基于MODIS NDVI数据的东北森林物候期监测

Monitoring Forest Phenophases of Northeast China based on MODIS NDVI Data

物候是指示气候与自然环境变化的重要指标.遥感技术的发展为物候监测和研究提供了新的手段.本文研究对象是中国东北森林,森林分布范围由Landsat TM影像解译得到的2000年土地利用数据确定.遥感数据源是2003年500m空间分辨率的MODIS NDVI 8天合成时间序列数据.通过分析东北主要森林树种的NDVI时间序列特征,表明不同树种的同一遥感参数时间序列基本形状近似,在关键物候期和变化振幅上存在差异,这为根据遥感参数时间序列曲线监测森林物候期奠定了理论基础.将MODIS NDVI 8天合成时间序列数据应用时间序列谐波分析法（HANTS）重构成每天的NDVI时间序列数据影像.基于每天的NDVI时间序列数据,研究采用动态阈值法获取了东北森林物候期及其空间分布格局.研究表明东北大部分地区树木在第100天～150天开始生长,到第260天～290天逐渐停止生长,生长季长度集中在140天～180天.通过与部分物候观测数据的比较验证,表明基于MODIS NDVI数据获取的树木生长始末日期与调查资料具有可比性,获取的森林物候期具有一定的可靠性.

Phenology is the study of recurring vegetation cycles and their connection to surrounding environmental factors such as climate, hydrology, soil, etc. Phenological records provide an integrative indication of the sensitivity of natural systems to climate changes and have a clear added value to climate impact assessment. Regional forest phenology is important in ecosystem simulation models and coupled biosphere/atmosphere models. The principal advantage of remote sensing data compared to traditional observations in the field is the possibility they offer to gather synoptic information at regular time intervals over large areas. Repeated observations from satellite-borne sensors can be used to monitor phenological dynamics at regional level. Northeastern China has abundant tree species and a variety of forest types, including evergreen conifer forest, deciduous conifer forest, deciduous broadleaf forest, and mixed forests. It is a suitable area to study forest phenology of China. To study forest phenology individually, the forest region of Northeast China was extracted from the whole Northeast China region based on the 1 ： 100 000 Land Use Map of China for 2000 which interpreted from Landsat-7 ETM ＋ images. This paper describes the application of Moderate Resolution Imaging Spectroradiometer （MODIS） NDVI time series data for forest phenological pattern in Northeastern China. The time series data of 500m MODIS NDVI were inferred from 8-day MODIS surface reflectance datasets,2003. The daily MODIS NDVI data were got from 8-day MODIS NDVI data after processing by Harmonic Analysis of Time Series （HANTS）. The dynamic threshold method was used to extract three key forest phenophases,which are the start of growing season （SOS）, end of growing season （EOS） and growing season length （GSL） based on the daily MODIS NDVI data. The start dates of growing season in Northeast China focus on 100th ～150th day. This is consistent with the period of tree leaf unfolding in spring. The end dates of growing season focus on 260th - 290th day. It is corresponding with the period of defoliation in the fall. The growing season length mainly ranges from 140 to180 days. The forest phenological variables inferred from MODIS data are related to the distribution of forest types. In Daxinganling forest region, which located in north of Northeast China, the dominant forest types are deciduous species such as larch, birch et al. The growing season begins later and ends earlier. The length of growing season is short. In Xiaoxinganling and Changbaishan forest regions, located in southeast of Northeast China, the dominant forest are evergreen species such as Korean pine, spruce, fir etc. The growing season begins earlier and ends later. The length of growing season is longer than that of deciduous tree species. Then the derived phenophases were validated by previous research achievements in the same area. Results indicate that forest phenophases from MODIS NDVI data is feasible.