Commit
•
ee4a07d
1
Parent(s):
abb7bde
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,5 +1,3 @@
|
|
| 1 |
-
goes, geospatial, computer vision, forecasting, nowcasting, radar, precipitation
|
| 2 |
-
|
| 3 |
# Dataset Card for Goes-MRMS
|
| 4 |
|
| 5 |
## Table of Contents
|
|
@@ -35,7 +33,9 @@ goes, geospatial, computer vision, forecasting, nowcasting, radar, precipitation
|
|
| 35 |
|
| 36 |
### Dataset Summary
|
| 37 |
|
| 38 |
-
This dataset is a combination of GOES-16 data and MRMS radar precipitation data to roughly match the unreleased dataset used to train Google Research's MetNet. In the
|
|
|
|
|
|
|
| 39 |
|
| 40 |
### Supported Tasks and Leaderboards
|
| 41 |
|
|
@@ -57,9 +57,9 @@ This dataset is a combination of GOES-16 data and MRMS radar precipitation data
|
|
| 57 |
|
| 58 |
### Data Splits
|
| 59 |
|
| 60 |
-
MetNet (January 2018-July 2019)
|
| 61 |
-
MetNet-2 (July 2017-August 2020)
|
| 62 |
-
Full (July 2017-January 2022)
|
| 63 |
|
| 64 |
## Dataset Creation
|
| 65 |
|
|
|
|
|
|
|
|
|
|
| 1 |
# Dataset Card for Goes-MRMS
|
| 2 |
|
| 3 |
## Table of Contents
|
|
|
|
| 33 |
|
| 34 |
### Dataset Summary
|
| 35 |
|
| 36 |
+
This dataset is a combination of GOES-16 data and MRMS radar precipitation data to roughly match the unreleased dataset used to train Google Research's MetNet. In the papers they used GOES-16 satellite imagery, MultiRadar/Multi-System (MRMS) instantaneous precipitation, hourly cumulative precipitation, and High Resolution Rapid Refresh NWP initializations as inputs to predict future MRMS precipitation rates. The precipitation rates were binned into 0.2mm/hr bins to make the output a classification task, and allow for the models to predict a probability distribution over the region of interest.
|
| 37 |
+
|
| 38 |
+
Additionally, the input image patches are much larger than the target image patches. For MetNet, the input images covered 512x512 km area, while the target was the center 64x64 km crop. For MetNet-2 the input covered 2048x2048 km with the target being the central 512x512 km.
|
| 39 |
|
| 40 |
### Supported Tasks and Leaderboards
|
| 41 |
|
|
|
|
| 57 |
|
| 58 |
### Data Splits
|
| 59 |
|
| 60 |
+
MetNet (January 2018-July 2019) (16 days training, 2 days validation, 2 days test)
|
| 61 |
+
MetNet-2 (July 2017-August 2020) (Non-overlapping time ranges with 12 hour black outs in between)
|
| 62 |
+
Full (July 2017-January 2022) (Train: 2017-2020. except for first of the month, Validation: first of the month July 2017-2020, Test: 2021-2022)
|
| 63 |
|
| 64 |
## Dataset Creation
|
| 65 |
|