Understanding the spatiotemporal patterns of seasonal, annual, and consecutive farmland abandonment in China with time-series MODIS images during the period 2005-2019

Farmland abandonment has substantial impacts on the environment, and its impacts vary with the duration of farmland abandonment. While consecutive and annual farmland abandonment (AFA) has been extensively studied, there is a need to investigate the spatiotemporal variation in seasonal farmland abandonment (SFA), particularly in areas where multiple-cropping practices are applied. In this study, we propose a method to monitor different types of farmland abandonment, such as SFA, AFA, and consecutive farmland abandonment (CFA), with time-series remote sensing data. The spatiotemporal pattern of farmland abandonment in mainland China from 2005 to 2019 was derived from time-series MODIS images. The overall accuracy for AFA maps was 79%-86% by comparing with MCD12Q1, and around 70% by comparing with China's Land Use/Cover Datasets. The results indicated that the total abandoned cultivable lands accounted for nearly 20% of the total cultivable lands and that SFA accounted for over 50% of the total abandoned cultivable lands in China. The total abandoned cultivable lands and total abandoned cultivable areas in mainland China did not change significantly across the study period. Moreover, the frequency of SFA gradually increased during the study period, and the frequency of CFA gradually decreased. Our study highlights that SFA accounts for a considerable proportion of the abandoned cultivable land in mainland China. SFA should be considered in related studies, as it could be a dominant type of farmland abandonment in areas where multiple cropping is practiced.