Dataset and results for "Comparing machine learning and deep learning models for probabilistic post-processing of satellite precipitation-driven streamflow simulation"

Dataset and results for "Comparing machine learning and deep learning models for probabilistic post-processing of satellite precipitation-driven streamflow simulation"

Yuhang Zhang1, Aizhong Ye1*, Phu Nguyen2, Bita Analui2, Soroosh Sorooshian2, Kuolin Hsu2

1 State Key Laboratory of Earth Surface Processes and Resource Ecology, Faculty of Geographical Science, Beijing Normal University, Beijing 100875, China.

2 Center for Hydrometeorology and Remote Sensing, Department of Civil and Environmental Engineering, University of California, Irvine, Irvine, California, CA 92697, USA.

## Dataset    

Streamflow simulations from one observed precipitation (CMA) and three satellite precipitation products (PDIR, IMERG-F, and GSMaP) for 522 sub-basins.

- Q-CMA (streamflow reference)
- Q-PDIR (uncorrected)
- Q-IMERGF (uncorrected)
- Q-GSMAP (uncorrected)

### Data structure

- Head section (row1-row5)
  - SubNO:    522 
  - BeginT:    2003-01-01 00:00 
  - EndT:    2019-12-31 00:00 
  - Interval:    1440s (daily)
  - Revise:    10 (scaling factor to keep int datatype)
  - Point1    Point2    ... (Subbasin No.)
- Data section
  - 6209 rows, 522 cols

## Results

Two post-processing model results for test period (2015-1-1 to 2018-12-31).

### Data structure

- 1462 rows, every row denotes each day from 2015-1-1 to 2018-12-31

- 100 columns, every column denotes each quantile from 0.005 to 0.995, total 100 quantiles.

### qrf-output

- pdir (single input)
- imergf (single input)
- gsmap (single input)
- all (multiple inputs)

### lstm-output

- pdir (single input)
- imergf (single input)
- gsmap (single input)
- all (multiple inputs)