Historical Rice Field Maps of Japan
Description
This dataset provides rice field maps of Japan from the 1980s. The rice field raster layers were created using Landsat images, by combining temporal aggregation and a phenology-base algorithm in Google Earth Engine. Each layer represents an averaged representation of the distribution of rice fields over a period of 5 years. Methods for creating the rice layers are fully described in the associated publication: Carrasco et. al. 2022. Historical mapping of rice fields in Japan using phenology and temporally aggregated Landsat images in Google Earth Engine. (Under review). The "Raw-maps" folder contains the seven rice field binary maps (seven different time periods; 1 = rice fields; 0 = non-rice fields) described in the associated publication. The "Post-processed-maps" folder contains the seven rice field maps with additional post-processing (1 = rice fields; Non-rice fields are set to NA values). This post-processing consists on (1) water-bodies masking, and (2) masking of Hokkaido municipalities where rice was not present during the last two decades. As discussed in the associated publication, rice fields were overestimated in Hokkaido. For this reason, by masking out rice fields in areas where rice fields were not present in recent times we can improve the accuracy of the mapping for this area. For this masking, we used official (field-based) data on rice field area of Hokkaido municipalities provided by the Ministry of Agriculture, Forestry, and Fisheries of Japan. Unfortunately, those data were not available at the municipality level for all prefectures. Each raster file is in geographic coordinates and has an original spatial resolution of 30 m. The accuracy of the maps is presented in the associated publication. We suggest users to check the associated documentation and to understand the uncertainties and limitations of this dataset.
Files
Steps to reproduce
These maps can be reproduced by following the methodology in "Carrasco et. al. 2022. Historical mapping of rice fields in Japan using phenology and temporally aggregated Landsat images in Google Earth Engine. (Under review)." Auxiliary input data is also provided as supplementary material in this publication. Additionally, we provide the code used to create the maps provided in the "Raw-maps" folder, together with the publication. Also available here: https://github.com/luiscartor/rice-mapping-japan Methods summary: We combined temporally aggregated Landsat images over five-year periods with a phenology algorithm to create rice field maps of Japan since the 1980s at a national scale. After pre-processing Landsat images, we extracted several spectral indices and performed the temporal aggregation for seven five-year periods. We then used rice transplantationdates, surface air temperature, and other ancillary data to determine rice transplantingseasons and other vegetation growing seasons at a prefecture level. This allowed us to identify several land covers and to classify rice fields using a pixel-based, phenology algorithm. We based our algorithm on Dong et al.’s (2015) Landsat-RICE algorithm,adapting it to be used with temporally aggregated Landsat data and at a national scalewhere transplantation dates and temperature vary greatly. Separate maps were therefore created for each of Japan’s prefectures using prefectural-level phenological parameters.These were merged to create a country-level rice field map for each of the seven five-year periods.