Knowledge Resource Center for Ecological Environment in Arid Area
| 基于卫星遥感数据的西非地区植被动态与气候关系的研究 | |
| 其他题名 | Vegetation Dynamics in relation to Climate over West Africa using Remote Sensing and Satellite Data |
| TERTSEA IGBAWUA | |
| 出版年 | 2018 |
| 学位类型 | 博士 |
| 导师 | 张佳华 |
| 学位授予单位 | 中国科学院大学 |
| 中文摘要 | 研究植被生长动态和气候变化在不同时空尺度下的相关性对应对即将来临的生态系统变化以及其对区域和全球变化的改变有极其重要的意义。在西非，农业是主要的收入来源，大部分人口依靠农业为生。最近，降水和SST出现了季节性的扰动，导致生长季开始期提前生长季结束期推后或者相反。降水扰动对自然资源尤其是森林，河流，湿地，牧场，农田以及湖泊有重要的长期和短期的影响，会导致食品安全和其他经济问题。较高的人口增长速率，森林退化，过度开垦耕地和原油以及不合理的土地保护政策致使大量的水土流失。因此，理解植被动态对气候变化和水汽通量在年内和年际尺度上的影响是非常必要的。本研究利用时间分辨率为半月的GIMMS AVHRR NDVI3g的遥感数据集和1981-2011年月尺度的MERRA（NASA’s Modern Era Retrospective Analysis for Research and Applications）气象数据集分析了尼日利亚境内的植被动态。气象数据集中涉及的气象要素包括距地2m的气温、降水和相对湿度。此外，研究还使用NDVI3g，CRU降水和气温数据分析了整个西非地区的植被的变化，并将以上两个空间尺度分析结果进行了比较。研究结果显示，尼日利亚境内植被区域的年均NDVI呈现0.03 x 10-4NDVI/year 的增长趋势，是整个西非区域年均NDVI增长趋势(21.6 x 10-4yr-1 ) 的0.14 %。尼日利亚境内约94.7%的区域的植被NDVI变化呈现增长趋势，5.3%的区域的呈减少趋势。其中，NDVI增长的区域主要位于南部地区，减少的区域主要分布在北部地区。在西非区域尺度，大约21.2%的区域的植被NDVI呈现减小趋势，67.2%的区域的呈现增长趋势，11.6%的区域的没有显著变化趋势。尼日利亚境内的NDVI和气象因子之间的时间相关性受NDVI对于气象要素响应的滞后时间（lags）的影响，研究区内lags的变化范围为0-4个月。NDVI对气温响应的滞后时间为2个月，而对相对湿度和降水的滞后时间分别为1个月和4个月。在西非区域尺度上，NDVI对降水和气温的滞后时间分别为2个月和4个月。结果还表面，无论是来自CRU还是在分析气象资料的降水均与西非区域植被变化呈现最强的相关性。在时间序列和空间分布上，植被的增长总是伴随降水的增加，而除森林以外，温度越高的区域植被越稀少。这意味着高温导致植被的减少。本研究利用植被和CRU观测数据进行的聚类分析对这一观点进行了详细阐述。K-Means聚类分析显示，西非四个聚类区域的植被NDVI的变化与降水的变化保持的高度的一致性，且这种一致性高于与温度。NDVI与温度、降水和水分收支数据之间的残余趋势分析（RESTREND）为研究非气象因素产生的土地退化提供了有用信息。利用西非地区的降水和气温数据进行分析，植被高度退化区主要分布在撒哈拉荒漠草原地区，水分收支平衡的地区则分布在几内亚海岸地区。这说明植被动态变化还受到人口和土地利用等其他因素的影响。因此，本研究对开展西非地区气候变化监测的研究是关键的一步，对于正在开展的植被动态和气候变化监测方面的研究工作具有重要意义。使用NASA的MERRA和MERRA-2高空气象模式（P-E*）模拟1980-2013年西非地区水分收支状况，并对模拟结果进行评估，结果显示近代发展的再分析资料在区域插值的降水和表面蒸散之间具有不平衡性，且该特点在新发布的MERRA-2数据中表现更为显著。本文将MERRA和 MERRA-2的大气水分平衡产品作了相互比较，之后针对欧洲中期天气预报中心(ECMWF)的再分析数据ERA-I 以及日本55年再分析数据（JRA-55），与它们的模型预测输出又做了对比。结果指出：在西非夏季季风期间， MERRA-2（ERA-I ）数据，12-20（5-13）毫米/月的偏差可以导致萨赫勒地区（14oN - 20oN）被划分为水分输出源。对比MERRA/MERRA-2以及ERA-I 和JRA-55的计算结果，文章发现在几内亚区域，MERRA的平均P-E* 要小于ERA-I (JRA-55)的结果，为18.94(52.24)毫米/月，而在撒哈拉区域，P-E* 的平均值为25.03(4.53)毫米/月，这要大于ERA-I (JRA-55)。对于MERRA-2数据，几内亚区域和撒哈拉区域的平均P-E* 都要小于ERA-I (JRA-55)，值分别为25.76(59.06)毫米/月和73.72(94.22)毫米/月。这些差异源于数据同化方法中的调整、卫星的校准，以及实测数据集。MERRA-2中对流P值的参数化的改变和P中提升的再蒸发值，可能是P与E估计中出现正偏差的原因。MERRA/MERRA-2与CRU降水之间存在的细微的不一致性突出说明西非地区气候研究中的一个主要挑战，而提高观测数据和再分析数据中的地表通量数据仍然十分重要。为了研究1982-2011年西非地区长时间序列的NDVI与LAI变化，文章使用了GIMMS的NDVI以及GLASS的LAI数据。文中分析了(a)一般线性回归得到的年变化与季节变化；(b) 去趋势lag-1自相关；（c）区别某一时间序列的长程相关性的赫斯特指数h；（d）NDVI 和 LAI 的多重分形特征。结果表显示明植被变化存在时间持续性的模式和趋势。在95%的置信区间内，NDVI 和 LAI的C(1)值都在95%的置信区间表现为马尔可夫线性类型的持续性。R/S 和 MF-DFA方法的赫斯特h指数的值表明NDVI和LAI数据序列具有长程相关性，但是小尺度的时间序列的相关性与大尺度的不同。在空间上，赫斯特h指数也揭示了依赖于降水、水文循环和人类活动的植被分布的长程相关性。另外，区域尺度、森林和萨赫勒地区的NDVI (LAI)的多重分形波谱宽度ω分别是0.519(0.513), 0.510(0.536) and 和 0.649(0.701)，这表明森林和萨赫勒地区的LAI相对于NDVI有更强的分形强度，区域尺度上的NDVI 分型程度强于LAI。因此，对比MF-DFA 方法在原始数据、随机变化的数据以及仅改变时间序列段的数据上的运行结果，发现区域尺度和萨赫勒地区的多重分形特性基本上是由长程相关导致的，而森林地区的多重分形主要是由重尾概率分布函数导致的。 |
| 英文摘要 | Understanding the relationship between vegetation growth dynamics and climate change at different spatial and temporal scales is vital in projecting upcoming ecosystem changes and their modification to regional and global changes. In West Africa, agriculture is the main source of earnings and majority of the population depends on agriculture for their livelihood. Recently, there has been seasonal fluctuation in rainfall and SST resulting in delayed start of season (SOS) or early end of season EOS or vice versa. Rainfall fluctuation has significant long and short term effects on natural resources particularly forests, lakes, wetlands, rangelands, croplands and rivers leading to food insecurity and other economic problems. High population growth rate, deforestation, crude agricultural practices, crude oil spills and improper land conservation policies have resulted in massive land degradation with losses of fertile soil. Therefore understanding the impact of vegetation dynamics in relation to climate variability and moisture flux at the inter-annual to decadal time scales are desirable. Vegetation dynamics in Nigeria were assessed using the GIMMS AVHRR NDVI3g bimonthly data and NASA’s Modern Era Retrospective Analysis for Research and Applications (MERRA) monthly climate data from 1982 to 2011. The climate data employed, included air temperature, precipitation and humidity (at 2 m). Further, the work also analyzed vegetation changes over the whole region of West Africa using NDVI 3g, CRU precipitation and temperature, and the results were then compared with the earlier results of the reanalysis data sets. The results indicated that the annual mean NDVI increased by 0.03 x 10-4NDVI/year of the Nigeria’s total vegetative area. The annual mean trend is about 0.14 % of the total increase of NDVI over West Africa (21.6 x 10-4yr-1 ). About 94.7 % of the area showed positive trends mostly in the southern part of the country, while 5.3 % showed negative trends mostly in the northern part. At regional scale, about 21.2 % pixels showed negative trends, 67.2 % showed positive trends while 11.6% pixels were unchanged over West Africa. The seasonal correlations between NDVI and climate elements over Nigeria varied with lags and the lags were between 0 and 4 months across all the zones: temperature lagged NDVI by 2 months while humidity and precipitation lagged by 1 and 4 months respectively, while at regional scale NDVI lagged precipitation and temperature at 2 and 4 months respectively. Results also show that, in all the analysis vegetation changes were best explained by precipitation dynamics over West Africa using reanalysis data sets and CRU gridded observation data sets. Vegetation increase is favored by increase in precipitation temporally and domain wise while areas with high temperatures showed vegetation decrease except the forest regions. This means a decrease in the vegetation resulted in temperature increase and vice versa. This was elaborated by the spatial cluster analysis performed in this work using vegetation and CRU observation data sets. The K-means cluster analysis suggested that the spatial variation in NDVI over the four clusters domain wide over West Africa is highly aligned to precipitation variation than temperature. The application of RESTREND analysis of NDVI with temperature, precipitation and moisture budget data sets has provided information about land degradation due to non-climate forcings. The highly degraded areas which are explained by precipitation and temperature over West Africa are spread over the Sahel while those for moisture budget are located around the Guinea coast. It was therefore suggestive that vegetation dynamics were also influenced by other factors like demography and land use. This work is therefore a step in climate change monitoring in West Africa, which will serve as a contribution to the ongoing research works on vegetation dynamics and climate change monitoring.In assessing the performance of NASA’s Modern-Era Retrospective Analysis for Research and Applications (MERRA) and MERRA-2 aerological (P-E*) model in reproducing the salient features of West Africa water balance (from 1980-2013), it was shown that the recent reanalysis efforts have generated imbalances between regional integrated precipitation (P) and surface evaporation (E), and the effect is more in the newly released MERRA-2. The atmospheric water balance of MERRA and MERRA-2 were inter-compared and thereafter compared with model forecast output of European Centre for Medium‐Range Weather Forecasts (ECMWF) Re‐Analysis (ERA‐I) and Japanese 55-year Reanalysis (JRA-55). Results indicated that a bias of 12-20 (5-13) mm/month in MERRA-2 (ERA-I) leads to the classification of the Sahel (14oN - 20oN) as a moisture source during the West African Summer Monsoon. Comparisons between MERRA/MERRA-2 and prognostic fields from two ERA-I and JRA-55 indicated that the average P-E* in MERRA is 18.94(52.24) mm/month which was less than ERA-I (JRA-55) over Guinea domain and 25.03(4.53) mm/month greater than ERA-I (JRA-55) over the Sahel. In MERRA-2, average P-E* indicated 25.76(59.06) mm/month which was less than ERA-I (JRA-55) over Guinea and 73.72(94.22) mm/month less than ERA-I (JRA-55) over the Sahel respectively. These imbalances are due to adjustments in data assimilation methods, satellite calibration and observational data base. The change in convective P parameterization and increased re-evaporation of P in MERRA-2 is suggestive of the cause of positive biases in P and E. The little disagreements between MERRA/MERRA-2 and CRU precipitation highlights one of the major challenges associated with climate research in West Africa and major improvements in observation data and surface fluxes from reanalysis remain vital. To study the long range correlations NDVI and LAI records over West Africa from 1982 to 2011, GIMMS NDVI and GLASS LAI were used. The analysis assessed (a) the annual and seasonal trends obtained using Ordinary Linear Regression, (b) the detrended lag-1-autocorrelation C(1), (c) the scaling hurst exponent h that distinguishes the long range correlation in a time series, and (d) the multifractal characteristics of NDVI and LAI. Results show that there exist some patterns or trends in the records that persist over time. The value of C(1) for both NDVI and LAI indicated a Markov linear type persistence which was significant at 95% confidence interval. Consequently, the scaling h values of the Hurst for R/S and MF-DFA methods showed that NDVI and LAI data series are long range correlated but that the correlations at small scales of the sequences are different from the large ones. Spatially, the scaling h exponent also revealed long range correlations which are consistent with the vegetation distribution based on precipitation, hydrologic cycle and human activities. Also, the multifractal spectrum width ω of NDVI (LAI) was obtained as 0.519(0.513), 0.510(0.536) and 0.649(0.701) at regional scale, forest and Sahel regions respectively, which shows that LAI has stronger fractal strength compared to NDVI over the Forest and Sahel regions while NDVI stronger than LAI over the forest region. Accordingly, the comparison of the MF-DFA results of original data to those of shuffled and surrogate series indicated that the multifractal nature of considered time-series at regional scale and Sahel is almost due to long-range correlations while over the forest regions, multi-fractility is dominantly due to heavy tail probability distribution function (PDF). |
| 中文关键词 | 西非 ; 净降水量 ; 遥感 ; 气候变化 ; 植被动态 |
| 英文关键词 | West Africa Moisture budget Remote Sensing Climate change Vegetation dynamic |
| 语种 | 中文 |
| 国家 | 中国 |
| 来源学科分类 | 地图学与地理信息系统 |
| 来源机构 | 中国科学院遥感与数字地球研究所 |
| 资源类型 | 学位论文 |
| 条目标识符 | http://119.78.100.177/qdio/handle/2XILL650/288055 |
| 推荐引用方式 GB/T 7714 | TERTSEA IGBAWUA. 基于卫星遥感数据的西非地区植被动态与气候关系的研究[D]. 中国科学院大学,2018. |
| 条目包含的文件 | 条目无相关文件。 | |||||
| 个性服务 |
| 推荐该条目 |
| 保存到收藏夹 |
| 导出为Endnote文件 |
| 谷歌学术 |
| 谷歌学术中相似的文章 |
| [TERTSEA IGBAWUA]的文章 |
| 百度学术 |
| 百度学术中相似的文章 |
| [TERTSEA IGBAWUA]的文章 |
| 必应学术 |
| 必应学术中相似的文章 |
| [TERTSEA IGBAWUA]的文章 |
| 相关权益政策 |
| 暂无数据 |
| 收藏/分享 |
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。
