2000—2014年青藏高原植被净初级生产力时空变化及对气候变化的响应

青藏高原是全球气候变化最敏感的地区之一。计算青藏高原生态系统净初级生产力(Net Primary Productivity, NPP)对精确估算全球碳循环具有重要意义。基于CEVSA模型，利用M-K趋势检验法、Sen’s斜率估计法及Pearson相关系数法，分析了2000—2014年青藏高原生态系统的净初级生产力时空变化特征。结果表明：（1） 青藏高原高寒生态系统净初级生产力在空间分布上表现出由东南向西北减小的趋势，在东部及东南部的森林区NPP在600~1 200 gC·m^-2·a^-1之间，中部草原和草甸区NPP在200~400 gC·m^-2·a^-1之间，西部和北部荒漠区，受水热条件的限制NPP很小，该趋势与水热分布趋势基本一致。（2） NPP年际变化与多年平均气温呈正相关，与降水量呈负相关。NPP与气温呈正相关的地区面积占研究区总面积的82.24%，与降水量呈负相关的地区面积占49.31%，表明气温是影响植被NPP空间分布的主要因子。（3） 近15 a来，青藏高原 NPP整体呈增加趋势，与气温趋势变化一致，降水量表现出微弱的减少趋势，气温的增加伴随降水量的减少是青藏高原NPP缓慢增加的主要原因。因此，准确描述NPP对气候变化响应的能力将使我们能够深入理解陆地生态系统应对全球变化做出的反应。

Response of net primary productivity of Tibetan Plateau vegetation to climate change based on CEVSA model

As the “Third Pole” of the world,the Qinghai-Tibetan Plateau,China is located at 74°-104°E and 25°-40°N.It is known for its high altitude,complex terrain,and harsh climate.The Qinghai-Tibetan Plateau is one of the most sensitive regions to global climate change.It can affect the productivity of plateau’s ecosystem.Calculating the net primary productivity of the Qinghai-Tibetan Plateau ecosystem is very important for accurately estimating the global carbon cycle.As NPP cannot be directly measured on a regional or global scale,so model based estimation is the only way to proceed.In this study,we used the CEVSA model to eatimate the NPP in the Tibetan Plateau between 2000 and 2014, and analyzed the spatio temporal patterns and trends of NPP with M-K Trend test method,the Sen’s slope estimation method,and the Pearson correlation coefficient method.The model was based on a 0.1°×0.1° resolution map of vegetation types,soil texture data,and daily meteorological data.The results indicated as follows:(1) the NPP for the Qinghai-Tibetan Plateau decreased from southeast to northwest and was consistent with the trend of water-heat distribution.The results were similar to those obtained by ZHOU Caiping et al,who applied a combination of terrestrial ecosystem model and MODIS data to estimate the net primary productivity of the Qinghai-Tibetan Plateau.In the spatial distribution,the NPP of the forests in the east and southeast was between 600 and 1 200 gC·m^-2·a^-1 and the NPP of the central grassland and meadows was from 200 to 400 gC·m^-2·a^-1.In the western and northern deserts,the NPP was limited by the moisture and temperature.(2) the annual average temperature increase had a significant positive effect,while the precipitation decrease had a significant negative effect on the Qinghai-Tibetan Plateau’s NPP.The annual NPP was positively correlated with annual mean temperature over 82.24% of the region,while negatively correlated with annual precipitation over 49.31% of the region.Therefore,temperature is considered to be the dominant factor determining spatial variations in NPP.The predecessors also obtained similar result.For example,LIU Gang et al. analyzed the spatiotemporal variation of net primary productivity and climate controls in China from 2001 to 2014.Based on their results,the correlation analysis between NPP and meteorological elements indicated that NPP was positively correlated with temperature in the Changbai Mountain area,Qinghai Tibetan Plateau,and southern areas.(3) from 2000 to 2014,the trend of increasing NPP was consistent with the changes in temperature.The precipitation showed a slight decrease change.A period of warming accompanied by a decrease in precipitation contributed to the trend of a gradual increase of NPP in Qinghai-Tibetan Plateau.Therefore,improving our ability to accurately describe the response of NPP to climate changes will provide a better understanding of terrestrial ecosystem responses to global changes.However,there are still some issues to be solved,such as the uncertainty of NPP prediction.These uncertainties mainly include the driving variables and parameters in the model.Overall,to predict the impact of climate change on ecosystems at the regional level,modeling uncertainty can be reduced by increasing the spatial resolution of the driving variables.In addition,optimizing the model parameters can also reduce the uncertainty in the model simulation.