摘要
当利用多普勒激光雷达进行测量时,由于断电、低云、降水等影响会产生风廓线数据的缺测。使用双向长短期记忆(Bi-LSTM)网络对风廓线进行预测,当风廓线观测结果缺失时使用预测结果进行插补,以提升风场数据的获取率和连续性。2021年4月,多普勒激光雷达在辽宁省觉华岛地区开展大气风场探测实验,针对因断电和天气条件导致的风廓线数据缺测,使用2021年4月18日-22日的10 min平均风廓线集,探究基于时序的Bi-LSTM模型在风廓线预测方面的性能,并与非时序卷积神经网络(CNN)的预测性能进行对比。实验结果表明,Bi-LSTM模型具有良好的风廓线短期预测能力,基于连续完整的443条10 min平均风廓线训练的Bi-LSTM模型,对未来15条10 min平均的风廓线预测效果较好:u分量和v分量的决定系数分别高于0.6和0.5,均方根误差分别低于2 m·s^(-1)和3 m·s^(-1),平均绝对误差分别低于2 m·s^(-1)和3 m·s^(-1),其预测评估指标值均优于CNN模型。在本案例中,使用Bi-LSTM模型的预测结果对缺测风廓线插补,1000 m高度以下的数据获取率平均提升8百分点。
Objective High spatiotemporal resolution atmospheric wind field detection has important applications in pollution transport and diffusion,extreme weather monitoring,numerical weather forecasting,wind resource assessment,and other areas.Coherent Doppler lidar,as an active laser remote sensing device,acquires high spatiotemporal resolution vector wind field verticalstructure information.However,in practical applications,factors such as platform or power supply stability,and weather conditions can lead to missing wind profiles,limiting the application scope of windsensing lidar.Deep learning methods based on historical data modeling have been widely used in wind field prediction.The long shortterm memory(LSTM)network shows good performance in wind field prediction.However,most studies mainly focus on onedimensional temporal or spatial wind fields,while atmospheric wind fields exhibit both temporal and vertical spatial characteristics.Doppler lidar,as a high spatiotemporal resolution atmospheric wind field detection tool,obtains spatiotemporal twodimensional wind field information.Therefore,we propose a method using a bidirectional long shortterm memory(BiLSTM)model applied to wind field detection with lidar for wind profile prediction.The aim is to fully utilize the spatiotemporal twodimensional wind field data observed by the lidar,train a temporal BiLSTM model to capture the temporal variation characteristics of wind profiles,predict future wind profiles,interpolate missing wind profiles,and acquire more continuous wind field information.Methods Our study focuses on Doppler lidar atmospheric wind field detection experiments in Juehua Island,Liaoning Province,China.We utilize complete wind profile data for modeling and validation to predict and interpolate deficient wind profiles detected by the lidar.Previous complete wind profile data segments serve as the training and validation sets to establish wind profile prediction models based on a timeseries BiLSTM model and a nontime series convolutional neural network(CNN)model for the zonal component u and meridional component v of the wind profiles.We train the models using the same parameter settings,including step size,number of iterations,loss function,and optimization algorithm.We evaluate the wind profile prediction performance of the BiLSTM and CNN models using various metrics such as coefficient of determination(R^(2)),root mean square error(RMSE),and mean absolute error(MAE).The BiLSTM model with superior validation wind profile prediction performance is then used for deficient wind profile prediction and interpolation to obtain more continuous wind field information.Results and Discussions Based on the evaluation results of wind profile prediction(Fig.4),the BiLSTM model shows similar trends and ranges in performance evaluation metrics R^(2),RMSE,and MAE for different lookback steps.As a temporal network,the BiLSTM model exhibits consistent performance across different lookback steps,indicating that wind profiles have shortterm or longterm temporal dependencies that allow prediction based on past wind profiles at various time steps.With an increase in prediction time steps,errors accumulate gradually.After the 16th time step iteration,the model’s predictive capability rapidly declines,with R2 values for predicting u and v components falling below 0.5,indicating an inability to accurately forecast wind profiles beyond that point.This suggests that the BiLSTM model demonstrates good shortterm predictive ability for the next 15 wind profiles(within the next 2.5 h).Comparing the wind profile prediction performance of the temporal BiLSTM model with the nontemporal CNN model,the box plot analysis(Fig.5)reveals that the CNN model shows greater variability in R^(2) values for wind profile prediction across different lookbacks,indicating a more pronounced influence of the lookback parameter on wind profile prediction and greater uncertainty introduced by the choice of lookbacks.The BiLSTM model outperforms the CNN model in predicting u and v profiles,likely due to its ability to capture temporal features of wind profiles.In shortterm wind prediction,the BiLSTM model exhibits lower variability in R^(2) values across different lookbacks,demonstrating greater robustness in wind profile prediction.Compared to the CNN model,the BiLSTM model achieves higher R2 values and lower errors in prediction.The differences in predictive performance may stem from the CNN model’s proficiency in extracting local features using convolutional kernels,while wind profiles,as time series data,exhibit features closely related to preceding and subsequent time steps,potentially limiting the CNN model’s performance in handling such timedependent wind profile data.In contrast,the BiLSTM network,with bidirectional LSTM layers,considers features of wind profiles from multiple time steps in both directions,enabling it to better capture dependencies in time series data and make more accurate wind profile predictions.Future work involves incorporating timeseries data such as boundary layer height,temperature,humidity,and pressure as input features to further explore the BiLSTM model’s wind profile prediction performance(Fig.8).Additionally,we find it necessary to increase the number of training time steps to achieve better wind profile prediction results.Conclusions In the present study,we propose a method for wind profile prediction using a BiLSTM model applied to wind field detection with lidar.The aim is to fully utilize the spatiotemporal twodimensional wind field data observed by the lidar.By training a temporal BiLSTM model to extract the temporal variations of wind profiles,we predict future wind profiles and interpolate missing wind profiles.We conduct a comparison between the temporal BiLSTM model and the nontemporal CNN model in wind profile prediction.Our study reveals that the temporal BiLSTM model exhibits higher robustness in shortterm wind field prediction compared to the nontemporal CNN model.
作者
廉文超
宋小全
郝朝阳
姜萍
Lian Wenchao;Song Xiaoquan;Hao Zhaoyang;Jiang Ping(College of Marine Technology,Faculty of Information Science and Engineering,Ocean University of China,Qingdao 266100,Shandong,China;Laboratory for Regional Oceanography and Numerical Modeling,Qingdao Marine Science and Technology Center,Qingdao 266237,Shandong,China)
出处
《光学学报》
CSCD
北大核心
2024年第24期50-59,共10页
Acta Optica Sinica
基金
国家重点研发计划(2022YFC3700402,2022YFB3901700)
崂山实验室科技创新项目(LSKJ202201202)。
