Knowledge Resource Center for Ecological Environment in Arid Area
基于 M-SEBAL 模型的复杂地表蒸散发的遥感估算 | |
其他题名 | Estimate the complex surface evapotranspiration based on the M-SEBAL model |
周彦昭 | |
出版年 | 2015 |
学位类型 | 硕士 |
导师 | 周剑 |
学位授予单位 | 中国科学院大学 |
中文摘要 | 蒸散发(Evapotranspiration,简写为ET)包括土壤蒸发和植被蒸腾,是水量平衡和能量平衡的重要组成部分。深入认识蒸散发,对于理解能-水循环的过程,加深陆表过程理解,有着十分重要的意义。特别是在干旱半干旱内陆河流域,蒸散发是该地区水分消耗的主要途径,对蒸散发的研究有助于了解能-水的空间分配,提高水资源的管理效率。因此如何准确地估算地表实际蒸散发成为了近30年来众多的学者不懈追求的目标。遥感由于其实时性、区域性和经济性等特点,使得基于遥感的蒸散发模型受到了广泛的关注。 目前常用的基于遥感的蒸散发模型有:地表能量平衡系统(SEBS)、陆地表面能量平衡算法(SEBAL)、改进的陆地表面能量平衡算法(M-SEBAL)、简化的地表能量平衡指数(S-SEBI)、MODIS蒸发比模型和双层蒸散模型(TTME)等。其中,基于地表辐射温度-植被指数特征空间的定标方案的M-SEBAL模型,不仅避免SEBAL 模型中冷热像元的人为主观选取造成的误差,同时温度-植被指数特征空间的定标方案,避免了单一冷热像元的定标方案引起的空间分布变形问题。然而目前对于M-SEBAL 模型的研究还不多,对于M-SEBAL 模型的在干旱半干旱区域的适应性以及其敏感性尚不清楚。 本文研究以2012年“黑河流域生态-水文过程综合遥感观测联合试验”(Heihe Watershed Allied Telemetry Experimental Research,简称HiWATER)中游专题试验“非均匀下垫面多尺度地表蒸散发观测试验”获取的地面气象和EC观测数据,针对M-SEBAL 模型展开一下几项研究: 1.评估SEBAL模型和M-SEBAL模型。本文使用2012年HiWATER试验获取的EC和气象观测数据,不同时期的3景ASTER 影像数据对SEBAL 模型和M-SEBAL 模型估算的感热通量、潜热通量和蒸发比进行了对比分析。结果表明:SEBAL模型在绿洲明显低估感热通量,高估潜热通量;在戈壁明显高估感热通量,低估潜热通量。M-SEBAL模型充分考虑不同下垫面地表辐射温度与植被覆盖度之间的关系,能很好地反映不同植被覆盖区域的湍流通量的异质性,估算黑河中游戈壁、绿洲蒸散发的精度明显高于SEBAL模型。 2.M-SEBAL模型的敏感性分析。利用2012年6-9月份、2013年4-5月份MODIS数据,采用单一参数扰动方法,综合分析M-SEBAL模型对地表特性、气象数据、冷热点性质等输入参量的敏感性,系统总结M-SEBAL模型参量不确定性特征。结果表明:M-SEBAL 模型最为敏感的参数为地表温度、空气温度和植被干点温度,较为敏感的参数是裸土干点温度、植被干点有效能量、植被覆盖度和裸土干点有效能量,一般敏感的参数是动量粗糙度和风速;不敏感的参数是地表比辐射率、反照率和相对湿度。 3.黑河中游整个生长季的蒸散发的时空分布。在明确M-SEBAL 模型敏感性参数基础上,使用MODIS影像基于M-SEBAL模型和参考蒸发比的时间尺度扩展方案,估算出2012年6-9月份和2013年4-5月份的黑河中游整个生长季的蒸散发,并使用地面观测数据对于M-SEBAL 模型和参考蒸发比的时间尺度扩展方案进行了验证和评估;并与ETMonitor 蒸散发产品进行了交叉对比分析。结果表明:(1)M-SEBAL模型和参考蒸发比的时间尺度扩展方法可以较为准确地估算出不同下垫面的蒸散发。其平均绝对偏差为0.39 mm,均方根误差为0.54 mm,纳什系数为0.95,平均绝对比例偏差为31.57%,相对偏差为2.60%;但在植被覆盖度低时,如戈壁和植被早期,由于降雨或灌溉的影响,M-SEBAL 模型和基于参考蒸发比的时间尺度扩展方案会产生较大的误差。(2)M-SEBAL 模型估算结果与ETMonitor估算结果趋势相对一致的,两者的直方图的分布趋势相对一致的,但两者的相对频率的大小存在着较大的差别,其中M-SEBAL模型低值高峰相对频率低于ETMonitor 模型。(3)M-SEBAL 模型估算结果与ETMonitor估算结果在生长季早期和末期相关系数较低,随着植被的生长,相关系数逐步升高。绿洲区域两个模型的相关系数较高,偏差较小,均方根误差较小,两个模型的差别不大;在戈壁沙地区域,两个模型的相关系数较低,偏差较大,均方根误差较大,两个模型差别较大。(4)黑河中游蒸散发总量的空间分布差异较大。其中水体的蒸发量最大,在700 mm以上;绿洲的蒸散发次之,在500 mm-650 mm之间。其中张掖绿洲的蒸散发最高,蒸散发总量在600 mm-650 mm;临泽绿洲次之,蒸散发总量在550 mm-600 mm之间;高台绿洲最小蒸散发总量在500 mm-550 mm之间。绿洲周边的荒漠、沙漠等区域蒸散发最低,整个生长季的蒸散发总量在0-100 mm之间; 4.不同土地利用蒸散发的时空变化。使用2011年黑河中游的土地利用/覆被数据,统计分析了不同月份、不同下垫面类型蒸散发的时空分布及其主要影响因子。结果表明:不同下垫面蒸散发的差异较大。(1)其中水体的日均蒸散发为3.13 mm/d,总平均蒸散发量为563.98 mm;林地的日平均蒸散发为1.45 mm/d,总平均蒸散发量为260.99 mm;耕地日平均蒸散发为2.52 mm/d,总平均蒸散发量为454.06 mm;草地日平均蒸散发为1.31 mm/d,总平均蒸散发量为236.75 mm;戈壁和沙地的日均蒸散发分别为0.64 mm/d和0.42 mm/d,总平均蒸散发量为115.16 mm和76.56 mm。不同下垫面蒸散发大小顺序为:水体>耕地>林地>草地>戈壁>沙地。(2)黑河中游区域蒸散发随时间的变化十分明显。黑河中游在7、8月份的蒸散发最大,6月份次之,4、5和9月份最低。 关键字:蒸散发;M-SEBAL 模型;黑河中游;温度-植被指数特征空间 |
英文摘要 | Evapotranspiration, including evaporation from soil and transpiration from vegetation, is a important item of heat balance and water balance. It also plays a important role in understanding the energy-water cycling and studying the land surface system, especially in the arid and semi-arid inland river basin,where Evapotranspiration is the main way of water consumption.So nearly 30 years how to accurately estimate evapotranspiration become lots of scholars’ goal. Due to the characteristics such as real-time, regional and economical, remote sensing technology is recognized as the only viable means to map regional scale patterns of ET on the Earth’s surface.With the advent of remote sensing technology, estimating evapotranspiration based on remote sensing draws widespread attention/become a reserch hotpoint. In recent years, there are many evapotranspiration model based on remote sensing, such as the Surface Energy Balance System (SEBS), the Surface Energy Balance Algorithm for Land (SEBAL), the Modified Surface Energy Balance Algorithm for Land (M-SEBAL), the Simplified Surface Energy Balance Index (S-SEBI), a MODIS global terrestrial evapotranspiration algorithm, a Hybrid dual-source scheme and Trapezoid framework–based Evapotranspiration Model (HTEM) and so on. Among those models, the M-SEBAL model, which uses a trapezoidal framework of the Ts-VI space that involves in the relationship between vegetation fraction of underlying surface and the surface temperature, to improve the SEBAL model estimation results, avoids errors of the subjectivity in extreme pixels selection and reduce ambiguity in flux estimation. However, Furture study about M-SEBAL model such as the adoption in the arid and semi-arid areas and sensitivity of M-SEBAL model has been not reported. In this thesis, the middle reaches of Heihe river basin is selected as study area to evaluates the performance of the SEBAL model and the M-SEBAL model, quantify uncertainty of the M-SEBAL model and estimate evapotranspiration by means of the M-SEBAL model. All data is acquired in the Multi-Scale Observation Experiment on Evapotranspiration over heterogeneous land surfaces of The Heihe Watershed Allied Telemetry Experimental Research (HiWATER-MUSOEXE). The methods and results can be summerised as the following: 1. Evaluates the performance of the SEBAL model and the M-SEBAL model. Three Advanced Space borne Thermal Emission and Reflection Radiometer (ASTER) images of different growth stages in 2012 is used in this study. The result shows that SEBAL model is to underestimate sensible heat flux in the oasis region, while the overestimating latent heat flux. The M-SEBAL model can capture the strong land surface heterogeneities because of the involved Ts-VI space. So It can provide more reliable and accuracy ET estimation about Gobi and oasis than SEBAL model. 2. Quantify uncertainty of the M-SEBAL model. The M-SEBAL model simulates evapotranspiration using 36 MODIS images including 24 images in the year of 2012 and 12 images in 2013 at a spatial resolution of 250 m × 250 m. A total of 36 sets of reference values (initial values for running M-SEBAL) are derived from 36 cloud-free MODIS images, spanning a broad range of atmospheric, soil moisture, and land cover conditions. These reference values are used to initialize the M-SEBAL H algorithm. M-SEBAL sensitivity is performed by varying each variable/parameter under a given set of reference values at 2% steps (perturbation) within ±10%. Results of sensitivity analysis indicate that the H estimates from M-SEBAL are most sensitive to temperatures of surface, air and the fully vegetated surface with the largest stress and are less sensitive to temperatures of the bare surface with the largest water stress, vegetation cover and available energy of the fully vegetated surface and the bare surface with the largest stress. The H estimates from M-SEBAL are least sensitive to emissitivity,albedo and relative humidity. 3. Estimate evapotranspiration of the middle reaches of Heihe river basin. The M-SEBAL model simulates instantaneous evapotranspiration using 36 MODIS images including 24 images in June-September of 2012 and 12 images in April-May of 2013 at a spatial resolution of 250 m × 250 m. Instantaneous evapotranspiration values are up-scaled to daily values by assuming that reference evaporative fraction is a constant throughout a day. For days without accessible remote sensing image, the daily evapotranspiration is estimated by linearly interpolating the reference evaporative fraction value over periods between two consecutive images and multiplying by the reference evaporative for each day. The results of M-SEBAL model have been successfully tested against numbers of in situ measurements and productions of the ETMonitor model. Results show that (1) the M-SEBAL model and up-scaled method based on reference evaporative fraction can accurately estimate the evapotranspiration over different overlying types, with the average absolute bias of 0.39 mm, the root mean square error of 0.54 mm, the Nash coefficient of 0.95, Mean absolute percentage difference of 31.57% and the relative error of 2.60%. However, duoe to the irrigation or precipitation, the performance of the M-SEBAL model and up-scaled method based on reference evaporative fraction is poor. (2) The spatial pattern of the M-SEBAL model and the ETMonitor model is similar. The distribution of these two histogram is relatively consistent, but there are lager difference between their relative frequency. And the relative frequency of the lower evapotraspiration values of the M-SEBAL model is lower than that of the ETMonitor model. (3) The correlation coefficients for the results of these two models are lower at the growing seasons of early than that at the middle growing seasons. And ET values in the oasis regions,with smaller bias and the root mean square error, are more similar than that in the Gobi regions. (4) The spatial distribution of the total ET during the growing season vary significantly. Water body has the largest ET with the mean value of 700 mm, while Gobi has the minimum ET ranging from 0 to 100 mm. Following water body, irrigated oasis has the second largest mean ET ranging from 500 mm to 650 mm. Among the oasis regions, Zhangye oasis has the largest ET ranging from 600 mm to 650 mm, while Gaotai oasis has the minimum ET ranging from 500 to 550 mm. Following Zhangye oasis, Linze oasis has the second largest mean ET ranging from 550 mm to 600 mm. 4. Spatial and temporal patterns of evapotranspiration over different landuse type. The statistic value of ET over different land use types shows that the ET value over different underlying surface has large difference. (1) The mean ET value of water bady, woodland, farmland, grassland, Gobi and desert are 3.13 mm/d, 1.45 mm/d, 2.52 mm/d, 1.31 mm/d, 0.64 mm/d and 0.42 mm/d respectively. The total ET value during the growing season of water bady, woodland, farmland, grassland, Gobi and desert are 563.98 mm/d, 260.99 mm/d, 454.06 mm/d, 236.75 mm/d, 115.16 mm/d and 76.56 mm/d respectively. (2)ET also shows obvious monthly changes. The value is the greatest in July and August, with ET decreasing a little in June, and then decreasing rapidly in April, May and September. Key words: evapotranspiration; M-SEBAL model; the middle reaches of Heihe river basin; Ts-VI; |
中文关键词 | 蒸散发 ; M-SEBAL 模型 ; 黑河中游 ; 温度-植被指数特征空间 |
英文关键词 | evapotranspiration M-SEBAL model the middle reaches of Heihe river basin Ts-VI |
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 | 中国科学院西北生态环境资源研究院 |
资源类型 | 学位论文 |
条目标识符 | http://119.78.100.177/qdio/handle/2XILL650/287528 |
推荐引用方式 GB/T 7714 | 周彦昭. 基于 M-SEBAL 模型的复杂地表蒸散发的遥感估算[D]. 中国科学院大学,2015. |
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 |
[周彦昭]的文章 |
百度学术 |
百度学术中相似的文章 |
[周彦昭]的文章 |
必应学术 |
必应学术中相似的文章 |
[周彦昭]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。