当前位置: X-MOL 学术Theor. Appl. Climatol. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Deep learning–based downscaling of summer monsoon rainfall data over Indian region
Theoretical and Applied Climatology ( IF 2.8 ) Pub Date : 2021-01-02 , DOI: 10.1007/s00704-020-03489-6
Bipin Kumar , Rajib Chattopadhyay , Manmeet Singh , Niraj Chaudhari , Karthik Kodari , Amit Barve

Downscaling is necessary to generate high-resolution observation data to validate the climate model forecast or monitor rainfall at the micro-regional level operationally. Available observations generated by automated weather stations or meteorological observatories are often limited in spatial resolution resulting in misrepresentation or absence of rainfall information at these levels. Dynamical and statistical downscaling models are often used to get information at high-resolution gridded data over larger domains. As rainfall variability is dependent on the complex spatio-temporal process leading to non-linear or chaotic spatio-temporal variations, no single downscaling method can be considered efficient enough. In the domains dominated by complex topographies, quasi-periodicities, and non-linearities, deep learning (DL)–based methods provide an efficient solution in downscaling rainfall data for regional climate forecasting and real-time rainfall observation data at high spatial resolutions. We employed three deep learning-based algorithms derived from the super-resolution convolutional neural network (SRCNN) methods in this work. Summer monsoon season data from India Meteorological Department (IMD) and the tropical rainfall measuring mission (TRMM) data set were downscaled up to 4 times higher resolution using these methods. High-resolution data derived from deep learning-based models provide better results than linear interpolation for up to 4 times higher resolution. Among the three algorithms, namely, SRCNN, stacked SRCNN, and DeepSD, employed here, the best spatial distribution of rainfall amplitude and minimum root-mean-square error is produced by DeepSD-based downscaling. Hence, the use of the DeepSD algorithm is advocated for future use. We found that spatial discontinuity in amplitude and intensity rainfall patterns is the main obstacle in the downscaling of precipitation. Furthermore, we applied these methods for model data post-processing, in particular, ERA5 reanalysis data. Downscaled ERA5 rainfall data show a much better distribution of spatial covariance and temporal variance when compared with observation. This study is the first step towards developing deep learning-based weather data downscaling model for Indian summer monsoon rainfall data.



中文翻译:

基于深度学习的印度地区夏季季风降雨数据的缩减

缩小规模对于生成高分辨率观测数据是必要的,以验证气候模型预测或在微区域级别上监控降雨。由自动气象站或气象台产生的可用观测值通常在空间分辨率上受到限制,从而导致这些级别的降雨信息显示不正确或不正确。动态和统计缩减模型通常用于在较大域上获取高分辨率网格数据的信息。由于降雨的可变性取决于导致非线性或混沌的时空变化的复杂时空过程,因此没有一种降尺度方法被认为是足够有效的。在以复杂的地形,准周期和非线性为主的领域中,基于深度学习(DL)的方法提供了一种有效的解决方案,可以将降雨数据按比例缩小以实现区域气候预测和高分辨率的实时降雨观测数据。在这项工作中,我们采用了从超分辨率卷积神经网络(SRCNN)方法派生的三种基于深度学习的算法。使用这些方法,印度气象局(IMD)的夏季季风季节数据和热带降雨测量任务(TRMM)数据集的分辨率下调至4倍。从基于深度学习的模型获得的高分辨率数据比线性插值提供更好的结果,分辨率最高可提高4倍。在此处采用的三种算法(即SRCNN,堆叠式SRCNN和DeepSD)中,基于DeepSD的降尺度可产生最佳的降雨幅度空间分布和最小均方根误差。因此,提倡使用DeepSD算法以备将来使用。我们发现振幅和强度降雨模式的空间不连续性是降水缩减的主要障碍。此外,我们将这些方法应用于模型数据的后处理,尤其是ERA5重新分析数据。与观测相比,缩减的ERA5降雨数据显示了空间协方差和时间方差的更好分布。这项研究是为印度夏季季风降雨数据开发基于深度学习的天气数据降尺度模型的第一步。我们发现振幅和强度降雨模式的空间不连续性是降水缩减的主要障碍。此外,我们将这些方法应用于模型数据的后处理,尤其是ERA5重新分析数据。与观测相比,缩减的ERA5降雨数据显示了空间协方差和时间方差的更好分布。这项研究是为印度夏季季风降雨数据开发基于深度学习的天气数据降尺度模型的第一步。我们发现振幅和强度降雨模式的空间不连续性是降水缩减的主要障碍。此外,我们将这些方法应用于模型数据的后处理,尤其是ERA5重新分析数据。与观测相比,缩减的ERA5降雨数据显示空间协方差和时间方差的分布要好得多。这项研究是为印度夏季季风降雨数据开发基于深度学习的天气数据降尺度模型的第一步。与观测相比,缩减的ERA5降雨数据显示空间协方差和时间方差的分布要好得多。这项研究是为印度夏季季风降雨数据开发基于深度学习的天气数据降尺度模型的第一步。与观测相比,缩减的ERA5降雨数据显示空间协方差和时间方差的分布要好得多。这项研究是为印度夏季季风降雨数据开发基于深度学习的天气数据降尺度模型的第一步。

更新日期:2021-01-02
down
wechat
bug