撂荒耕地信息获取方法研究进展与展望

撂荒作为耕地利用边际化的最终表现形式,是中国近20 a以来耕地利用亟需解决的重要问题之一,而有效获取撂荒耕地信息是探究撂荒耕地时空变化与影响因素的基础前提,也是政府进行耕地政策调整,保障耕地资源可持续利用的重要依据。该文应用文献综述法与归纳总结法,基于国内外学者在撂荒耕地信息获取研究上的最新成果,对撂荒耕地信息获取方法进行归纳总结,并对未来研究方向进行了展望。结果表明：1）综合前人研究成果,撂荒耕地信息获取方法可以分为基于抽样调查、文献荟萃分析以及遥感获取的3种类型。2）基于抽样调查的撂荒耕地信息获取方法在案例研究上应用较广,且有着近乎一致的研究范式,但数据空间表征能力较弱。全国家庭调查数据在一定程度上增加了数据的空间属性,但对原始数据的二次筛选减少了样本容量,降低了数据的可信性。3）文献荟萃法基于"二手"的文献或数据,研究结果受已公开发表文章数量的限制,且需要研究人员对相关领域热点关键词有全面了解,目前该方法在撂荒耕地信息获取应用上相对较少。4）遥感卫星与计算机技术发展给撂荒耕地信息获取提供了便利,基于地物特征规则与土地利用信息变化,多种耕地撂荒检测方法被开发,但受限于高空间分辨率与大范围提取需求,仍存在较大的优化空间,未来研究可在数据选择与处理、特征和方法融合上进行更深一步的探索。

Progress and prospects on information acquisition methods of abandoned farmland

A farmland served as the carrier of grain production is an important material basis to the food security. The abandonment of farmland has become one of the most important problems in China''s farmland utilization over the past 20 years, particularly on a series of social, economic, and environmental issues. Therefore, it is necessary to quickly capture the spatial and temporal information, including the accurate location, duration, and the scale of abandoned farmland, in order to evaluate the marginal trend, and further to maintain the sustainable use of farmland. According to the current status of information acquisition methods for the abandoned farmland, this study was first to summarize the advanced methods of acquiring abandoned farmland information, and then to make a comprehensive prospect for the directions of future research. The results show that: 1) Three types can be divided in the methods of information acquisition: sampling survey, literature meta-analysis, and remote sensing. 2) In a sampling survey, the information acquisition method of abandoned farmland was widely used in small-scale case studies, indicating a nearly consistent research paradigm, but the spatial representation ability was relatively weak. To a certain extent, the national household survey data increased the spatial attribute of data, but the secondary screening of original data reduced the sample size and the credibility. Moreover, most panel data has made it difficult to trace the abandoned farmland information in different historical stages. 3) In literature meta-analysis, the information statistics method of abandoned farmland applied the idea of big data analysis, and further integrated the previous research data. It can be used to not only represent the abandonment rate and spatial-temporal changes of different regions, but also compare the main driving factors of abandoned farmland in different regions. However, the research findings of this method were confined to only a few published articles, while, a comprehensive understanding of hot keywords was highly demanding in related fields. At present, only a relatively few applications were found in the acquisition of abandoned farmland information. 4) Remote sensing can be expected to the mainstream for the information acquisition of abandoned farmland, as the development of satellites and computer technologies in the future. A variety of detection methods for the abandoned farmland have been developed, according to the object features and information changes of land use. There was still a large optimization space in the remote sensing, due to the high spatial resolution and large-scale extraction requirements. With the emergence of cloud computing and machine learning, the remote sensing has the promising potential to explore large-scale and high-resolution detection of abandoned farmland. Therefore, future research can be further explored in data selection and processing, feature fusion, and method fusion. The findings can provide an appropriate reference for the detection and management of farmland use.