不同时序植被指数重构方法的江苏省冬小麦物候期提取的影响

基于2009~2010年MODIS数据,以江苏省为研究区域,比较两种常见的重构植被指数序列方法的重构能力以及应用到冬小麦物候期提取上的效果;并且用观测数据对提取结果进行了验证。结果表明,两种重构方法都能较好的去除噪声,但Savitzky-Golay滤波方法在保持原始数据的真实性方面要优于HANTS(harmonic anazysis of time series)方法,而HANTS方法在呈现原始数据的周期性方面更有优势。通过动态阈值法提取冬小麦物候期时,经两种重构方法处理后的植被指数时间序列,在返青期和成熟期提取上均方根误差均在8 d之内,抽穗期效果较差。S-G滤波重构后的植被指数时间序列提取结果要优于HANTS方法。总体来说,经两种方法滤波后提取的江苏全省冬小麦关键物候期,都能够体现全省范围分布的变化性。

Phenology Extraction of winter wheat based on different time series vegetation index reconstructing methods in Jiangsu province

Aiming to compare the ability of two common methods in reconstructing the vegetation index series based the MODIS data from 2009 to 2010 which cover Jiangsu Province and then application to extract the phenology of winter wheat, and with the observation data on the extraction results are verified. The results showed that two kinds of reconstruction method all can better remove the noise, but S-G filter method is superior to maintain the integrity of the original data in HANTS method, HANTS method have more advantages in the aspect of present original data periodically. When through dynamic threshold method to extract winter wheat phenology, the vegetation index time series after dispose by two reconstruction methods, the returning green stage and mature stage extraction on root mean square error were within 8 days, heading stage effect is poor. About the extraction results, S-G filter method is better than HANTS method.