基于S G滤波的陕西关中地区冬小麦生育期遥感识别和长势监测

采用SPOT VGT S10产品数据，通过S G滤波处理，重构NDVI时间序列，分析陕西关中地区冬小麦返青期和抽穗期在遥感影像上NDVI时序曲线变化特征，对冬小麦的返青期和抽穗期进行遥感识别。通过与站点观测的生育期数据比较，遥感识别的均方根误差大部分在±20d之内，返青期均方根误差（RMSE）为14.29d，抽穗期RMSE为9.16d，表明基于SPOT VGT数据对关中地区冬小麦返青期和抽穗期进行遥感识别是可行的。采用差值模型基于环境减灾小卫星（HJ星）数据进行渭南市2012年冬小麦抽穗期长势监测研究，结果显示大部分地区2012年冬小麦长势稍好于2011年，经与实地调查资料比较，监测结果与实际情况基本一致。采用HJ星数据进行冬小麦长势监测，结果纹理清晰，且以同源数据提取的冬小麦种植面积作为掩膜剔除非麦区域，降低了重采样带来的误差，用其进行长势监测研究具有明显优势。

Remote Recognition and Growth Monitoring of Winter Wheat in Key Stages Based on S G Filter in Guanzhong Region

The NDVI time series was constructed by using Savitzky Golay filter method based on SPOT VGT S10 product data.The temporal variation of the NDVI values during reviving stage and tasseling stage of winter wheat was obtained.The temporal variation of NDVI were applied for detecting the reviving stage and the tasseling stage of winter wheat in Guanzhong region in Shaanxi province.Compared with the field measurements，the prediction uncertainty of remote sensing approach was within20d，and the corresponding RMSE values of reviving stage and tasseling stage were 14.29d and 9.16d，respectively.This meant remote recognition to winter wheat during reviving stage and tasseling stage based on SPOT VGT data was available.Combing with HJ satellite data，the deference model was used to study the winter wheat growth in 2012，the results showed that the winter wheat growth in 2012 was better than 2011 in Weinan region，which was consistence with the in situ measurements.Based on HJ data，the satellite derived spatial distributing veins of winter wheat growth were clear，and using the plant area of winter wheat extracted from the same satellite data as the mask to exclude the non winter wheat region could reduce the uncertainty generated during the data resample procedure.The results indicated that HJ data was pronounced advantage for winter wheat growth monitoring．