High-Resolution Land-Use and Land-Cover Map of Japan 2024
(Released in February 2026 / 2024JPN_v26.02_orchard)

Scientific Paper Under Review: A research paper describing this dataset and methodology is currently to be published (peer-review ongoing). The full title and DOI link will be provided here shortly.

Interactive Map Preview

You can switch background maps and toggle LULC visibility using the layer panel in the top-right corner.

Legend:

Download link

2024JPN_v26.02_orchard: 27 categories

Download (.zip)

Category Definitions

Definition of Categories

Figure 1: Definition and number of reference data for each category. Note that the original category “Cropland” in JAXA HRLULC-Japan has been renamed to “Herbaceous cropland” in this study, as woody crops (orchards) are segregated and classified as independent categories.

Classification Workflow

Classification Workflow

Figure 2: Workflow of the hierarchical classification in this study. Note that the 1st and 2nd level classifications are executed sequentially as independent processes, each encompassing classification, ensemble learning, and post-processing. The 2nd level classification is performed exclusively for the pixels classified as "Deciduous orchard" in the finalized 1st level LULC map. Both levels utilize the same input data and the same classifier SACLASS 2.5.

Input Data

Input Data Sources

Figure 3: List of input data.

Classification Algorithm (SACLASS2.5)

Algorithm Framework

Figure 4: Processing flow of SACLASS2.5 developed by JAXA.

Accuracy Assessment

  • 1st step classification: 93.2 ± 0.5 %
  • 2nd step classification: 85.2 ± 3.5 %
1st Level Matrix

Figure 5a: Confusion matrix for the 1st step classification (Major categories). Error bars represent the 95 % confidence interval.

2nd Level Matrix

Figure 5b: Confusion matrix for the 2nd step classification focusing on detailed orchard categories. Error bars represent the 95 % confidence interval.

Regional Classification Results

Regional Comparisons

Figure 6: Comparison of JAXA HRLULC-Japan 2024JPN_v25.04 and the output of this study in major fruit-growing regions, along with true color Sentinel-2/MSI image (median composite of July–August 2024). JAXA HRLULC-Japan 2024JPN_v25.04 consists of 15 categories (Category ID: 1 – 14, 20), while the output of this study consists of 27 categories (Category ID: 1 – 14, 16– 28).