README.md.orig

# Radar Forecast Model – Deep Learning

## Long Short-term Memory
LSTM is good at processing time series data.

The model structure can be decomposed by two networks, one is encoder, and the other is predictor.


Model configurations:

Model Name|hidden layers|loss function|# training samples|tsize|# epochs|training process|Date
--------|--------|----------|----------|-------|--------------|--------|------|
newest-5_8| 16 |built-in RSME| 400| 10| 500| Fig.1|2019.5.8
newest-5-13|16|RMSE+FAR|400|10|500|Fig.2|2019.5.13

<p align="center">
<img src="images/training-5_8.PNG" style="width:60%"><br>
Fig.1 Training loss on 5.8
<img src="images/201811230600-predict.gif" style="width: 80%"><br>
Fig.2 Comparison of predction and observation
</p>

### Wind Effect
To add wind effect in, one way is to customize the loss function by comparing LSTM modeled image at t and predicted image with Lagrangian plus wind at t. In this way, we drive the LSTM to the correct advection direction.

## GAN

<<<<<<< HEAD

## GRU

## Optical Flow

## Kalman Filter
=======
In the GAN repository, we implement generative adversorial neural network to train the model. The reason behind is that, we hope GAN could help us to discriminate the input data.

## Semi-Lagrangian movement

## Optical Flow

We select one event from our radar data shown below
<p align="center">
    <img src="demo.gif" style="width:60%">
</p>

## Kalman Filter 
>>>>>>> 6c7a9f25cbe0e8611e7e7b7b6034746961336007


To be continued...