Knowledge Resource Center for Ecological Environment in Arid Area
| 非参数化蒸散计算方法的遥感反演适用性研究 | |
| 其他题名 | Retrieval of Surface Evapotranspiration with Nonparametric Approach: Applicability and Accuracy Assessment |
| 潘鑫 | |
| 出版年 | 2016 |
| 学位类型 | 博士 |
| 导师 | 刘元波 |
| 学位授予单位 | 中国科学院大学 |
| 中文摘要 | 蒸散包括陆面蒸发和植被蒸腾，它是水文循环中与水量和热量具有密切关系的重要物理过程，同时蒸散过程中消耗的潜热也是地-气之间能量转移的重要组成部分。蒸散的研究对定量化水量平衡，规划管理水资源利用有重要意义，对生态、气象、水文、农林业等领域的研究发展起推动作用。传统蒸散估算方法，能较好地反映蒸散的物理机制，但是基本都是气象点位尺度的估算，无法扩展到下垫面复杂多样的区域尺度。由于遥感的时间连续性和空间广阔性，遥感反演区域蒸散使得传统方法可以从点位尺度扩展到区域尺度，因而逐渐成为监测区域蒸散的研究热点。近年来，基于遥感技术和传统蒸散估算方法发展出很多区域蒸散遥感方法，但这些方法大多存在阻抗参数化复杂，地-气传输系数不唯一等缺点，使得区域遥感反演方法计算复杂，反演结果误差溯源困难，限制了区域蒸散遥感反演方法的进一步发展。非参数化（NP）方法作为一种新近提出的估算地表蒸散的物理方法，避免了对阻抗的参数化，具有明确的物理意义，可以避免传统方法的种种缺陷。目前基本只在气象点位尺度上应用，但NP方法在大范围长时序反演区域蒸散方面的适用性仍需要进一步研究。因此，本文基于非参数化（NP）方法发展了全空条件下多时间尺度区域蒸散遥感反演模型算法，该模型算法以MODIS（the MODerate-resolution Imaging Spectroradiometer）遥感产品及CLDAS（China Meteorological Administration Land Data Assimilation System）陆面数据同化产品为模型输入，可分别对晴空和云空条件下的区域蒸散进行估算。利用本文算法，分别对干旱区（黑河流域张掖地区）和湿润区（鄱阳湖流域）范围内的瞬时蒸散和长时间尺度蒸散进行反演，并使用干旱区6个站点和湿润区3个站点处的地表观测数据和MOD16遥感蒸散产品，对反演结果进行直接检验和交叉检验。通过对晴空和云空条件下的两个区域蒸散反演结果误差溯源，分析不同条件下算法的适用性，探讨反演结果精度可能的改进途径。本文的主要研究结果和基本结论如下：1. 基于NP方法遥感反演了全空条件下2012年6月至9月的张掖地区和2014（2015）年3月至5月鄱阳湖流域的区域蒸散，结果表明，从空间分布上看，蒸散在干旱区符合沙漠-绿洲的空间分布，湿润区符合湖区-非湖区的水体-草地-林地的空间分布特征。从瞬时蒸散反演结果上看，晴空条件下的蒸散数值明显高于云空条件下的。干旱区晴空条件下植被区站点（湿地、菜地、果园）处的蒸散数值高于非植被区站点（村庄、戈壁、沙漠）处，云空条件下也是如此。湿润区，全空条件下，非湖区（森林）站点处蒸散高于湖区站点（湖面、草地）处。2. 基于NP方法遥感反演的蒸散，在全空条件下，瞬时蒸散总体上被低估，相对误差在4% ~ 40%之间，R2在0.04 ~ 0.82之间，湿润区站点（相对误差4% ~ 40%）精度优于干旱区站点（相对误差20% ~ 30%）。晴空条件下，植被区站点处蒸散反演精度优于非植被区站点处，云空条件下则相反。在时间序列蒸散变化特征上，瞬时反演结果与实测值具有很好的一致性。与MOD16蒸散产品的交叉检验结果表明，本文算法反演的8天尺度蒸散与MOD16蒸散产品相比，二者在湿润区比在干旱区更加接近。8天尺度的反演结果在体现蒸散时间变化趋势方面，优于MOD16产品，可以模拟蒸散时序变化的峰谷，尤其是在干旱区。3. 通过对蒸散反演结果误差来源的分析，表明，在干旱区，非植被区站点蒸散反演的主要误差源是地表净辐射、表面热通量的反演误差，及地表温度和气温遥感产品输入误差；在植被区主要是NP方法误差，云空时还受地表温度输入误差的影响。在湿润区，蒸散反演结果在湖区站点处主要误差源是表面热通量的反演误差，在非湖区主要是地表温度的输入误差和NP方法误差。针对反演结果的误差来源，可通过改进不同条件下主要误差源的输入精度和改进NP方法的低估提高算法的反演精度。本文建立了基于MODIS的NP方法遥感全空多时间尺度蒸散反演算法，选取两个气候、下垫面条件差异很大的试验区（干旱区/湿润区）进行区域蒸散反演，并对反演结果进行精度验证和误差来源分析，表明了蒸散反演结果在干旱区和湿润区范围内精度可靠，反演算法适用，揭示了反演结果的误差来源，探讨了反演改进的途径。研究结果为基于NP方法使用长时间序列多源遥感数据反演区域蒸散研究提供了理论基础和实践经验。未来可通过对反演结果的改进及更多类型区域处的验证等方法进一步拓展NP方法遥感反演区域蒸散的应用。 |
| 英文摘要 | As a component in hydrologic cycle, surface evapotranspiration (ET) including evaporation from land and transpiration from vegetation is closely related to water and energy. ET is interchangeable to the associated latent energy, and it is a key parameter that regulates regional hydrological and climatic processes. Accurate estimation of evapotranspiration is important for quantification of water balance and management of regional water resources. Moreover, it promotes the sciences of meteorology, hydrology, argriculture and forestry. Traditional site-scale methods for ET estimation are generally based on physical mechanism, however, they are difficult to accommodate complex underlying surfaces. For regional ET estimation, remote sensing technology has the advantage to extend site-scale algorithm to a wide area. As a result, more attentions are currently being paid on ET estimation from remotely sensed data. Although there have been recently substantial studies to combine remote sensing with traditional methods in an effort to estimate ET, the complicated parameterization of resistence and the inconsistent coefficient are generally unavoidable. Therefore, these methods are computationally inefficient and the associated errors cannot be easily tracked. Overall, the development of these methods is largely limited.A nonparametric approach (NP) has been recently proposed to estimate ET. This approach has specific physical background and provides a simple analytical form. More importantly, it avoids the parameterization processes. Currently, the approach has only been applicated at site scales with relatively satisfactory accuracy, yet not examined using remotely sensed data at the large scale in the long temporal series. Based on the NP approach, a retrieval algorithm is proposed for ET estimation under all sky using MODIS (MODerate-resolution Imaging Spectroradiometer) and CLDAS (China Meteorological Administration Land Data Assimilation System) datasets. In addition, the algorithm can also estimate the ET under clear sky and cloudy sky. The instantaneous and daily/monthly ETs are estimated for a typical arid region (Zhangye, Heihe basin) and a typical moist region (Poyang Lake basin). Subsequently, the estimated ETs are validated with surface measurements and MOD16 products and the error sources quantified. Then the applicability of the NP approach is analyzed. Lastly, the possible improvements to the NP approach are discussed. The main conclusions are as follows:1. ET is estimated in the Zhangye region during June - September, 2012 and in the Poyang lake basin during March - May in 2014 and 2015. The ET distribution is in good accordance with the oasis-desert distribution in the arid region, and in good accordance with the water-grassland-forest (lake-land) distribution in the moist region. For the instantaneous estimation, ET is higher under clear sky than cloudy sky. In the arid region, ET is higher over vegetated sites (wetland, vegetable and orchard) than non-vegetated sites (village, Gobi and desert) under both clear and cloudy sky. In the moist region, ET is higher over the non-lake area (forest) than the lake area (water and grassland).2. ET is generally underestimated under all sky with relative error (RE) of 4% ~ 40% and R2 of 0.04 ~ 0.82. ET estimation is more accurate in the moist region (RE: 4% ~ 40%) than the arid region (RE: 20% ~ 30%). ET estimation is more (less) accurate over vegetated than nonvegetated sites under clear (cloudy) sky. On temporal scales, instantaneous ET estimates are in good accordance with surface measurements. Additionally, a cross validation between ET estimation and MOD16 reveals smaller differences over the moist region than the arid region. Our ET estimation can reveal temporal variations in ET and show featured peak and valley ET values, especially in the arid region, which cannot be revealed from MOD16. 3. In the arid region, ET error mainly results from the errores of MODIS products or the net radiation and soil heat flux in non-vegetated region. ET error is mainly attributed to the self-underestimation of NP approach in the vegetated region. Moreover, land surface temperature error also affects ET under cloudy sky. In the moist region, surface heat flux error dominates ET estimation error in the lake region, and land surface temperature error and the self-underestimation of NP approach affect the estimation in the non-lake region. Therefore, ET estimation can be more accurate by improving both the input data and the NP approach. Based on the NP approach and MODIS products, we develop a remote sensing algorithm for ET estimation at multi-temporal scales under all sky. ET is estimated for two typical regions with different climates and underlying surfaces. Then, we validate the ET estimation and investigate the error sources. All results support the satisfactory accuracy of ET estimation and the applicability of the algorithms. Besides, the error sources of ET estimation have been ascertained. The potential improvements have been discussed. Our study provides basic theory and practical experience for regional LE estimation using the NP approach and long-term multi-source remote sensing data. In the future, to expand the application of the algorithm, we need to improve the ET eatimation and validate the ET estimation in more typical regions. |
| 中文关键词 | NP方法 ; 遥感反演 ; 蒸散 ; 精度评价 ; 适用性 |
| 英文关键词 | Nonparametric evaporation approach remote sensing retrieve evapotranspiration accuracy assessment algorithm applicablity |
| 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 | 中国科学院南京地理与湖泊研究所 |
| 资源类型 | 学位论文 |
| 条目标识符 | http://119.78.100.177/qdio/handle/2XILL650/287733 |
| 推荐引用方式 GB/T 7714 | 潘鑫. 非参数化蒸散计算方法的遥感反演适用性研究[D]. 中国科学院大学,2016. |
| 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 |
| [潘鑫]的文章 |
| 百度学术 |
| 百度学术中相似的文章 |
| [潘鑫]的文章 |
| 必应学术 |
| 必应学术中相似的文章 |
| [潘鑫]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。
