【作者】 李宗南；

【导师】 陈仲新；

【作者基本信息】 中国农业科学院 ， 农业遥感， 2014， 博士

【摘要】 及时准确的作物生长信息是作物生产管理和早期估产的重要依据，对保障国家粮食安全至关重要。作物生长遥感监测通过非接触、远距离探测的方式可快速获取大面积作物生长信息。定量遥感通过辐射传输模型和数学方法从遥感波谱信息中提取多种作物参数及环境参数，有助于定量化监测作物生长状况，是作物生长监测方法发展的重要方向。作物光能利用率模型是一种使用气象、遥感数据快速估算作物生产力、生物量的模型；模型中的光能利用率、光合有效辐射吸收比可通过定量遥感方法、产品参与估算，在发挥定量遥感作用方面有独特优势。因此本论文以生物量为监测指标，展开基于光能利用率模型和定量遥感的玉米生长监测方法研究。本研究在玉米主产区之一的华北平原设立研究区，通过玉米生长的地面遥感试验和区域遥感调查获取数据。针对一般作物光能利用率模型缺乏考虑叶片叶绿素含量、光合有效辐射强度影响生物量累积的问题，研究了叶片叶绿素含量参数化光能利用和光合有效辐射吸收比的方法、光合有效辐射强度参数化光能利用的方法，建立了叶绿素含量参数化光能利用的模型(LUEChl模型)、叶绿素含量参数化光合有效辐射吸收比的光能利用率模型(FAPARChl模型)、叶绿素含量同时参数化光能利用率和光合有效辐射吸收比的光能利用率模型(LUEChl+FAPARChl模型)、叶绿素含量和光合有效辐射强度参数化光能利用率的模型(LUEPAR模型)；鉴于叶绿素含量在作物生长定量监测中的关键作用，进行了光谱指数法和冠层反射率模型反演法估算叶片叶绿素含量的研究；最后在作物光能利用率模型和叶绿素含量遥感方法的研究基础上，研究应用HJ-1多光谱遥感数据估算玉米生物量、监测玉米生长的方法，在区域上实现基于叶绿素含量参数化光能利用率模型和定量遥感的玉米生长监测方法。通过各项研究得到以下主要发现和结论。1.对比不使用叶绿素含量信息的MODIS GPP模型(RMSE=390.9g/m2，RE=21.5%)，叶片叶绿素含量参数化的光能利用率模型，即LUEChl模型(RMSE=91.7g/m2，RE=5.1%)、FAPARChl模型(RMSE=173.8g/m2，RE=9.4%)、LUEChl+FAPARChl模型(RMSE=52.9g/m2，RE=2.6%)和LUEPAR模型(RMSE=75.2g/m2，RE=4.1%)显著降低了生物量估算误差；其中，LUEChl+FAPARChl模型首次在玉米生物量估算中使用叶绿素含量同时参数化光能利用率和光合有效辐射吸收比，生物量估算误差最低，为应用叶绿素含量遥感产品定量监测作物生长提供了可靠的模型。对比LUEChl模型和LUEPAR模型估算的净初级生产力，LUEPAR模型可降低光合有效辐射强度引起的误差，能更准确地估算不同天空散射比条件下的净初级生产力，在作物生长监测中有很好的应用潜力。2.叶片叶绿素含量光谱指数估算方法研究表明MTCI、Datt及SRCI指数算法简便且在玉米各生长期与叶片叶绿素含量的相关性稳定，适用于多个时期的叶片叶绿素含量遥感估算。冠层反射率模型反演法研究表明：先通过红光、近红外波段反演叶面积指数，然后使用叶面积指数和绿光或红边反射率反演叶片叶绿素含量的方法可有效可行；且使用红边反射率反演的效果与绿光反射率的相当。3.首次建立了综合多光谱定量遥感和叶绿素含量参数化光能利用率、光合有效辐射吸收比的光能利用率模型的玉米生长监测方法，并使用HJ-1多光谱数据在区域上实现该方法。该方法先使用多时相的HJ-1多光谱数据建立归一化植被指数曲线，根据玉米归一化植被指数曲线特征识别夏玉米并提取其空间分布；然后通过ACRM模型和HJ-1多光谱地表反射率先后反演得到玉米叶面积指数和叶片叶绿素含量；最后基于定量反演结果和LUEChl+FAPARChl模型估算生物量，实现玉米生长定量监测。综上，本研究通过建立叶片叶绿素含量参数化的光能利用率模型、研究叶片叶绿素含量遥感光谱指数估算方法和冠层反射率模型反演法，为应用多光谱定量遥感产品监测作物生长提供了相关理论依据和适用的光能利用率模型；通过HJ-1多光谱数据，实现了综合多光谱定量遥感和叶绿素含量参数化的光能利用率模型的区域的玉米生长监测方法，为使用高时空分辨率多光谱遥感数据定量监测作物生长提供很好的参考案例。更多还原

【Abstract】 Timely and accurate monitoring of crop growth is very important for the decision-making onnational food security such as food pricing, grain reserve and food trade. The crop growth monitoringmethod based on remote sensing vegetation indices through the way of non-contact and long distancedetection can timely get large area crop growth information, is the main method in current cropmonitoring, and plays an important role in the decision making process of the national food security.The monitoring method based on quantitative remote sensing extracts crops and environmentalparameters from spectral information by using radiation transfer model and mathematical method, thenconducts a comprehensive monitoring of crop growth status. It is one of the important directions of cropgrowth monitoring method research, and how to apply more quantitative remote sensing methods andproducts in crop growth monitoring is the main content of this study. Crop light use efficiency modelhas its unique advantages in application of quantitative remote sensing. By improving the algorithm forlight use efficiency (LUE) or fraction of absorbed photosynthetically active radiation (FAPAR) in theLUE model, more data products of quantitative remote sensing can be used in crop growth monitoring,which is an efficient methodology to integrate quantitative remote sensing and LUE model into cropgrowth monitoring. This study defines biomass as the monitor index. The main objective of this study isto improve the method of crop growth monitoring by combining the LUE model and quantitativeremote sensing techniques.A series of the crop growth monitoring field experiments and regional remote sensing experimentswere conducted in North China Plain, which is the main maize growing region in China as well as in theworld. Regarding the lack of modeling of leaf chlorophyll content (LCC) and photosynthetically activeradiation (PAR) density effects in most crop LUE models, new models that biomass responds to LCCand PAR, including LUEChlmodel, FAPARChlmodel, LUEChl+FAPARChlmodel and LUEPARmodel werebuilt by improving the algorithms of LUE or FAPAR. LCC is a key role in estimating LUE and FAPAR,so algorithms of leaf chlorophyll content derived from remote sensing were investigated. CombinedLUEChl+FAPARChlmodel and chlorophyll content inversion method, the author implemented the maizegrowth monitoring using HJ-1multi-spectral remote sensing data. Through this study, the mainconclusions are as follows.1. Compared to the results (RMSE=390.9g/m2，RE=21.5%) of a LUE model that lack of modelingof leaf chlorophyll content effect, LUEChlmodel (RMSE=91.7g/m2，RE=5.1%), FAPARChlmodel(RMSE=173.8g/m2, RE=9.4%), LUEChl+FAPARChlmodel (RMSE=52.9g/m2, RE=2.6%) and LUEPARmodel (RMSE=75.2g/m2, RE=4.1%) significantly reduced biomass estimation error. LUEChl+FAPARChlmodel especially broke through limitations of many traditional LUE models by improving thealgorithms of LUE and FAPAR, is suitable for using chlorophyll content for crop growth monitoring.Compared NPP results of LUEChlmodel and LUEPARmodel, LUEPARmodel can lower NPP errors thatinduce by PAR density, is more accurate for NPP estimation in clear and overcast sky, performs a goodpotential in crop growth monitoring. 2. The study of retrieve LCC using spectral index showed that, MTCI, Datt and SRCI performsgood correlation with LCC at whole growth period, is suitable for retrieve LCC at different periods.Canopy reflectance model inversion research showed that the method that retrieved leaf area index(LAI) firstly according to red and near infrared reflectance then retrieved LCC by LAI and green or rededge reflectance is effective, and the results of red edge reflectance has the same accuracy with greenreflectance in chlorophyll content inversion.3. A monitoring method of maize growth based on a LUE model that LCC parameterized LUE andFAPAR was established for the first time, and realized using HJ-1multispectral images. In this method,the spatial distribution of maize was extracted by normalized difference vegetation index curve thatbuilt by multi-temporal multispectral images, maize LAI and LCC were retrieved based on ACRMmodel and surface reflectivity, finally LUE, FAPAR and biomass response to LCC were estimated.In conclusion, this study established a series of new light use efficiency models that consideredLCC effects, which prepared theory and models for applying more quantitative remote sensing productsto monitoring crop growth. By HJ-1multispectral data, a monitoring method of maize growth based onquantitative remote sensing and a LUE model parameterize by LCC was carry out, which provide agood reference case for quantitative monitoring of crop growth using high resolution multispectral data.更多还原