Datasets:

Modalities:
Image
ArXiv:
Libraries:
Datasets
License:
amitmisra commited on
Commit
b3fb022
1 Parent(s): 55bbe9a

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +56 -49
README.md CHANGED
@@ -1,52 +1,59 @@
1
  ---
2
  license: mit
3
  ---
4
-
5
- # Overview
6
-
7
- Here you will find the model output parquet files and geotiffs from the following paper:
8
- <placeholder for arxiv link>
9
-
10
- Each file is for a 3 degree by 3 degree tile. The folder structure is as follows:
11
- - Main directory by latitude
12
- - Subdirectory for each tile
13
- - Each subdirectory contains 3 files:
14
- - parquet files with tabular data
15
- - geotiff with 90 meter buffer applied on filtered dataset (see paper for details)
16
- - same as above but with 240 meter buffer applied
17
-
18
- The folder name and file names denote the southwest corner of the tile. For example, The file N27E030 would be for latitudes between 27 and 30 degrees north, and longitudes between 30 and 33 degrees east. Its files would be in the N27 folder.
19
-
20
- # Schema
21
-
22
- The schema for each file is as follows:
23
- - year: year of the observation (int)
24
- - month: month of the observation (int)
25
- - day: day of the month of the observation (int)
26
- - lat: latitude of the potential flood detection (center of pixel) (float)
27
- - lon: longitude of the potential flood detection (center of pixel) (float)
28
- - filename: Sentinel-1 file name corresponding to observation in which flooding was detected (str)
29
- - land_cover: Land cover mapping from ESA WorldCover for the pixel (int)
30
- - dem_metric_1: slope calculated for the individual pixel (float)
31
- - dem_metric_2: maximum slope within 240m of the individual pixel (float)
32
- - soil_moisture: absolute soil moisture as estimated by the LPRM algorithm, units in percentage points (float)
33
- - soil_moisture_zscore: soil moisture anamoly relative to the period of Oct 2014-Sep 2024 (float)
34
- - soil_moisture_sca: absolute soil moisture as estimated by the SCA algorithm, units in percentage points (float)
35
- - soil_moisture_sca_zscore: soil moisture sca anamoly relative to the period of Oct 2014-Sep 2024 (float)
36
- - temp: average daily minimum temperature for the day of the year of the observation, under current climate conditions (1991-2020) (float)
37
- - edge_false_positives: were any edge effects (near the edge of the image) detected for this observation (int)
38
-
39
- # Suggested Filters
40
-
41
- Here are the filters we typically apply in post processing, as described in the paper:
42
- - dem_metric_2 < 10
43
- - soil_moisture_sca > 1
44
- - soil_moisture_zscore > 1
45
- - soil_moisture > 20
46
- - temp > 0
47
- - land_cover: exclude values of 60 for bare ground
48
- - edge_false_positives = 0
49
-
50
- # Spatial Resolution
51
-
52
- The data is run on input data with 20 meter resolution (Status: Oct 31 2024: Africa and South America are available at 20 meters, rest of world is at 30 meters with update expected later this year)
 
 
 
 
 
 
 
 
1
  ---
2
  license: mit
3
  ---
4
+ # Flood Detection Dataset
5
+
6
+ ## Overview
7
+ This repository contains model output parquet files and geotiffs from our flood detection research. The data is organized in 3° × 3° tiles, with each tile containing both tabular data and processed geotiff images.
8
+
9
+ ## Directory Structure
10
+ ```
11
+ ├── N27 # Main directory by latitude
12
+ │ ├── N27E030 # Subdirectory for specific tile
13
+ │ │ ├── N27E030.parquet # Tabular data
14
+ │ │ ├── N27E030_90m.tif # Geotiff with 90m buffer
15
+ │ │ └── N27E030_240m.tif # Geotiff with 240m buffer
16
+ ```
17
+
18
+ The naming convention uses the southwest corner coordinates of each tile. For example, `N27E030` represents:
19
+ - Latitude: 27°N to 30°N
20
+ - Longitude: 30°E to 33°E
21
+
22
+ ## Schema
23
+ | Column | Type | Description |
24
+ |--------|------|-------------|
25
+ | year | int | Year of observation |
26
+ | month | int | Month of observation |
27
+ | day | int | Day of observation |
28
+ | lat | float | Latitude of potential flood detection (pixel center) |
29
+ | lon | float | Longitude of potential flood detection (pixel center) |
30
+ | filename | str | Corresponding Sentinel-1 file name |
31
+ | land_cover | int | ESA WorldCover land cover classification |
32
+ | dem_metric_1 | float | Individual pixel slope |
33
+ | dem_metric_2 | float | Maximum slope within 240m |
34
+ | soil_moisture | float | LPRM algorithm absolute soil moisture (%) |
35
+ | soil_moisture_zscore | float | Soil moisture anomaly (Oct 2014-Sep 2024 baseline) |
36
+ | soil_moisture_sca | float | SCA algorithm absolute soil moisture (%) |
37
+ | soil_moisture_sca_zscore | float | SCA soil moisture anomaly (Oct 2014-Sep 2024 baseline) |
38
+ | temp | float | Average daily minimum temperature (1991-2020 climate) |
39
+ | edge_false_positives | int | Edge effect detection flag |
40
+
41
+ ## Recommended Filtering
42
+ For optimal results, we recommend applying the following filters:
43
+
44
+ ```python
45
+ filters = {
46
+ 'dem_metric_2': '< 10',
47
+ 'soil_moisture_sca': '> 1',
48
+ 'soil_moisture_zscore': '> 1',
49
+ 'soil_moisture': '> 20',
50
+ 'temp': '> 0',
51
+ 'land_cover': '!= 60', # exclude bare ground
52
+ 'edge_false_positives': '= 0'
53
+ }
54
+ ```
55
+
56
+ ## Spatial Resolution
57
+ - Current Status (Oct 31, 2024):
58
+ - Africa and South America: 20-meter resolution
59
+ - Rest of world: 30-meter resolution (20-meter update expected later this year)