基于NOAA NDVI的植被生长季模拟方法研究

在近几年里， 大尺度的植被生长季监测已经成为全球气候变化研究中的一个重要科学问 题。NOAA/NDVI 数据为研究植被生长季时空变化规律提供了重要手段。文章综述并分析比较了 基于NDVI 估测植被生长季开始、结束、长度等特征参数的方法: NDVI 阈值、时间序列分析、物候 期的频率分布型与NDVI 相结合、主分量分析、利用曲线进行拟合等方法。受不同因素影响， 各方 法有不同的应用局限性， 因此， 在以前研究的基础上， 利用较常用的四种方法: 阈值法、滑动平均 法、最大变化斜率、曲线拟合模型模拟了锡林浩特1991～1999 年的草原生长季， 最后利用野外实 测的草原返青期验证了监测结果。结果表明: 与地面观测数据相结合， 基于阈值可得到较好的草 原返青期; 基于曲线拟合模型能适用于大尺度上的植被生长季变化监测， 但存在问题是拟合曲线 很难接近于实际曲线， 因此， 需要进一步研究的内容是选择合适的曲线估测模型， 监测不同植被 类型生长季的年际变化规律。

The Methods of Simulating Vegetation Growing Season Based on NOAA NDVI

In recent years, estimating vegetation growing season on large scale has become a significant scientific field in global climate change studies. NOAA/AVHRR NDVI provides important means of research on temporal and spatial variability of vegetation growing season. This paper has summarized, compared and analyzed methods of estimating beginning, end, length of vegetation growing season on the basis of NDVI: NDVI threshold, NDVI time series analysis, combination of frequency distribution pattern of plant phenology with NDVI, principal component analysis,curves fitting model and so on. Influenced by different factors, each method has different limitations in application. Therefore, it is necessary to seek better methods for determining interannual and regional distribution variability of vegetation growing season. And on the basis of previous studies, growing season of grassland in Xilinhaote from 1991 to 1999 is estimated by NDVI threshold method, smoothed moving average, the greatest change slope, and curves fitting model. Then the results were verified with grassland greenup by field survey. Results shows that, combining with field surveying data, grassland greeup with better accuracy is acquired with NDVI threshold method, and curves fitting model can be applied to monitoring vegetation growing season on large scale. However, it is difficult for actual curves to match the simulated curves. Therefore, it is very important to choose appropriate fitting curves for subsequent studies. Furthermore, it is necessary to carry out further research on interannual variability of vegetation growing season.