reported different effects of N addition on litter decomposition, exhibiting positive, negative, or neutral effects. Previous meta-analysis of litter decomposition under N addition was mainly based on a small number of samples to allow comparisons among ecosystem types. This study presents the results of a meta-analysis incorporating data from 53 published studies (including 617 observations) across forests, grasslands, wetlands, and croplands in China, to investigate how environmental and experimental factors impact the effects of N addition on litter decomposition. Averaged across all of the studies, N addition significantly slows litter decomposition by 7.02%. Considering ecosystem types, N addition significantly accelerates litter decomposition by 3.70% and 11.22% in grasslands and wetlands, respectively, clearly inhibits litter decomposition by 14.53% in forests, and has no significant effects on litter decomposition in croplands. Regarding the accelerated litter decomposition rate in grasslands due to N addition, litter decomposition rate increases slightly with increasing rates of N addition. However, N addition slows litter decomposition in forests, but litter decomposition is at a significantly increasing rate with increasing amounts of N addition. The responses of litter decomposition to N addition are also influenced by the forms of N addition, experiential duration of N addition, humidity index, litter quality, and soil pH. In summary, N addition alters litter decomposition rate, but the direction and magnitude of the response are affected by the forms of N addition, the rate of N addition, ambient N deposition, experimental duration, and climate factors. Our study highlights the contrasting effects of N addition on litter decomposition in forests and grasslands. This finding could be used in biogeochemical models to better evaluate ecosystem carbon cycling under increasing N deposition due to the differential responses of litter decomposition to N addition rates and ecosystem types.
摘要：The Loess Plateau, located in Gansu Province, is an important energy base in China because most of the oil and gas resources are distributed in Gansu Province. In the last 40 a, ecological environment in this region has been extremely destroyed due to the over-exploitation of crude-oil resources. Remediation of crude-oil contaminated soil in this area remains to be a challenging task. In this study, in order to elucidate the effects of organic compost and biochar on phytoremediation of crude-oil contaminated soil (20 g/kg) by Calendula officinalis, we designed five treatments, i.e., natural attenuation (CK), planted C. officinalis only (P), planted C. officinalis with biochar amendment (PB), planted C. officinalis with organic compost amendment (PC), and planted C. officinalis with co-amendment of biochar and organic compost (PBC). After 152 d of cultivation, total petroleum hydrocarbons (TPH) removal rates of CK, P, PB, PC and PBC were 6.36%, 50.08%, 39.58%, 73.10% and 59.87%, respectively. Shoot and root dry weights of C. officinalis significantly increased by 172.31% and 80.96% under PC and 311.61% and 145.43% under PBC, respectively as compared with P (P<0.05). Total chlorophyll contents in leaves of C. officinalis under P, PC and PBC significantly increased by 77.36%, 125.50% and 79.80%, respectively (P<0.05) as compared with PB. Physical-chemical characteristics and enzymatic activity of soil in different treatments were also assessed. The highest total N, total P, available N, available P and SOM (soil organic matter) occurred in PC, followed by PBC (P<0.05). C. officinalis rhizospheric soil dehydrogenase (DHA) and polyphenol oxidase (PPO) activities in PB were lower than those of other treatments (P<0.05). The values of ACE (abundance-based coverage estimators) and Chao 1 indices for rhizospheric bacteria were the highest under PC followed by PBC, P, PB and CK (P<0.05). However, the Shannon index for bacteria was the highest under PC and PBC, followed by P, PB and CK (P<0.05). In terms of soil microbial community composition, Proteiniphilum, Immundisolibacteraceae and Solimonadaceae were relatively more abundant under PC and PBC. Relative abundances of Pseudallescheria, Ochroconis, Fusarium, Sarocladium, Podospora, Apodus, Pyrenochaetopsis and Schizothecium under PC and PBC were higher, while relative abundances of Gliomastix, Aspergillus and Alternaria were lower under PC and PBC. As per the nonmetric multidimensional scaling (NMDS) analysis, application of organic compost significantly promoted soil N and P contents, shoot length, root vitality, chlorophyll ratio, total chlorophyll, abundance and diversity of rhizospheric soil microbial community in C. officinalis. A high pH value and lower soil N and P contents induced by biochar, altered C. officinalis rhizospheric soil microbial community composition, which might have restrained its phytoremediation efficiency. The results suggest that organic compost-assisted C. officinalis phytoremediation for crude-oil contaminated soil was highly effective in the Loess Plateau, China.
摘要：Saltation is the major particle movement type in wind erosion process. Saltating sand grains can rebound up to tens of times larger in length and height over hard surface (such as gravel surface) than over loose sand surface. Gravels usually have different faces, causing distinct response of the impacting grains, but the effects of the grain and gravel-surface contact angle on grain rebound are not yet well quantified. We performed full-range controlled experiments of grain saltation using different contact angles, grain sizes and impact speeds in still air, to show that contact angle increases the height of representative saltation path but decreases particle travel length. The results were compared with outputs from the COMprehensive numerical model of SALTation (COMSALT). Large saltation height of 4.8 m and length of 9.0 m were recorded. The maximum and representative saltation height over the gravel surface were found to be about 4.9 times and 12.8 times those over the loose sandy surface, respectively. The maximum saltation length may be reduced by 58% and the representative saltation height may be increased by 77% as contact angle increases from 20° to 40°. We further showed that the collision inertia contributes 60% of the saltation length, and wind contributes to the other 40%. These quantitative findings have important implications for modeling saltation trajectory over gravel surface.
摘要：The accurate simulation and prediction of runoff in alpine glaciated watersheds is of increasing importance for the comprehensive management and utilization of water resources. In this study, long short-term memory (LSTM), a state-of-the-art artificial neural network algorithm, is applied to simulate the daily discharge of two data-sparse glaciated watersheds in the Tianshan Mountains in Central Asia. Two other classic machine learning methods, namely extreme gradient boosting (XGBoost) and support vector regression (SVR), along with a distributed hydrological model (Soil and Water Assessment Tool (SWAT) and an extended SWAT model (SWAT_Glacier) are also employed for comparison. This paper aims to provide an efficient and reliable method for simulating discharge in glaciated alpine regions that have insufficient observed meteorological data. The two typical basins in this study are the main tributaries (the Kumaric and Toxkan rivers) of the Aksu River in the south Tianshan Mountains, which are dominated by snow and glacier meltwater and precipitation. Our comparative analysis indicates that simulations from the LSTM shows the best agreement with the observations. The performance metrics Nash-Sutcliffe efficiency coefficient (NS) and correlation coefficient (R2) of LSTM are higher than 0.90 in both the training and testing periods in the Kumaric River Basin, and NS and R2 are also higher than 0.70 in the Toxkan River Basin. Compared to classic machine learning algorithms, LSTM shows significant advantages over most evaluating indices. XGBoost also has high NS value in the training period, but is prone to overfitting the discharge. Compared with the widely used hydrological models, LSTM has advantages in predicting accuracy, despite having fewer data inputs. Moreover, LSTM only requires meteorological data rather than physical characteristics of underlying data. As an extension of SWAT, the SWAT_Glacier model shows good adaptability in discharge simulation, outperforming the original SWAT model, but at the cost of increasing the complexity of the model. Compared with the oftentimes complex semi-distributed physical hydrological models, the LSTM method not only eliminates the tedious calibration process of hydrological parameters, but also significantly reduces the calculation time and costs. Overall, LSTM shows immense promise in dealing with scarce meteorological data in glaciated catchments.
摘要：Changes in atmospheric aerosols have profound effects on ecosystem productivity, vegetation growth and activity by directly and indirectly influencing climate and environment conditions. However, few studies have focused on the effects of atmospheric aerosols on vegetation growth and activity in the vulnerable arid and semi-arid regions, which are also the source areas of aerosols. Using the datasets of aerosol optical depth (AOD), normalized difference vegetation index (NDVI) and multiple climatic variables including photosynthetically active radiation (PAR), surface solar radiation (SSR), surface air temperature (TEM) and total precipitation (PRE), we analyzed the potential responses of vegetation activity to atmospheric aerosols and their associated climatic factors in arid and semi-arid regions of Asia from 2005 to 2015. Our results suggested that areas with decreasing growing-season NDVI were mainly observed in regions with relatively sparse vegetation coverage, while AOD tended to increase as NDVI decreased in these regions. Upon further analysis, we found that aerosols might exert a negative influence on vegetation activity by reducing SSR, PAR and TEM, as well as suppressing PRE in most arid and semi-arid regions of Asia. Moreover, the responses of atmospheric aerosols on vegetation activity varied among different growing stages. At the early growing stage, higher concentration of aerosol was accompanied with suppressed vegetation growth by enhancing cooling effects and reducing SSR and PAR. At the middle growing stage, aerosols tended to alter microphysical properties of clouds with suppressed PRE, thereby restricting vegetation growth. At the late growing stage, aerosols exerted significantly positive influences on vegetation activity by increasing SSR, PAR and TEM in regions with high anthropogenic aerosols. Overall, at different growing stages, aerosols could influence vegetation activity by changing different climatic factors including SSR, PAR, TEM and PRE in arid and semi-arid regions of Asia. This study not only clarifies the impacts of aerosols on vegetation activity in source areas, but also explains the roles of aerosols in climate.
摘要：Drought is an inevitable condition with negative impacts in the agricultural and climatic sectors, especially in developing countries. This study attempts to examine the spatial and temporal characteristics of drought and its trends in the Koshi River Basin (KRB) in Nepal, using the standardized precipitation evapotranspiration index (SPEI) over the period from 1987 to 2017. The Mann-Kendall test was used to explore the trends of the SPEI values. The study illustrated the increasing annual and seasonal drought trends in the KRB over the study period. Spatially, the hill region of the KRB showed substantial increasing drought trends at the annual and seasonal scales, especially in summer and winter. The mountain region also showed a significant increasing drought trend in winter. The drought characteristic analysis indicated that the maximum duration, intensity, and severity of drought events were observed in the KRB after 2000. The Terai region presented the highest drought frequency and intensity, while the hill region presented the longest maximum drought duration. Moreover, the spatial extent of drought showed a significant increasing trend in the hill region at the monthly (drought station proportion of 7.6%/10a in August), seasonal (drought station proportion of 7.2%/10a in summer), and annual (drought station proportion of 6.7%/10a) scales. The findings of this study can assist local governments, planners, and project implementers in understanding drought and developing appropriate mitigation strategies to cope with its impacts.