English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

. Eastern China is one of the most economically developed anddensely populated areas in the world. Due to its special geographicallocation and climate, eastern China is affected by different weather systems,such as monsoons, shear lines, typhoons, and extratropical cyclones. In thenear future, the rainfall rate becomes difficult to predict precisely due tothese systems. Traditional physics-based methods such as numerical weatherprediction (NWP) tend to perform poorly on nowcasting problems due to thespin-up issue. Moreover, various meteorological stations are distributed inthis region, generating a large amount of observation data every day, whichhave great potential for application to data-driven methods. Thus, it isimportant to train a data-driven model from scratch that is suitable for thespecific weather situation of eastern China. However, due to the high degreesof freedom and nonlinearity of machine learning algorithms, it is difficultto add physical constraints. Therefore, with the intention of using variouskinds of data as a proxy for physical constraints, we collected three kindsof data (radar, satellite, and precipitation data) in the flood season from2017 to 2018 in this area and preprocessed them into tensors (256×256) that cover eastern China with a domain of 12.8×12.8∘.The developed multisource data model (MSDM) combines the optical flow,random forest, and convolutional neural network (CNN) algorithms. It treatsthe precipitation nowcasting task as an image-to-image problem, which takesradar and satellite data with an interval of 30 min as inputs andpredicts radar echo intensity with a lead time of 30 min. To reduce thesmoothing caused by convolutions, we use the optical flow algorithm topredict satellite data in the following 120 min. The predicted radarechoes from the MSDM together with satellite data from the optical flowalgorithm are recursively implemented in the MSDM to achieve a 120 minlead time. The MSDM predictions are comparable to those of other baselinemodels with a high temporal resolution of 6 min. To solve blurry imageproblems, we applied a modified structural similarity (SSIM) index as a lossfunction. Furthermore, we use the random forest algorithm with predictedradar and satellite data to estimate the rainfall rate, and the resultsoutperform those of the traditional, nonlinear radar reflectivity factor andrainfall rate (Z–R) relationships that use logarithmic functions. Theexperiments confirm that machine learning with multisource data providesmore reasonable predictions and reveals a better nonlinear relationshipbetween radar echo and precipitation rate. Apart from developing complicatedmachine learning algorithms, exploiting the potential of multisource datawill yield more improvements.

MSDM v1.0: A machine learning model for precipitation nowcasting over eastern China using multisource data