摘要
利用1982～2000年NOAA AVHRR NDVI时间序列数据，分剐采用中值法、经验公式法、延迟滑动平均法和原型曲线法，划分我国温带草原区7个牧业试验站和暖温带落叶阔叶林区5个物候站所在像元的植被生长季节。通过比较各种方法划分结果与物候观测结果的标准差，以及遥感生一长季节划分结果的绝对误差分级百分比、相关误差指标和相关系数，对比分析了各种方法生长季节划分结果的稳定性、准确性和有效性。研究结果表明，划分生长季节的开始，在草原区和森林区均以原型曲线法最为适宜。划分生长季节的结束，在草原区以中值法的划分结果最好，其次是原型曲线法衰落点的划分结果；而在森林区，则以原型曲线法休眠点的划分结果最好。本文为遥感生长季节划分结果的地面检验与划分方法的筛选，提供了典型的案例和方法。
Satellite-derived beginning and end dates of growing season at 7 sites of the temperate steppe area and 5 sites of the warm temperate deciduous broad-leaved forest area in China were determined with the measures of seasonal midpoint NDVI （SMN）, inflection point by Moulin （MOULIN） and by Zhang （LOGISTIC）, and moving average （MA） using NOAA/AVHRR NDVI time series data from 1982 to 2000. The four measures were matched to surface observations and then their stabilities, accuracies, and validities were studied via compare of absolute error distributions, standard deviations, mean absolute errors, and correlation coefficients and so on. The results show that LOGISTIC is the most appropriate of the four measures in determining beginning date of growing season in both the steppe area and the forest area. As to satellite-derived end date of growing season in the steppe area, SMN performs best, followed by LOGISTIC with the senescence point defined as end date. In the forest area, of the four satellite- derived end date of growing season measures, LOGISTIC with the dormancy point defined as end date, is the most preponderant one. This paper provided a typical case and method to surface validation of the satellite-derived vegetation growing season and selection of various measures.
出处
《遥感技术与应用》
CSCD
2007年第3期382-388,共7页
Remote Sensing Technology and Application
基金
国家自然科学基金项目(40671028
40371042)~~
关键词
NDVI
遥感生长季节
地面检验
方法比较
NDVI, Satellite-derived growing season, Surface validation, Measures compare
作者简介
丁登（1982-），女，硕士研究生，主要从事植被物候学和气候变化响应研究。
