耕地复种指数遥感监测研究进展

复种指数是进行粮食估产、耕地集约利用评价、农业生态系统模拟等的关键参数,及时、准确地提取复种指数对于粮食安全、土地管理和生态环境安全具有重要意义。在传统的研究中,复种指数主要来源于地面统计数据。使用统计数据来计算复种指数虽然过程简单,但是计算结果存在信息滞后、无法体现统计单元内部的空间异质性、精度低等不足。遥感技术因具有大范围、高时效、低成本等优点而被用于耕地复种指数监测,已有学者对耕地复种指数的遥感监测开展了大量工作。本文以复种指数遥感提取的关键环节为主线,对1997—2020年国内外相关研究进行综述：首先,梳理了已有研究中的监测方法、高质量时间序列遥感数据获取方法及提取结果精度验证方法,并对不同方法的优缺点进行了总结;其次,对已有研究中存在的不足进行了探讨,并提出未来研究的侧重点：① 开展已有监测方法的对比和分析;② 加强地形复杂地区、小农尺度的监测力度;③ 提高遥感数据时空分辨率及处理效率;④ 对提取结果进行多尺度验证。

Research Progress on Remote Sensing Monitoring of Cultivated Land Cropping Intensity

Cropping intensity refers to the frequency of crop planting in the same cultivated land in one year. It is a key parameter for grain yield estimation, land use intensity evaluation, and agroecosystem modeling. Understanding the spatiotemporal change of cropping intensity provides support for food security, land management, and eco-environment security. The demand for timely and accurate information of cropping intensity is expected to increase in the future. Traditionally, cropping intensity is calculated from the statistical data. However, there are several shortcomings in the output from statistics, such as time-lag effect, homogeneity in one administrative unit, and low accuracy. In the past two decades, remote sensing technology has been widely used to monitor cropping intensity at different scales due to its multiple advantages such as high-efficiency and low-cost. However, the performance of remote sensing in monitoring cropping intensity has not been well evaluated. In this paper, remote sensing data, extraction algorithms, and accuracy evaluation methods employed in cropping intensity researches are summarized elaborately: ① cropping intensity extraction algorithms can be grouped into the following types: feature discrimination algorithm, curve feature comparison algorithm, peak detection algorithm, temporal mixture analysis algorithm, hierarchical training algorithm, continuous wavelet transform algorithm, growth cycle judgment algorithm, and time-series banning algorithm; ② time-series data from a single sensor is still the main data source for cropping intensity monitoring and can no longer meet the requirement of higher precision. As a result, data fusion has gradually become an effective way to obtain high-quality time-series data from remote sensing; and ③ results from different extraction algorithms are evaluated by statistics, visual interpretation, field survey, and previous studies. Furthermore, we conclude the advantages and disadvantages of remote sensing data and compare different extraction algorithms and accuracy evaluation methods. Finally, we discuss the deficiencies of previous studies, and put forward several tips for future studies: ① a reasonable evaluation system is expected to be established for comparing different extraction algorithms; ② more attention should be paid to regions having complex terrain and smallholder farms; ③ to obtain higher quality time-series data from remote sensing and improve the efficiency of data processing, denoising algorithms, data fusion, big data, and cloud computing techniques should be considered; ④ multi-scale validation of results is needed if data is available.