中文摘要：

[目的]NDVI时序数列能够模拟植物的生长过程，反映其生长状况。目前重建NDVI时序数列的方法有很多，由于模型和参数的不同导致结果存在不确定性以及偏差。本研究旨在对比3种模型（Whittaker平滑、HANTS和Savitzky-Golay滤波）在物候提取和复种指数提取中的应用，以探讨各模型的优缺点。[方法]采用16 d间隔的MODIS MOD13Q1 2000—2012年陕西地区影像，利用3种模型拟合重建NDVI时序数列。首先，将研究区划分为3个气候区，每区分别选择林地和耕地两个采样点，视觉比较各采样点3种模型拟合效果。其次，通过均方根误差、相关系数和信噪比对比各模型拟合精度，并探讨不同植被类型之间拟合精度的差异。然后，采用动态阈值法提取13年植被物候参数（生长开始日期SOS、生长结束日期EOS、生长周期LOS），对比模型提取不同植被类型物候参数均值和标准差的差异。最后，利用二次差分算法和提取规则获取陕西13年复种指数，对比3种模型提取和统计年鉴计算复种指数之间的差异。[结果]Savitzky-Golay滤波拟合精度较高，复种指数提取精度较高，但是提取物候参数方面存在较大误差；HANTS提取物候参数效果较好，但参数设置复杂以及精度较差；Whittaker平滑参数设置简单，能有效降低原始影像的信噪比，在精度和物候参数提取均表现良好；均方根误差和相关系数作为精度检验的标准，二者存在负相关，相关系数比均方根误差更灵敏。[结论]Whittaker平滑能够很好地平衡NDVI时序数列的保真度和粗糙度，在提取物候参数方面表现良好，在提取复种指数方面还有待进一步研究。

英文摘要：

[Objective]NDVI time-series can simulate plant growth and reflect its growing status. Several models have been fitted in the past to smooth time-series vegetation index data from different satellite sensors. However, differences between the models and fine tuning of model parameters lead to uncertainty and bias between the results amongst users. The current research assessed three techniques： Whittaker smoother, HANTS, and Savitzky-Golay filter for smoothing multi-temporal satellite sensor observations with the ultimate purpose of estimating phenological parameters and deriving the annual multiple cropping index （MCI） reliably. [Method]The research used MODIS MOD13Q1 data over the year 2000 to 2012 composed at sixteen day intervals covering the Shaanxi Province, and three models were fitted to reconstruct NDVI time-series. First, samples time-series of different vegetation types in three climate zones of research area were picked, and the fitting effect was contrasted by vision. Second, three evaluation indexes （RMSE, the correlation coefficient, and SNR） were used to compare the quality of each model, and then the differences between different types of vegetation were discussed. Third, three phenological parameters （Start of Season, End of Season, and Length of Season） were estimated by the dynamic threshold method, while the means and standard deviations of phenological parameters of different vegetation types estimated from three models were compared. Finally, the MCIs of each year were derived by quadratic difference algorithm and rules using the NDVI time-series of each model, and then were compared with the MCIs derived from statistics.[Result]Savitzky-Golay performed better than others on RMSE and the correlation coefficient, but has a bigger error in extracting phenological parameters. HANTS performed well in extracting phenological parameters, but which needs more complex parameter settings and has a poor accuracy. Whittaker can effectively reduce the signal to noise ratio of the original i