Understanding the distribution and dynamics of glaciers is of great significance to the management and allocation of regional water resources and socio-economic development in arid regions of Northwest China.In this s...Understanding the distribution and dynamics of glaciers is of great significance to the management and allocation of regional water resources and socio-economic development in arid regions of Northwest China.In this study,based on 36 Landsat images,we extracted the glacier boundaries in the Manas River Basin,Northwest China from 2000 to 2020 using eCognition combined with band operation,GIS(geographic information system)spatial overlay techniques,and manual visual interpretation.We further analyzed the distribution and variation characteristics of glacier area,and simulated glacial runoff using a distributed degree-day model to explore the regulation of runoff recharge.The results showed that glacier area in the Manas River Basin as a whole showed a downward trend over the past 21 a,with a decrease of 10.86%and an average change rate of–0.54%/a.With the increase in glacier scale,the number of smaller glaciers decreased exponentially,and the number and area of larger glaciers were relatively stable.Glacier area showed a normal distribution trend of increasing first and then decreasing with elevation.About 97.92%of glaciers were distributed at 3700–4800 m,and 48.11%of glaciers were observed on the northern and northeastern slopes.The retreat rate of glaciers was the fastest(68.82%)at elevations below 3800 m.There was a clear rise in elevation at the end of glaciers.Glaciers at different slope directions showed a rapid melting trend from the western slope to the southern slope then to the northern slope.Glacial runoff in the basin showed a fluctuating upward trend in the past 21 a,with an increase rate of 0.03×10^(8) m^(3)/a.The average annual glacial runoff was 4.80×10^(8) m^(3),of which 33.31%was distributed in the ablation season(June–September).The average annual contribution rate of glacial meltwater to river runoff was 35.40%,and glacial runoff accounted for 45.37%of the total runoff during the ablation season.In addition,precipitation and glacial runoff had complementary regulation patterns for river runoff.The findings can provide a scientific basis for water resource management in the Manas River Basin and other similar arid inland river basins.展开更多
以风电为代表的新能源发电装机容量占比逐年增长,精确的风电功率超短期预测对提高风能利用率、助力双碳实现有重要意义。该文提出一种基于多元注意力框架与引导式监督学习的闭环风电功率超短期预测策略,从特征筛选、模型优化、策略改良...以风电为代表的新能源发电装机容量占比逐年增长,精确的风电功率超短期预测对提高风能利用率、助力双碳实现有重要意义。该文提出一种基于多元注意力框架与引导式监督学习的闭环风电功率超短期预测策略,从特征筛选、模型优化、策略改良3个角度全面提高预测准确性与模型智能性。首先,采用动态权重特征选择算法、孤立森林算法以及最邻近节点算法筛选并处理数据,便于预测模型更好把握其中特征;其次,对长短期记忆(long short term memory,LSTM)基模型多角度优化,并根据基模型中不同信息的特点,构建关于LSTM的多元注意力框架(Multielement-attention-LSTM),将此框架用于对LightGBM集成学习模型的引导,并通过多种可视化方法提高了模型可解释性;最后,将Bland-Altman应用于模型输出与实际风电出力一致性检验,在预测数据与实际数据交互的基础上实现训练–预测闭环机制。仿真结果表明,所构建的Multielement-attention-LSTM框架具有提高模型预测精度的作用,且闭环更新机制具备合理性。展开更多
基金supported by the National Natural Science Foundation of China(52169005)the Support Plan for Innovation and Development of Key Industries in southern Xinjiang,China(2022DB024)the Corps Science and Technology Innovation Talents Program Project of China(2023CB008-08).
文摘Understanding the distribution and dynamics of glaciers is of great significance to the management and allocation of regional water resources and socio-economic development in arid regions of Northwest China.In this study,based on 36 Landsat images,we extracted the glacier boundaries in the Manas River Basin,Northwest China from 2000 to 2020 using eCognition combined with band operation,GIS(geographic information system)spatial overlay techniques,and manual visual interpretation.We further analyzed the distribution and variation characteristics of glacier area,and simulated glacial runoff using a distributed degree-day model to explore the regulation of runoff recharge.The results showed that glacier area in the Manas River Basin as a whole showed a downward trend over the past 21 a,with a decrease of 10.86%and an average change rate of–0.54%/a.With the increase in glacier scale,the number of smaller glaciers decreased exponentially,and the number and area of larger glaciers were relatively stable.Glacier area showed a normal distribution trend of increasing first and then decreasing with elevation.About 97.92%of glaciers were distributed at 3700–4800 m,and 48.11%of glaciers were observed on the northern and northeastern slopes.The retreat rate of glaciers was the fastest(68.82%)at elevations below 3800 m.There was a clear rise in elevation at the end of glaciers.Glaciers at different slope directions showed a rapid melting trend from the western slope to the southern slope then to the northern slope.Glacial runoff in the basin showed a fluctuating upward trend in the past 21 a,with an increase rate of 0.03×10^(8) m^(3)/a.The average annual glacial runoff was 4.80×10^(8) m^(3),of which 33.31%was distributed in the ablation season(June–September).The average annual contribution rate of glacial meltwater to river runoff was 35.40%,and glacial runoff accounted for 45.37%of the total runoff during the ablation season.In addition,precipitation and glacial runoff had complementary regulation patterns for river runoff.The findings can provide a scientific basis for water resource management in the Manas River Basin and other similar arid inland river basins.
文摘以风电为代表的新能源发电装机容量占比逐年增长,精确的风电功率超短期预测对提高风能利用率、助力双碳实现有重要意义。该文提出一种基于多元注意力框架与引导式监督学习的闭环风电功率超短期预测策略,从特征筛选、模型优化、策略改良3个角度全面提高预测准确性与模型智能性。首先,采用动态权重特征选择算法、孤立森林算法以及最邻近节点算法筛选并处理数据,便于预测模型更好把握其中特征;其次,对长短期记忆(long short term memory,LSTM)基模型多角度优化,并根据基模型中不同信息的特点,构建关于LSTM的多元注意力框架(Multielement-attention-LSTM),将此框架用于对LightGBM集成学习模型的引导,并通过多种可视化方法提高了模型可解释性;最后,将Bland-Altman应用于模型输出与实际风电出力一致性检验,在预测数据与实际数据交互的基础上实现训练–预测闭环机制。仿真结果表明,所构建的Multielement-attention-LSTM框架具有提高模型预测精度的作用,且闭环更新机制具备合理性。
