黄土高原地区NDVI与气候因子空间尺度依存性及非平稳性研究

基于MODIS传感器的植被指数产品(MOD13Q1)及50年气候数据,通过地理加权回归与普通最小二乘回归模型对比,对中国黄土高原地区NDVI与气候因子间的空间尺度依存性及非平稳性进行研究,以期准确建立二者间关系.结果表明:① 研究区域内,NDVI与气候因子间存在很强的空间尺度依存关系,相同空间尺度下,年均降水较年均温对NDVI影响的波动性更大;② 与普通最小二乘回归模型相比,地理加权回归模型能够更准确地展现二者间关系;③气候因子对该地区NDVI的影响差异明显,降水存在直接正向影响,而温度的影响则较复杂;④ NDVI与气候因子间沿东北--西南的分布格局体现出区域内不同植被--气候区差异特征.二者间的异质情况还反映出除气候外,人类活动,地形等其他因素对NDVI的影响.

Spatial scale-dependent and non-stationarity relationships between NDVI and climatic factors in the Loess Plateau

Understanding the relationship between vegetation and climate is the premise and foundation to reveal the distribution pattern of vegetation in large areas. Normalized Differentiation Vegetation Index (NDVI) has been regarded as an effective indicator for vegetation growth and distribution, especially for the large scope. To establish the accurate relationship between NDVI and climatic factors, this paper, based on the vegetation index product (MOD13Q1) relating to the Loess Plateau Area, northern China, and the climatic data observed in resent 50 years from the same area, has conducted a comparison between the two models named Geographically Weighted Regression, GWR, and Ordinary Least Squares, OLS, respectively. We analyzed the non-stationarity and scale-dependent characteristics between the two models with validation tool of corrected Akaike's Information Criterion, AICc, and calculated Moran's Index. The results showed: (1) the NDVI and the climatic factors had a strong scale-dependent relationship in the study area, and when the bandwidth approached to about 330 km in scale, they came up to a stable status. The annual mean precipitation, AMP, presented a larger fluctuation than the annual mean temperature, AMT, at the same scale of bandwidth. (2) Compared with OLS, the results of GWR showed a more accurate spatial distribution of vegetation, through validation by its model performance (AICc, R², R² adjusted) and Moran's Index of residuals (P<0.01). (3) The predicated result of GWR reflected the heterogeneity to some extent between the NDVI and the climatic factors. Precipitation had direct and positive influence on NDVI, whereas that of temperature was complicated. (4) The northeastern to southwestern distribution pattern between the NDVI and the climatic factors indicated a remarkable difference of climate-vegetation distribution pattern within the Loess Plateau. The heterogeneity between them also showed that some other factors such as human activities and/or orographic rains exerted influence on NDVI.