Precipitation-Nowcasting

This is an easy-to-understand implementation of ConvLSTM model(fisrt proposed by [Xinjian Shi et al.])(https://arxiv.org/abs/1506.04214https://arxiv.org/abs/1506.04214) in a real-world precipitation nowcasting problem with Pytorch. Here presents the guidance on how to run this project by yourself. Have fun!

DATA

Two open-sourced datasets are available for training and testing in this project.

1. A pre-masked radar datasets.(Included in the package)
2. Tianchi CNKI 2017 dataset（Provided by Shenzhen Meteorological Bureau）.This dataset is not included yet. However, You can download the datasets here

Getting Started

Prerequisites

Environment:

• Win10 or Win7
• Anaconda 3-5.1
• Python 3.6
• CUDA 8(BOTH 0.3.1 & 1.4.0) or CUDA10.1(ONLY FOR 1.4.0)

Installing

1. Install CUDA8(BOTH 0.3.1 & 1.4.0) or CUDA10.1(ONLY FOR 1.4.0)

2. Download and install Anaconda environment

3. Install an environment

   conda create -n project python=3.6 
   conda create -n old python=3.6 
  

4. Activate your new-built environemt and install Pytorch and torchvision (For Nowcasting)

   activate old 
   conda install -c peterjc123 pytorch (WIN10)
   conda install -c peterjc123 pytorch cuda80 (WIN7)
   pip install torchvision===0.2.1 -f https://download.pytorch.org/whl/torch_stable.html
   pip install requests
   pip install arrow
   pip install pillow===6.0.0
   pip install tqdm
   pip install colorama
   conda deactivate

5. Activate your new-built environemt and install Pytorch and torchvision (For Training)

   activate project
   pip install torch===1.4.0 torchvision===0.5.0 -f https://download.pytorch.org/whl/torch_stable.html
   pip install requests
   pip install arrow
   pip install pillow
   pip install tqdm
   pip install colorama
   conda deactivate

Train the model

1. Download the all package and unpack it
   Note: you also need to unpack the files in the original data directory before training

2. Train the model

  Python training.py

Running the test

Run the test.py with the command.

  python test.py  

Evaluate your model's performance by running

  python evaluate.py

Authors

 cxxixi
 pqx
 Amy Hsiao
 Thomas Liu 
 Garry Lai

Notes

1. Notes on ConvLSTM

Lnk

%windir%\System32\cmd.exe "/K" C:\ProgramData\Anaconda3\Scripts\activate.bat C:\ProgramData\Anaconda3 && call activate project && f: && cd Precipitation-Nowcasting

%windir%\System32\cmd.exe "/K" C:\ProgramData\Anaconda3\Scripts\activate.bat C:\ProgramData\Anaconda3 && call activate old && f: && cd Precipitation-Nowcasting