摘  要  【目的】 利用水稻叶片光谱反射率建立白背飞虱Sogatella furcifera虫量的监测模型，为稻飞虱的光谱监测提供方法支持。【方法】 在水稻分蘖期和孕穗期分别人工接种不同数量的白背飞虱3龄若虫，使水稻受到不同程度的危害；在暗室利用ASD手持式光谱辐射仪测定白背飞虱为害1-4周后水稻倒4单叶和全部叶片的光谱反射率；采用相关性分析方法确定叶片光谱反射率和植被指数与白背飞虱虫量的关系；运用逐步回归建模方法组建基于植被指数的白背飞虱虫量监测模型。【结果】 稻株倒4单叶和全部叶片的光谱反射率不能很好表征水稻分蘖期受20头/株以内白背飞虱为害1-2周和孕穗期受30头/株以内白背飞虱为害1周的虫量大小，但为害3-4周时叶片反射光谱均有稳定波段区域与虫量显著相关。分蘖期倒4单叶和全叶在440和680 nm处反射率可区分16头/株白背飞虱为害3周时的虫量，680和760 nm处反射率可区分12头/株白背飞虱为害4周时的虫量；孕穗期全叶680和760 nm处反射率可区分30头/株白背飞虱为害3周时的虫量，倒4单叶在760 nm处反射率可区分12头/株白背飞虱为害4周时的虫量。在白背飞虱为害分蘖期和孕穗期稻株3和4周时，倒4单叶在400-1 000 nm范围内的差值（DVI）、比值（RVI）和归一化（NDVI）植被指数与虫量相关程度高的光谱波段区域较为稳定，并且DVI与虫量的相关性强于RVI和NDVI。分别建立了基于倒4单叶DVI的分蘖期和孕穗期稻株为害3和4周时的白背飞虱虫量的逐步回归模型，模型均方根误差在两水稻生育期分别约为4和5-8头/株。【结论】 稻叶光谱反射率对白背飞虱为害虫量的表征能力受水稻生育期和为害时长的影响。稻株受白背飞虱为害2周后，倒4单叶和全叶的光谱反射率在稳定波段区域内均与白背飞虱虫量存在显著相关性，叶片光谱的差值植被指数与白背飞虱虫量相关稳定，可用于建立虫量监测模型。

Abstract  [Aim]  Models for monitoring the population size of the white-backed planthopper (WBPH), Sogatella furcifera, were established using the spectral reflectance of rice leaves to provide methodological support for the spectral monitoring of rice planthoppers. [Methods]  The degree of WBPH damage to rice plants was manually controlled by transplanting different numbers of third instar nymphs onto the tillering and booting stages of rice plants. Reflectance measurements were taken from the fourth leaf from top (4LFT) and all leaves of the rice plant 1-4 weeks following WBPH infestation using an ASD FieldSpec handheld spectroradiometer in a dark room. Correlation analysis was used to determine the relationships between leaf spectral reflectance, spectral vegetation index, and the number of WBPHs on the rice plant. Regression models for monitoring the WBPH population size were established using the stepwise regression modeling method based on the spectral vegetation index. [Results]  Spectral reflectance from the 4LFT and all leaves of a rice plant was unable to characterize a population size of up to 20 WBPHs per plant after 1-2 weeks of damage during the tillering stage, or 30 individuals per plant after 1 week of damage during the booting stage. However, after 3-4 weeks of damage, the leaf spectral reflectance had stable bands and was significantly correlated with WBPH population size. At 440 and 680 nm, the reflectance of single and whole leaves at the tillering stage was sufficient to distinguish a population size of 16 WBPHs per plant damaged over 3 weeks, and at 680 and 760 nm, the reflectance was sufficient to distinguish 12 WBPHs per plant after 4 weeks of damage. The reflectance of all leaves at 680 and 760 nm during the booting stage was sufficient to detect 30 WBPHs per plant after 3 weeks of damage, while the reflectance of the 4LFT at 760 nm could detect 12 WBPHs per plant after 4 weeks of damage. In rice plants damaged by WBPH during the tillering and booting stages for 3 and 4 weeks, there was a significant correlation between the spectral bands and the difference vegetation index (DVI), ratio vegetation index (RVI), and normalized vegetation index (NDVI) of the 4LFT single leaf in the spectral range of 400-1 000 nm. Additionally, the WBPH population size remained relatively stable, showing a stronger correlation with DVI than with RVI and NDVI. Stepwise regression models were established to monitor WBPH population size during the tillering and booting stages of rice plants damaged for 3 and 4 weeks, based on the DVI of the 4LFT. The root mean square error of these models was approximately 4 and between 5-8 WBPHs/plant for the tillering and booting stages, respectively. [Conclusion]  The ability of rice leaf spectral reflectance to determine WBPH population size is affected by the growth stage of rice and the duration of WBPH damage. After two weeks of WBPH damage, there was a significant correlation between the spectral reflectance of both the 4LFT single leaf and all leaves and WBPH population size in the stable band regions. The DVI of leaf spectra shows a stable correlation with WBPH population size, which can be used to establish a pest monitoring model.