基于CERES-Maize与PROSAIL模型耦合的冠层反射率模拟分析

发布时间：2018-08-17 14:41

【摘要】：植被冠层是生态系统中物理和化学进程的重要场所之一,植被冠层辐射传输模型理论奠定了植被遥感界的理论基础。LAI(全称,叶面积指数,下同)是表征陆地表面冠层状况的重要指标,作为辐射传输模型的重要特征参量,对模型模拟和反演精度至关重要。地面LAI测量方法主要有直接测量法和间接测量法两种,适合小区域的LAI估算。基于作物生长模型模拟可获取作物全生长期的LAI,时间连续性强,目前将遥感与作物生长模型进行耦合的研究逐渐增多,但研究的重点多集中在将遥感数据反演、推算的一些宏观信息嵌入作物模拟模型或校正有关参数,目的多为监测作物长势和估产,而针对遥感模型在特定生育期模拟结果的评价和误差来源分析的相关研究鲜少出现。本文在对PROSAIL模型输入参数的进行敏感性分析的基础上,首先对CERES-Maize玉米生长模型进行标定,得到最优的作物遗传参数组合,模拟得到抽穗期LAI变化特征。然后结合分光光度计对特定生育期玉米的理化参数(叶绿素、类胡萝卜素)观测试验,建立了PROSAIL模型输入参数数据库,模拟得到了玉米特定生育期的多角度冠层光谱数据,最后利用四维塔吊观测平台观测的不同时间点的多角度光谱信息和不同穗数下的多角度光谱信息对模型模拟结果进行对比,评价模型模拟精度,确定误差来源。主要方法和结论如下:(1)耦合CERES-Maize模型PROSAIL辐射传输模型模拟玉米抽穗期冠层反射率的变化情况,结果显示冠层反射率呈随时间的推移不断下降。结合多角度冠层反射率观测试验数据验证结果表明:五个时间点的相同观测方位角和天顶角的冠层反射率观测值与模型模拟值虽总体变化趋势一致,但实测值高于模拟值,尤其在可见光波段(绿光、红光)表现明显,但在近红外波段差异不明显。(2)参数敏感性分析显示:C(6(7的变化对绿光波段的反射率影响最大;褐色素对红光波段约700nm附近的反射率值影响较大;干物质量对近红外波段、短波近红外波段和中红外波段的反射率影响较大;等效水厚度变化主要影响900nm以后的少数区间的冠层反射率值;叶片结构和热点系数的变化会对全波段反射率存在影响;LAI对可见光波段的反射率影响较大;C(6的变化对反射率的影响较小,主要集中在450nm附近。(3)利用河北怀来遥感试验站2013年地面观测数据及田间试验资料,对CERES-Maize模型有关生长发育的品种遗传特性参数进行了标定,同时利用2014、2015年的观测数据对模型模拟结果进行验证,两年的LAI模拟均较为准确,可用于玉米长时间序列模拟。结合2016年站内气象站的观测数据模拟得出:玉米全生育期LAI分布区间为0.01~5.48,抽穗期的LAI在4.75左右,与实测数据吻合度较高,说明该模型对玉米抽穗期LAI模拟效果较好。对2016年玉米抽穗期的地面观测数据进行计算和分析发现:等效水厚度、干物质量、叶倾角、叶绿素含量、类胡萝卜素含量等所有参数值均波动较小,基本保持不变。(4)采用人工剪穗的方法探究雄穗数量对冠层反射率的影响,结果表明:无穗的冠层反射率最高,全穗的冠层反射率最低。对比五个穗梯度的观测值和五个时间点的模拟和实测误差发现,雄穗是产生误差的主要原因。
[Abstract]:Vegetation canopy is one of the important sites of physical and chemical processes in ecosystem. The theory of radiative transfer model of vegetation canopy lays a theoretical foundation for vegetation remote sensing. Ground LAI measurement methods mainly include direct measurement and indirect measurement, which are suitable for LAI estimation in small areas. Based on crop growth model simulation, LAI of whole crop growth period can be obtained, and the time continuity is strong. In this paper, the input parameters of the PROSAIL model are sensitized to the input parameters of the PROSAIL model. On the basis of perceptual analysis, the CERES-Maize maize growth model was calibrated to obtain the optimal combination of crop genetic parameters, and the LAI variation characteristics at heading stage were simulated. Finally, the model simulation results were compared by using the multi-angle spectral information at different time points and under different spike numbers from the four-dimensional tower crane observation platform to evaluate the model simulation accuracy and determine the error sources. The following: (1) The coupling CERES-Maize model PROSAIL radiation transfer model simulated the change of canopy reflectance at heading stage of maize, and the results showed that the canopy reflectance decreased with time. Although the observed emissivity is consistent with the simulated value, the measured value is higher than the simulated value, especially in the visible band (green light, red light), but the difference is not obvious in the near-infrared band. (2) Parameter sensitivity analysis shows that the change of C (6 (7) has the greatest influence on the reflectivity of the green band; brown pigment has the greatest influence on the red band about 700 nm. Near reflectance has a great influence; dry matter has a great influence on near-infrared reflectance, short-wave near-infrared reflectance and mid-infrared reflectance; the change of equivalent water thickness mainly affects the canopy reflectance in a few areas after 900 nm; the change of blade structure and hot spot coefficient has an effect on the full-band reflectance; LAI has an effect on visible light; C (6) had a little effect on the reflectance, and mainly concentrated around 450 nm. (3) Based on the ground observation data and field experiment data of Huailai Remote Sensing Station in Hebei Province in 2013, the genetic parameters of varieties related to the growth and development of C ERES-Maize model were calibrated, and the observation data of 2014 and 2015 were used. The simulation results show that the two-year LAI simulation is more accurate and can be used to simulate the long time series of maize. Combined with the observation data of meteorological stations in 2016, the LAI distribution range of Maize in the whole growth period is 0.01-5.48, and the LAI at heading period is about 4.75, which shows that the model is in good agreement with the measured data. The results showed that all parameters, such as equivalent water thickness, dry matter mass, leaf inclination, chlorophyll content and carotenoid content, fluctuated slightly and remained basically unchanged. (4) Artificial pruning was used to explore the effect of male panicle number on canopy reflectance. The results showed that the canopy reflectance of panicle-free was the highest and that of whole panicle was the lowest.
【学位授予单位】：石河子大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：S513;S127

【参考文献】