Annual Crop & Land Cover Land Use (CLCLU) Dataset for the Middle Rio Grande (MRG) Region (1994–2024)

This dataset presents the first long-term, scalable, and data-driven Crop and Land Cover Land Use (CLCLU) classification product for both the U.S. and Mexican sides of the Middle Rio Grande (MRG) region. Covering the years 1994 to 2024, it was developed using multi-temporal optical satellite imagery, with crop-specific annotations and expert-informed land use categories to support transboundary environmental monitoring and agricultural policy efforts.

Dual-month median composites (July & December) derived from Landsat 5 and 8 imagery were utilized as inputs for a semantic segmentation model, the Multi-Attention Network (MANet), employing a ResNeXt-101 encoder. The Crop Data Layer (CDL) served as the reference ground truth for model training and evaluation.

To assess the accuracy and reliability of the dataset, cross-validation was performed against the National Land Cover Dataset (NLCD) and the MODIS Land Cover Product (MCD12Q1-UMD).

The dataset maintains consistent 30-meter spatial resolution and annual temporal granularity. The processing workflow includes data preprocessing, training and validation dataset construction, semantic segmentation, accuracy assessment, and inter-comparison with benchmark products.

Each annual classified map consists of pixel-level land cover and crop type assignments. The pixel values correspond to the following CLCLU classes:

Pixel Value
Class Label

0
Alfalfa/Hay

1
Cotton

2
Pecan

3
Othercrops

4
Forest/Shrublands

5
Grassland/Barren

6
Water bodies

7
Developed

8
Other/ background

Key Features:

Temporal Coverage: 1994–2024

Spatial Resolution: 30 meters

Geographic Scope: U.S.–Mexico transboundary region along the Middle Rio Grande

Data Type: Annual raster classification maps (GeoTIFF format)

Use Cases: Agricultural monitoring, land use/land cover change analysis, transboundary resource planning, AI-based remote sensing applications