基于通量塔净生态系统碳交换数据的植被物候遥感识别方法评价

选择北美洲72座通量塔观测的净生态系统碳交换（NEE）数据来计算植被物候，并以此作为参考数据，从可行性和准确性两方面对阈值法、移动平均法和函数拟合法三大类常用的植被物候遥感识别方法进行了综合评价.结果表明： 基于局部中值的阈值法对植被物候识别的可行性和准确性均最优；其次为Logistic函数拟合法中的一阶导数方法；移动平均法对植被物候识别的可行性和准确性与移动窗口的大小有关，对于16 d合成的归一化差值植被指数（NDVI）时间序列数据来说，移动窗口大小为15时能获得较优的结果；而全局阈值法对植被物候识别的可行性和准确性均最差；Logistic函数拟合法中的曲率变化率方法在识别植被物候时虽然与基于NEE数据得到的植被物候在数值上存在较大偏差，但二者之间具有较高的相关性，说明基于曲率变化率方法识别出的植被物候能较真实地反映植被物候在时空上的变化趋势.

Relationships between soil moisture and environmental factors at different spatial scales

Taking Yanhe basin as a case, this paper studied the relationships between soil moisture and environmental factors at slope surface, small watershed, and regional scales by using canonical correspondence analysis (CCA). In the basin, slope position and land use type were the main factors controlling soil moisture change at slope surface scale, slope and land use type were the dominating factors at small watershed scale, while evapotranspiration, precipitation, and land use type were the key factors at regional scale, which suggested that with the enlargement of studied spatial scale, the information contained by topographic factors (slope direction, slope position, and slope) gradually converted to climatic factors (evapotranspiration and precipitation), but the information contained by land use type was almost invariant. This study could have significance in studying soil moisture model and hydrological model and their extrapolation.