KMS XINJIANG INSTITUTE OF ECOLOGY AND GEOGRAPHY，CAS
| 基于多源数据的新疆土壤水分时空变异特征研究 | |
| Alternative Title | Research on the Spatiotemporal Variation Characteristics of Soil Moisture Based on Multi-source Data in Xinjiang, China |
| 王云倩 | |
| Subtype | 博士 |
| Thesis Advisor | 杨井 ; 陈亚宁 |
| 2019-06-30 | |
| Degree Grantor | 中国科学院大学 |
| Place of Conferral | 北京 |
| Degree Discipline | 理学博士 |
| Keyword | 土壤水分 降水 温度 NDVI 格兰杰因果关系检验方法 收敛交叉映射 因果关系 干旱 soil moisture precipitation temperature NDVI Granger Causality Convergent Cross Mapping causality drought |
| Abstract | 土壤水分是水文学、气象学和农业科学研究领域的一个重要参数，影响着全球尺度的水循环、能量循环和生物地球化学循环， 在天气和气候预报、 干旱和洪水预测、 农业生产力估算等方面有很大的应用潜力。 然而针对土壤水分的研究在很多方面存在着不足和争议， 例如气候和植被是影响土壤水分变化的两大因素，但是由于土壤水分、气候和植被之间存在复杂的相互作用，量化不同因素对土壤水分变化的贡献仍然是一个挑战。此外土壤水分会影响大气和陆地表面之间水和能量的传递，进而影响天气，但是土壤水分对降水的影响涉及到复杂的大气过程，难以分析和量化，所以一直存在很大争议。因此有必要针对这些问题进一步开展土壤水分的研究。新疆维吾尔自治区位于欧亚大陆腹地，降水稀少，生态系统脆弱，是最易受气候变化影响的地区之一。对新疆土壤水分的研究有助于当地水资源管理、 生态系统保护以及干旱等自然灾害的预防。本文基于遥感和再分析数据，应用不同的统计方法系统地研究了土壤水分的时空变化特征、与气候因子的相互作用及其在干旱监测和预测中的应用， 构建了一个完整的土壤水分研究框架。 本文的主要研究结论如下：新疆土壤水分在空间上有明显差异， 准噶尔盆地和塔里木盆地中心区域土壤水分含量最低，两个盆地之间的区域土壤水分含量最高，且高海拔区域的土壤水分高于低海拔区域的土壤水分。 1980-2016 年间不同季节土壤水分的空间平均值都呈现出增加的趋势，其中夏季土壤水分增加速率明显高于其他季节。生长季和夏季的土壤水分在 2009 年之后增加速率显著提升；春季土壤水分在 1980-2006年间呈现下降趋势， 2006 年之后呈现增加趋势；秋季土壤水分在 2006 年之前无明显变化， 2006 年之后呈显著增加的趋势。不同季节土壤水分的空间变化有明显差异，春季土壤水分显著变化的区域面积最大，约占整个区域的 27.2%，秋季土壤水分显著变化的区域面积最小，约占整个区域的 20%。除了春季以外，其他季节土壤水分显著增加的区域面积均大于显著减少的区域，夏季土壤水分显著增加的区域面积最大。不同土地利用类型的土壤水分含量及年际变化有明显差异，森林土壤水分最高（14.61%），其次是耕地（13.32%）、草地（13.02%）和灌木丛（12.46%），裸地的土壤水分含量最低约为 10.15%。在 2001-2013 年期间，只有草地和裸地上的土壤水分呈现出显著增加趋势，其他土地利用类型上的土壤水分变化不显著。利用非线性格兰杰因果关系检验方法分别量化了降水、温度和植被对土壤水分的影响。结果表明在新疆地区降水、温度和植被对土壤水分的影响有很大差异，在研究区内大约 91%的区域的土壤水分受到降水的影响，降水能够解释土壤水分变化的 0-40%；研究区内大约 59%的区域的土壤水分受到温度的影响，温度能够解释土壤水分变化的 0-8.2%；研究区内大约 44%的区域的土壤水分受到植被的影响，植被对土壤水分的影响较小，仅能解释土壤水分变化的 0-3.3%。研究区内 87%的区域的土壤水分主要受降水影响，然而仅有大约 10%的区域的土壤水分主要受温度的影响。滞后分析表明，当月的降水和温度对当月土壤水分的影响是最大的， 这表明土壤水分在应对降水和温度变化时的弹性较低。此外，降水和温度极值也会对土壤水分产生影响，研究区内约 76%的区域的土壤水分受到降水极值的影响，降水极值能够解释土壤水分变化的 0-10%；研究区内约 54%的区域的土壤水分受到温度极值的影响，温度极值能够解释土壤水分变化的 0-8%。降水和温度极值对土壤水分的影响小于降水和温度平均值对土壤水分的影响。我们的研究结果还表明，降水对土壤水分的影响随着降水量、土壤水分和海拔的升高而降低。利用收敛交叉映射（CCM） 方法分别检测了北半球中低纬度地区和新疆地区土壤水分对降水的影响。 在北半球中低纬度地区， CCM 结果显示土壤水分对降水有显著的影响，土壤水分对降水的影响在滞后一个月时（τ = −1）最强，且在 4 个月之后显著减弱，这表明土壤水分对降水的强影响可以持续 3-4 个月。 在不同气候区土壤水分对降水的影响强度不同，随着干旱指数（AI） 增加， 土壤水分对降水的影响呈现出先增强后减弱的趋势，最强的影响出现在半干旱和半湿润地区（0.3 < AI < 0.6）。在新疆， CCM 结果显示土壤水分对降水有显著影响，且最强影响出现在 τ = 0 处， 且土壤水分对降水的强影响持续时间较短（一个月之内）。对比分析以上结果可以得出不同地区土壤水分对降水影响的持续时间存在差异， 气候越干旱， 土壤水分对降水影响的持续时间可能越短。利用降水和土壤水分数据分别计算了标准化降水指数（SPI）、 标准化土壤水指数（SSI）和多变量干旱指数（MSDI）三种干旱指数来监测和预测新疆的干旱。SPI、 SSI 和 MSDI 三种干旱指数的空间平均值能够较一致地捕获新疆历史上的干旱事件。然而，三种干旱指数在监测干旱发生的开始时间、持续性和强度等方面存在差异，与 SPI 相比， SSI 监测到的干旱事件开始发生的时间有所滞后，但SSI 监测到的干旱事件的持续时间比 SPI 长， MSDI 监测到的干旱事件的强度比SPI 和 SSI 更严重。干旱概率预测结果显示基于这三个干旱指数提前 1 个月进行预测时得到的结果最好，随着提前期的增加，预测能力逐渐衰退。基于 SPI 的预测在提前 1 个月时得到的结果较准确，而基于 MSDI 和 SSI 提前 2 个月和 3 个月进行干旱预测时仍可以得到较准确的结果。从空间上来看，基于 SPI、SSI 和 MSDI的预测在北疆比在南疆更准确。 此外，对于这三个干旱指数，预测值通常低估了实际监测到的干旱强度。 |
| Other Abstract | Soil moisture is an important parameter in the fields of hydrology, meteorology,and agricultural science. Its spatial and temporal distributions have strong effects onwater, energy, and biogeochemical balances. Study of soil moisture is essential forseveral applications, such as weather and climate forecasting, drought and floodprediction, agricultural productivity estimation, and so on. However, there are manydeficiencies and controversies in the study of soil moisture. Climate and vegetationare two main factors for soil moisture variation. Due to their complex interactions,quantifying individual effects of these factors on soil moisture variation remains achallenge. Through its impact on the exchanges of water and energy between theatmosphere and the land surface, soil moisture has an impact on climate processes.However, the effect of soil moisture on precipitation is hard to quantify due tocomplex atmospheric processes and has been strongly debated. Therefore, futurestudies are worthy to be carried out to explore these subjects further.Xinjiang is located in the Eurasian hinterland and features with extremely scarcewater resources and a fragile ecosystem. Xinjiang is one of the most sensitive regionsto climate change. The study of soil moisture in Xinjiang is helpful for local waterresources management, ecosystem protection, and the prevention of natural disasterssuch as drought. Based on multi-source soil moisture data, this dissertation hasdeveloped a complete framework for soil moisture research, in which a systematicstatement is made on spatiotemporal variation characteristics of soil moisture,influencing factors, and its applications in drought and precipitation prediction. Themain conclusions can be drawn as follows:In Xinjiang, the soil moisture showed a distinctive spatial pattern. The minimumsoil moisture appeared in the centers of Tarim and Junggar basins, while themaximum appeared between these two basins. Soil moisture generally increased withelevation. For different seasons, the spatially averaged soil moisture posed anincreasing trend during 1980-2016. The increasing rate of soil moisture in summerwas obviously higher than that in other seasons. For the growing season and summer,the increasing rate of soil moisture was higher after 2009 than before. In spring, soilmoisture showed a slightly decreasing trend during 1980-2006, and an increasingtrend after 2006. In autumn, soil moisture showed no obvious change during 1980-2006, and a significant increasing trend after 2006. The spatial variations of soilmoisture in different seasons are obviously different. For spring, pixels where soilmoisture exhibited significant change account for about 27.2% of the study area,which was larger than other seasons. In autumn, about 20% of the study area showedsignificant change. Except for spring, the increasing trend of soil moisture waspredominant in the study area. Summer was the most important season for theregional soil moisture change. Furthermore, soil moisture content and variation weredifferent in different land cover types. Soil moisture was highest in the forest(14.61%), followed by cropland (13.32%), grassland (13.02%), and shrubland(12.46%), and it was lowest in the bare land (10.15%). During 2001-2015, soilmoisture exhibited significant increasing trends only in grassland and bare land.A non-linear Granger causality framework was applied to quantify the individualeffects of precipitation, temperature, and vegetation on soil moisture variability inXinjiang. Precipitation, temperature, and vegetation contributed differently to soilmoisture. Precipitation had effects on soil moisture in about 91% of the study area andexplained 0-40% of the soil moisture variability. Temperature contributed 0-8.2% tosoil moisture variability in about 59% pixels. Vegetation had effects on soil moisturein about 44% of the area and explained 0-3.3% of the soil moisture variability. About87% of the study area was precipitation-limited during 1982-2015. The temperaturewas the primary driving of soil moisture variability for 10% of the study area. The laganalysis shows that the effects of precipitation and temperature on soil moisture wereimmediate and dissipated shortly, indicating that soil moisture has less resilience tothe variations of precipitation and temperature. In addition, precipitation andtemperature extremes also had effects on soil moisture. The precipitation extremeswere able to explain 0-10% of soil moisture variability in about 76% of the study area.The temperature extremes contributed 0-8% to soil moisture variability in about 54%of the study area. The effects of climatic extremes were weaker than that of thegeneral climate factors. Our results also reveal that the effects of precipitation on soilmoisture decreased with the increase of precipitation, soil moisture, and elevation.The Convergent Cross Mapping (CCM) method was applied to detect the effectsof soil moisture on precipitation in both the low- and mid-latitude regions of theNorthern Hemisphere and Xinjiang. In the low- and mid-latitude regions of theNorthern Hemisphere, the CCM results indicated significant causal effects of soilmoisture on precipitation. Soil moisture had the strongest effect on precipitation with a one-month lag (τ = −1), and the effect clearly decreased after four months, whichindicate that the strong effect of soil moisture on precipitation could last three to fourmonths. As the Aridity Index (AI) increased, the effects of soil moisture onprecipitation increased first and decreased afterward, with peaks in the semi-arid andsemi-humid areas (0.3 < AI < 0.6). In Xinjiang, the CCM results indicated significantcausal effects of soil moisture on precipitation, and the optimal effect of soil moistureon precipitation occurred with no time lag (τ = 0). The strong effect of soil moistureon precipitation only lasted for a short time (within one month) in Xinjiang. Thecomparisons of these results suggest that the persistence time in the effect of soilmoisture on precipitation was highly region dependent and proved to be shorter in aridthan in humid regions.Three drought indices, i.e., the Standardized Precipitation Index (SPI), theStandardized Soil-moisture Index (SSI), and the Multivariate Standardized DroughtIndex (MSDI), have been calculated from soil moisture and precipitation data andwere used to monitor and predict drought in Xinjiang. As a monitoring tool, SPI, SSI,and MSDI were able to capture severe historic drought events in Xinjiang. However,the spatial coverage, persistence, and severity of drought varied with different droughtindices. The SSI captured the onset of drought events with lags to the SPI, while itindicated longer drought persistence than SPI. The MSDI monitored more severedrought conditions than either SPI or SSI. As a prediction tool, the 1-month leadforecast based on these three drought indices was the most accurate, and thepredictive skill gradually decayed as the lead time increased. The SPI only showedpredictive skill at a 1-month lead time. The MSDI performed best in capturingdroughts at 1- to 2-month lead times, and the SSI was accurate up to 3-month leadtime. From a spatial perspective, predictions based on these three drought indiceswere more accurate in north Xinjiang than in south Xinjiang. In addition, for thesethree drought indices, the predicted ensemble median generally underestimated themonitored drought degree. |
| Subject Area | 自然地理学 |
| Language | 中文 |
| Document Type | 学位论文 |
| Identifier | http://ir.xjlas.org/handle/365004/15361 |
| Collection | 中国科学院新疆生态与地理研究所 研究系统 |
| Affiliation | 中国科学院新疆生态与地理研究所 |
| First Author Affilication | 中国科学院新疆生态与地理研究所 |
| Recommended Citation GB/T 7714 | 王云倩. 基于多源数据的新疆土壤水分时空变异特征研究[D]. 北京. 中国科学院大学,2019. |
| Files in This Item: | There are no files associated with this item. | |||||
Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.
Edit Comment