# Landscape Change Monitoring System

USDA — version 2024.10

## Description

This dataset maps annual land cover, land use, and landscape change across the conterminous United States at 30 m spatial resolution. Three thematic variables are provided: land cover, land use, and change. An additional quality assessment variable is included as a packed bit string.

The dataset is produced using the LandTrendr time-series segmentation algorithm applied to Landsat and Sentinel-2 imagery. The change variable identifies the cause of short- and long-term vegetation disturbance events, as well as water desiccation or inundation and snow/ice transition, using a ruleset approach that incorporates ancillary datasets such as burn severity records, insect and disease surveys, and storm event data.

## Summary

- **ID:** 446d2f29-0d7a-48a3-8407-e84d43df909b
- **Type:** Raster
- **CRS:** See description
- **Spatial coverage:** Conterminous US
- **Spatial resolution:** 30 m
- **Temporal coverage:** 1985+
- **Temporal resolution:** Annual

## Usage notes

New time points are delivered approximately annually. Spatial coverage is limited to the conterminous United States. This dataset is best suited for regional to national scale assessments, including analysis of vegetation cover, land cover, or land use change trends, total extents, and aggregated summaries.

No minimum mapping unit has been applied and known features such as roads have not been manually incorporated. Users should exercise caution when applying this data at site or fine-scale levels.

## Variables

### change
- **Type:** uint8
- **NoData:** 0
Index of the predicted landscape change class describing the cause of short- or long-term vegetation disturbance events, as well as water desiccation or inundation and snow/ice transition. Predictions are made using a machine learning model whose outputs (slow loss, fast loss, gain) are reclassified using a ruleset incorporating ancillary fire, storm, insect, disease, and land cover datasets.

### land_cover
- **Type:** uint8
- **NoData:** 0
Index of the predicted land cover class describing biophysical attributes of the land surface, including vegetation type and other surface conditions. Predictions are made using a machine learning model trained on Landsat and Sentinel-2 time-series data.

### land_use
- **Type:** uint8
- **NoData:** 0
Index of the predicted land use class describing the intended human use of the land, representing the economic and cultural activities practised at a given place. Predictions are made using a machine learning model trained on Landsat and Sentinel-2 time-series data, with class assignments further refined using ancillary land cover and built-up datasets.

### qa_bits
- **Type:** uint16
- **NoData:** 65535
Quality assessment data provided as a packed bit string. Contains ancillary information on the origin of annual pixel values, including whether an interpolated value was used, which sensor the pixel came from, and the Julian day of the year the LandTrendr pixel value was derived from.

## Pricing

This dataset has no data acquisition cost.