Land Cover Estimates for the Kenai Peninsula Lowlands, 1973-2017

Circa 1973, 2002, and 2017 raster images of estimated land cover types for the western Kenai Peninsula lowlands. The data includes nine land cover classes derived from a random forest classifier using Landsat imagery and topographic data, with 3,600 training points per era. The dataset was created by Carson A. Baughman of the United States Geological Survey.

Use Cases

• Analyzing land cover change over time based on the three temporal snapshots (1973, 2002, 2017).
• Training or validating remote sensing classification models based on the nine land cover types.
• Studying the impact of environmental factors on forest and wetland distribution based on the described predictor variables.

Strengths

• Includes three distinct time points (circa 1973, 2002, 2017) for temporal analysis.
• Uses a substantial training set of 3,600 points per era (400 per land cover type).
• Derived from a defined methodology using a random forest classifier with Landsat and DEM data.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file size are unknown, which may limit suitability assessment.
• The 'circa' dates and opportunistic training area distribution may introduce temporal and spatial bias.

Provenance

Source

United States Geological Survey

Collection Method

Land cover types were derived from a random forest classifier executed in R, using predictor variables from Landsat mosaics, a digital elevation model, and training points manually interpreted from high-resolution aerial imagery.

Time Range

circa 1973, circa 2002, circa 2017

Geography

Western Kenai Peninsula lowlands, Alaska, USA