2020 Fiscal Year Annual Research Report
| Project/Area Number |
20J21462
|
|---|
| Research Institution | Tohoku University |
|---|
Principal Investigator |
王 心月 東北大学, 理学研究科, 特別研究員(DC1)
|
|---|
| Project Period (FY) |
2020-04-24 – 2023-03-31
|
| Keywords | Deep neural network / Cloud retrieval / Himawari-8 |
|---|
| Outline of Annual Research Achievements |
In the last year, we developed a cloud retrieval model with a deep neural network (DNN) algorithm. When comparing to the traditional physics-based models, the new breakthrough is that as an infrared method in nature, our new model extends the predictable cloud optical thickness to ~200 with an overall relative bias less than 20%. This new model can be realistically applied to severe weather monitoring and mesoscale convective system studies. This work has been published on Remote Sensing of Environment, which is a Top journal in the field of remote sensing. With the DNN retrieved cloud properties, we further conducted a case study on the cloud evolutions in offshore and inland mesoscale convective systems over the south China coastal area.
|
| Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
Currently the technical report of the newly-developed deep neural network model has been organized as a paper, which has been published on a top scientific journal named as Remote Sensing of Environment. We also worked to improve the model interfaces for easier batch-process of the input and output data.
|
| Strategy for Future Research Activity |
Next plan is to apply the data retrieved by our new model, and study the daily evolution of cloud properties over the south China coastal region.
|
