Data for the paper titled "Investigation of the spatio-temporal transferability of large-scale random forest crop classification models and potential influencing factors"

This dataset comprises the training and validation data used in the paper titled "Investigation of the spatio-temporal transferability of large-scale random forest crop classification models and potential influencing factors". It consists of 21 CSV files. The training data are files prefixed with "training_", containing training samples from Kansas, South Dakota, and Minnesota for the years 2019 to 2023, sourced from CDL. The validation data, identified by filenames starting with "LUCAS_", comprise validation samples from France, Germany, Italy, Poland, Romania, and Spain for the year 2018, sourced from LUCAS. In each CSV file, the scales for band, precipitation, and temperature data are 0.001, 0.001, and 0.01, respectively. Band/index names prefixed with the digit i denote the (i+1)th month's band/index value. Numeric suffixes in precipitation and temperature field names denote the month. The field named "cropland" represents crop type labels, where 1 indicates Corn and 20 indicates Wheat.<br>Shuai Xie, Naijie Zhang, Liangyun Liu &amp; Lin Sun (2025) Investigation of the spatio-temporal transferability of large-scale random forest crop classification models and potential influencing factors, International Journal of Digital Earth, 18:2, 2563211, https://doi.org/10.1080/17538947.2025.2563211