小时步长森林碳循环模型（BEPS）参数优化及应用研究
发布时间：2018-08-29 14:37
【摘要】：森林是陆地生态系统的主体,森林碳循环在陆地碳循环系统中占有重要地位。本研究在站点尺度对小时步长碳循环模型BEPSHourly进行了模型优化和验证,并在区域尺度上探究了小时步长碳循环模型与日步长碳循环模型的协同应用方法。首先在站点尺度(通量塔有效覆盖区)收集BEPSHourly模型的驱动及验证数据,利用迭代方法对主要光合作用参数—最大羧化速率(Vc max)和最大电子传递速率Umax)进行参数优化；根据东北地区不同林分类型各器官生物量分配的研究结果,对BEPSHourly模型中不同林分类型的生物量分配参数进行优化；以辐射传输模型为切入点,考虑林冠层二次透过现象及多次反射的影响,优化辐射传输模型,以期在更准确求得净辐射的基础上,提高BEPSHourly模型对潜显热通量的模拟能力；利用数据同化算法—集合卡尔曼滤波(EnKF)对BEPSHourly模型预测的土壤湿度进行校正,减少计算误差随着时间的累积；利用优化后的BEPSHourly模型进行土壤温度模拟及验证分析、地表雪深度变化模拟及验证分析和冠层温度模拟；对NPP的主要影响因子—叶面积指数和气象因子(温度、降水、风速、太阳总辐射、相对湿度)进行定量敏感性分析。在区域尺度上,收集BEPSDaily模型的区域驱动数据,由于现有的MODIS LAI产品无法满足应用需求,开展适用于中小区域尺度的时间序列林地LAI快速估测方法研究；研究BEPSDaily和BEPSHourly各自的特点及适用性,重点分析了主要光合作用参数Vcmax、Jmax由BEPSHourly模型向BEPSDaily模型进行传递的可靠性,在参数传递可靠性分析的基础上制定了BEPSHourly模型与BEPSDaily模型协同应用方法；基于优化后的BEPSHourly在站点尺度模拟和分析GPP、NPP的日内变化规律,并基于优化后的区域BEPSDaily模型进行区域NPP估测和碳源／汇空间分析。综上,研究结果表明：1、对于东北落叶阔叶林,当Vc max为41.1 μmol·m-2·s-1、Jmax 为 82.8 μmol·m-2·s-1时,基于BEPSHourly模拟的2011年逐日GPP与观测数据进行比较的RMSE最小,为1.10 g C·m-2·d-1,R2最高,为0.95。经过光合作用参数Vc max和Jmax优化后,BEPSHourly模型能更能好地模拟GPP的季节变化。2、阔叶林叶子占地上生物量的4%,枝干占地上生物量的96%；混交林叶子占地上生物量的5%,枝干占地上生物量的95%；针叶林叶子占地上生物量的6%,枝干占地上生物量的94%；地下生物量与地上生物量存在显著的线性关系,将模型中地下地上生物量比率常数优化为两者的线性关系表达式。3、考虑林冠层二次透过现象及多次反射的影响,模型优化后模拟的潜热通量的R2由0.769提高到了0.792, RMSE由50.77 W／m2减少到47.84WW／m2,显热通量的R2由0.684提高到了0.705, RMSE由48.42 W／m2减低为45.86 W／m2,潜显热通量模拟值与实测值存在显著的相关关系；潜显热月平均日内变化均为单峰曲线,中午达到最大值,晚上及凌晨较低；相比显热通量,潜热通量的季节性变化较为显著,在生长季潜热通量远远高于非生长季,与植被的生长呈现伴随的正相关关系,显热通量在植被旺盛期呈现低峰状态。4、数据同化前,模型模拟与站点观测的土壤湿度的RMSE为0.1198,间隔0.5h引入观测数据进行同化后,RMSE降低到0.0293,模拟结果得到较明显的改善；同化数据引入频率越高,同化后模拟的土壤湿度与观测土壤湿度的RMSE越小,即同化模拟效果越好,当同化频率为15d时,同化系统对土壤湿度的模拟能力与未经同化的模型基本相当。5、NPP与LAI呈现非线性正相关关系,敏感度S为0.292,敏感等级为Ⅲ; NPP先随着温度的增加而增加,后随着温度的增加而减少,其敏感度S为0.594,敏感等级为Ⅳ; NPP随着降水量的增加基本保持不变,说明在目前降水量降低30%到增加30%的范围内未包含对植被生长有抑制或者促进作用的降水量范围,敏感度S为0.0005,敏感等级为Ⅰ; NPP与太阳总辐射呈现非线性正相关关系,敏感度S为0.310,敏感等级为Ⅲ; NPP随着风速的增加基本保持不变,敏感度S为0.015,敏感等级为Ⅰ; NPP与相对湿度呈现非线性正相关关系,敏感度S为0.159,敏感等级为Ⅱ。6、本研究提出了中小区域逐日LAI快速估测方法,利用MODIS LAI产品提取LAI归一化生长曲线,基于传统的遥感统计模型估计LAI最大值,将两者简单相乘估测时间序列LAI。与实测值进行比较,阔叶林2样地的RMSE分别为0.40、0.49,混交林样地RMSE为0.59,此方法可简单、快速为中小区域尺度的其他研究提供有效的时间序列LAI数据。7、BEPSHourly模型优化的主要光合作用参数可以直接引入BEPSDaily模型,使其达到较高的模拟能力；GPP、NPP的季节性变化较为显著,与植被的生长和温度的增加呈现伴随的正相关关系；在区域尺度,利用优化后的BEPSDaily模型进行区域NPP估测,2011年帽儿山地区GPP、NPP单位面积平均值分别为1265.56 g C·m-2·a-1、628.40 gC.m-2·a-1；不同林分类型的初级生产力为阔叶林混交林针叶林。
[Abstract]:Forest is the main part of terrestrial ecosystem, and forest carbon cycle plays an important role in terrestrial carbon cycle system. This study optimized and validated the hourly carbon cycle model BEPS Hourly at site scale, and explored the synergistic application of hourly carbon cycle model and daily carbon cycle model at regional scale. Firstly, the driving and validating data of BEPS Hourly model were collected at site scale (flux tower effective coverage area), and the main photosynthetic parameters, maximum carboxylation rate (Vc max) and maximum electron transfer rate (Umax), were optimized by iterative method. The biomass allocation parameters of different forest types in BEPS Hourly model were optimized, and the radiation transfer model was used as a breakthrough point to optimize the radiation transfer model considering the secondary canopy penetration and multiple reflectance, so as to improve the simulation ability of BEPS Hourly model for latent sensible heat flux on the basis of more accurate calculation of net radiation. The data assimilation algorithm-EnKF was used to correct the soil moisture predicted by BEPS Hourly model to reduce the cumulative error with time; the optimized BEPS Hourly model was used to simulate and validate the soil temperature, and the change of snow depth was simulated and validated, and the canopy temperature was simulated. Leaf area index (LAI) and meteorological factors (temperature, precipitation, wind speed, total solar radiation, relative humidity) were quantitatively analyzed. At the regional scale, the regional driving data of BEPS Daily model were collected. Because the existing MODIS LAI products could not meet the application requirements, the time series suitable for small and medium-sized regional scale were developed. Study on the method of fast LAI estimation in woodland; study the characteristics and applicability of BEPS Daily and BEPS Hourly, especially analyze the transfer reliability of main photosynthetic parameters Vcmax, Jmax from BEPS Hourly model to BEPS Daily model. On the basis of reliability analysis of parameter transfer, the BEPS Hourly model and BEPS Daily model are established. Based on the optimized BEPS Hourly, the intraday variation of GPP and NPP was simulated and analyzed at the site scale, and the regional NPP estimation and carbon source/sink spatial analysis were carried out based on the optimized BEPS Daily model. When L.m-2.s-1 was used, the RMSE of daily GPP based on BEPS Hourly simulation was the smallest, 1.10 g C.m-2.d-1, R2 was the highest, 0.95. After optimization of photosynthetic parameters Vc Max and Jmax, the BEPS Hourly model could better simulate the seasonal variation of GPP. 2. Broadleaf forest accounted for 4% of the total biomass and branches accounted for 4% of the total biomass. 96% of the total biomass, 5% of the total aboveground biomass, 95% of the total aboveground biomass, 6% of the total aboveground biomass, 94% of the total aboveground biomass, and 96% of the total aboveground biomass were mixed forest leaves and branches. Formula 3. Considering the secondary canopy penetration and multiple reflections, the latent heat flux of the model was increased from 0.769 to 0.792, RMSE from 50.77 W/m2 to 47.84 W W/m2, sensible heat flux from 0.684 to 0.705, RMSE from 48.42 W/m2 to 45.86 W/m2, and latent heat flux from simulated value to measured value. There is a significant correlation; the monthly mean diurnal variation of latent sensible heat is a single peak curve, reaching the maximum at noon, lower at night and early morning; compared with sensible heat flux, the seasonal variation of latent heat flux is more significant, in the growing season is much higher than the non-growing season, and has a positive correlation with vegetation growth, sensible heat flux. Before the data assimilation, RMSE was 0.1198, 0.5 h interval, RMSE was reduced to 0.0293, and the simulation results were significantly improved. The higher the frequency of assimilation data, the simulated soil moisture and the observed soil moisture after assimilation. The smaller the RMSE, the better the assimilation simulation effect. When the assimilation frequency was 15 days, the ability of the assimilation system to simulate soil moisture was almost the same as that of the non-assimilation model. The sensitivity S was 0.594, the sensitivity grade was IV; NPP remained basically unchanged with the increase of precipitation, indicating that the range of precipitation which could inhibit or promote vegetation growth was not included in the range of 30% to 30% precipitation, the sensitivity S was 0.0005, the sensitivity grade was I; NPP and solar radiation showed a nonlinear positive phase. The sensitivity S is 0.310, the sensitivity grade is III; the sensitivity S is 0.015 and the sensitivity grade is I with the increase of wind speed; the sensitivity S is 0.159 and the sensitivity grade is II.6, and the sensitivity S is 0.310 and 0.015 respectively; the sensitivity S is non-linear positive correlation with the relative humidity, the sensitivity S is 0.159 and the sensitivity grade is II.6. The normalized growth curve of LAI was extracted and the maximum value of LAI was estimated based on the traditional remote sensing statistical model. The time series LAI was estimated by the simple multiplication of the two methods. Compared with the measured values, the RMSE of broadleaf forest and mixed forest were 0.40 and 0.49, respectively, and 0.59 respectively. The main photosynthetic parameters optimized by BEPS Hourly model can be directly introduced into the BEPS Daily model to achieve a higher simulation capacity; the seasonal variation of GPP and NPP is more significant, showing a positive correlation with the growth of vegetation and the increase of temperature; at the regional scale, the optimized BEPS Daily model is used to carry on According to the regional NPP estimation, the average GPP and NPP per unit area in Maoershan area in 2011 were 1265.56 g C.m-2.a-1,628.40 g C.m-2.a-1, respectively. The primary productivity of different forest types was coniferous forest of broadleaf mixed forest.
【学位授予单位】：东北林业大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：S718.5
本文编号：2211505
[Abstract]:Forest is the main part of terrestrial ecosystem, and forest carbon cycle plays an important role in terrestrial carbon cycle system. This study optimized and validated the hourly carbon cycle model BEPS Hourly at site scale, and explored the synergistic application of hourly carbon cycle model and daily carbon cycle model at regional scale. Firstly, the driving and validating data of BEPS Hourly model were collected at site scale (flux tower effective coverage area), and the main photosynthetic parameters, maximum carboxylation rate (Vc max) and maximum electron transfer rate (Umax), were optimized by iterative method. The biomass allocation parameters of different forest types in BEPS Hourly model were optimized, and the radiation transfer model was used as a breakthrough point to optimize the radiation transfer model considering the secondary canopy penetration and multiple reflectance, so as to improve the simulation ability of BEPS Hourly model for latent sensible heat flux on the basis of more accurate calculation of net radiation. The data assimilation algorithm-EnKF was used to correct the soil moisture predicted by BEPS Hourly model to reduce the cumulative error with time; the optimized BEPS Hourly model was used to simulate and validate the soil temperature, and the change of snow depth was simulated and validated, and the canopy temperature was simulated. Leaf area index (LAI) and meteorological factors (temperature, precipitation, wind speed, total solar radiation, relative humidity) were quantitatively analyzed. At the regional scale, the regional driving data of BEPS Daily model were collected. Because the existing MODIS LAI products could not meet the application requirements, the time series suitable for small and medium-sized regional scale were developed. Study on the method of fast LAI estimation in woodland; study the characteristics and applicability of BEPS Daily and BEPS Hourly, especially analyze the transfer reliability of main photosynthetic parameters Vcmax, Jmax from BEPS Hourly model to BEPS Daily model. On the basis of reliability analysis of parameter transfer, the BEPS Hourly model and BEPS Daily model are established. Based on the optimized BEPS Hourly, the intraday variation of GPP and NPP was simulated and analyzed at the site scale, and the regional NPP estimation and carbon source/sink spatial analysis were carried out based on the optimized BEPS Daily model. When L.m-2.s-1 was used, the RMSE of daily GPP based on BEPS Hourly simulation was the smallest, 1.10 g C.m-2.d-1, R2 was the highest, 0.95. After optimization of photosynthetic parameters Vc Max and Jmax, the BEPS Hourly model could better simulate the seasonal variation of GPP. 2. Broadleaf forest accounted for 4% of the total biomass and branches accounted for 4% of the total biomass. 96% of the total biomass, 5% of the total aboveground biomass, 95% of the total aboveground biomass, 6% of the total aboveground biomass, 94% of the total aboveground biomass, and 96% of the total aboveground biomass were mixed forest leaves and branches. Formula 3. Considering the secondary canopy penetration and multiple reflections, the latent heat flux of the model was increased from 0.769 to 0.792, RMSE from 50.77 W/m2 to 47.84 W W/m2, sensible heat flux from 0.684 to 0.705, RMSE from 48.42 W/m2 to 45.86 W/m2, and latent heat flux from simulated value to measured value. There is a significant correlation; the monthly mean diurnal variation of latent sensible heat is a single peak curve, reaching the maximum at noon, lower at night and early morning; compared with sensible heat flux, the seasonal variation of latent heat flux is more significant, in the growing season is much higher than the non-growing season, and has a positive correlation with vegetation growth, sensible heat flux. Before the data assimilation, RMSE was 0.1198, 0.5 h interval, RMSE was reduced to 0.0293, and the simulation results were significantly improved. The higher the frequency of assimilation data, the simulated soil moisture and the observed soil moisture after assimilation. The smaller the RMSE, the better the assimilation simulation effect. When the assimilation frequency was 15 days, the ability of the assimilation system to simulate soil moisture was almost the same as that of the non-assimilation model. The sensitivity S was 0.594, the sensitivity grade was IV; NPP remained basically unchanged with the increase of precipitation, indicating that the range of precipitation which could inhibit or promote vegetation growth was not included in the range of 30% to 30% precipitation, the sensitivity S was 0.0005, the sensitivity grade was I; NPP and solar radiation showed a nonlinear positive phase. The sensitivity S is 0.310, the sensitivity grade is III; the sensitivity S is 0.015 and the sensitivity grade is I with the increase of wind speed; the sensitivity S is 0.159 and the sensitivity grade is II.6, and the sensitivity S is 0.310 and 0.015 respectively; the sensitivity S is non-linear positive correlation with the relative humidity, the sensitivity S is 0.159 and the sensitivity grade is II.6. The normalized growth curve of LAI was extracted and the maximum value of LAI was estimated based on the traditional remote sensing statistical model. The time series LAI was estimated by the simple multiplication of the two methods. Compared with the measured values, the RMSE of broadleaf forest and mixed forest were 0.40 and 0.49, respectively, and 0.59 respectively. The main photosynthetic parameters optimized by BEPS Hourly model can be directly introduced into the BEPS Daily model to achieve a higher simulation capacity; the seasonal variation of GPP and NPP is more significant, showing a positive correlation with the growth of vegetation and the increase of temperature; at the regional scale, the optimized BEPS Daily model is used to carry on According to the regional NPP estimation, the average GPP and NPP per unit area in Maoershan area in 2011 were 1265.56 g C.m-2.a-1,628.40 g C.m-2.a-1, respectively. The primary productivity of different forest types was coniferous forest of broadleaf mixed forest.
【学位授予单位】：东北林业大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：S718.5
【参考文献】
相关期刊论文 前10条
1 卢伟;范文义;;中小区域尺度时间序列林地LAI快速估测方法[J];农业工程学报;2016年05期
2 赵俊芳;延晓冬;贾根锁;;东北森林净第一性生产力与碳收支对气候变化的响应[J];生态学报;2008年01期
3 朱文泉;潘耀忠;张锦水;;中国陆地植被净初级生产力遥感估算[J];植物生态学报;2007年03期
4 张生雷;谢正辉;田向军;师春香;陈锋;;基于土壤水模型及站点资料的土壤湿度同化方法[J];地球科学进展;2006年12期
5 毛留喜;孙艳玲;延晓冬;;陆地生态系统碳循环模型研究概述[J];应用生态学报;2006年11期
6 王培娟;孙睿;朱启疆;谢东辉;陈镜明;;复杂地形条件下提高BEPS模型模拟能力的途径[J];中国图象图形学报;2006年07期
7 李世华,牛铮,李壁成;植被净第一性生产力遥感过程模型研究[J];水土保持研究;2005年03期
8 赵敏,周广胜;中国北方林生产力变化趋势及其影响因子分析[J];西北植物学报;2005年03期
9 何勇,董文杰,季劲均,丹利;基于AVIM的中国陆地生态系统净初级生产力模拟[J];地球科学进展;2005年03期
10 王秋凤,牛栋,于贵瑞,任传友,温学发;长白山森林生态系统CO_2和水热通量的模拟研究[J];中国科学(D辑:地球科学);2004年S2期
，本文编号：2211505
本文链接：https://www.wllwen.com/shoufeilunwen/nykjbs/2211505.html
最近更新
教材专著
