Hugging Face's logo

Datasets:
ibm-nasa-geospatial
/
multi-temporal-crop-classification

Modalities:
Image
Text
Formats:
text
Languages:
English
Size:
1K - 10K
Tags:
remote sensing
segmentation
crop type
foundation model
DOI:
Libraries:
Datasets
License:
Dataset card Viewer Files Community
2
multi-temporal-crop-classification / README.md
HamedAlemo's picture
HamedAlemo
update README
8e3da83
preview code
|
raw
history blame
No virus
2.89 kB
metadata
license: cc-by-4.0
language:
  - en
tags:
  - remote sensing
  - segmentation
  - crop type
  - foundation model
size_categories:
  - 1K<n<10K

Dataset Card for Multi-Temporal Crop Classification

Dataset Description

Dataset Summary

This dataset contains temporal Harmonized Landsat-Sentinel imagery of diverse land cover and crop type classes across the Contiguous United States for the year 2022. The target labels are derived from USDA's Crop Data Layer (CDL). It's primary purpose is for training segmentation geospatial machine learning models.

Dataset Structure

TIFF Files

Each tiff file covers a 224 x 224 pixel area at 30m spatial resolution. Each input satellite file contains 18 bands including 6 spectral bands for three time steps stacked together. Each GeoTIFF file for the mask contains one band with the target classes for each pixel.

Band Order

In each input GeoTIFF the following bands are repeated three times for three observations throughout the growing season: Channel, Name, HLS S30 Band number
1, Blue, B02
2, Green, B03
3, Red, B04
4, NIR, B8A
5, SW 1, B11
6, SW 2, B12

Masks are a single band with values:
0 : "No Data" 1 : "Natural Vegetation" 2 : "Forest" 3 : "Corn" 4 : "Soybeans" 5 : "Wetlands" 6 : "Developed/Barren" 7 : "Open Water" 8 : "Winter Wheat" 9 : "Alfalfa" 10 : "Fallow/Idle Cropland" 11 : "Cotton" 12 : "Sorghum" 13 : "Other"

Class Distribution

Training Data Distribution

Training Data

Validation Data Distribution

Validation Data

Data Splits

The 3,854 chips have been randomly split into training (80%) and validation (20%) and corresponding ids recorded in cvs files train_data.txt and validation_data.txt.

Dataset Creation

Query and Scene Selection

First, a set of 5,000 chips are defined based on samples from the USDA CDL to ensure a representative sampling across the CONUS. Next, for each chip the corresponding HLS S30 scenes between March and September 2022 are queried and scenes with low cloud cover are retrieved. Then, three scenes are selected among the low cloudy scenes to ensure we get a scene early in the season, one in the middle, and one toward the end. The three final scenes are then reprojected to CDL's projection grid (EPSG:5070) using bilinear interpolation.

Chip Generation

In the final step, the three scenes for each chip are clipped to the bounding box of the chip, and 18 spectral bands are stacked together. In addition, a quality control is applied to each chip using the Fmask layer of HLS dataset. Any chip containing clouds, cloud shadow, adjacent to cloud or missing values is discarded. This resulted in 3,854 chips.