STUDY ON MODEL FOR REMOTE SENSING ESTIMATION OF MAIZE YIELD

Through analysis of perpendicular vegetation index (PVI) from combination of visible and nearinfrared spectrums reflecting the feature of crop reflectance, we come to the conclusion that the index can better indicate crop instantaneous photosynthesis whereas people generally regard it as the representation of crop leaf area index(LAI). Exploration of crop photosynthesis within a day and its period of duration leads to production of photosynthetic vegetation index (PST) that can reflect the whole crop accumulated photosynthesis, which means the total biomass produced by crop, moreover the method simulating PST is put forward by employment of multitemporal spectrum parameters. On the basis of the achievements mentioned above, a new comprehensive model for remote sensing estimation of maize yield is established, which can comprehensively show major physiological actions of maize and the course of its yield formation, organically integrate various effective ways of crop yield estimation. It lays a solid foundation for carrying out remote sensing estimation of maize yield on a large scale.

STUDY ON GIS FOR YIELD ESTIMATION BY REMOTE SENSING IN JILIN MAIZE BELT

The integration of remote sensing and geographical information system (GIS) technology is an optimal method for maize yield estimation, because it needs the support of various data including remote sensing information and others.This paper introduces the objective, components, data acquisition and implementing way of maize yield estimation information system, and uses it in the study on maize acreage calculation, growth vigour monitoring, regional soil moisture content assessment and final yield forecast.

Rice Yield Estimation by Integrating Remote Sensing with Rice Growth Simulation Model

Since remote sensing can provide information on the actual status of an agricultural crop, the integration between remote sensing data and crop growth simulation models has become an important trend for yield estimation and prediction.The main objective of this research was to combine a rice growth simulation model with remote sensing data to estimate rice grain yield for different growing seasons leading to an assessment of rice yield at regional levels. Integration between NOAA (National Oceanic and Atmospheric Administration) AVHRR (Advanced Very High Resolution Radiometer) data and the rice growth simulation model ORYZA1 to develop a new software, which was named as Rice-SRS Model, resulted in accurate estimates for rice yield in Shaoxing, China, with an estimation error reduced to 1.03% and 0.79% over-estimation and 0.79% under-estimation for early, single and late season rice, respectively. Selecting suitable dates for remote sensing images was an important factor which could influence estimation accuracy. Thus, given the different growing periods for each rice season, four images were needed for early and late rice, while five images were preferable for single season rice.Estimating rice yield using two or three images was possible, however, if images were obtained during the panicle initiation and heading stages.

Regionalization for Rice Yield Estimation by Remote Sensing in Zhejiang Province

In order to provide a scientific basis for rice yield estimation and improve the accuracy of yield estimation in Zhejiang Province, regionalization indices for rice yield estimation by remote sensing (RS) in the province were determined by considering the special features of yield estimation by RS, and based on analysis of the natural conditions of Zhejiang Province. The indices determined included rice cropping system, agroclimate, landform, surface feature structure and rice yield level, where rice planting system was considered as the main one. Then regionalization for rice yield estimation by RS was completed by spatial neighboring analysis with the Geographical information System (GIS) technology combined with using of tree algorithm. The province was divided into two regions, i. e., the single-cropping rice region which was subdivided into 3 regions including those in mountains of northwest Zhejiang, water network area of north Zhejiang and mountains of south Zhejiang, and double-cropping rice region which was subdivided into 5 regions including those on plain of north Zhejiang, coastal plains and hills of southeast Zhejiang, Jin-Qu Basin of middle Zhejiang, hills of east Zhejiang, and hills and mountains of northwest Zhejiang. This regionalization took the county borders as the region boundaries, kept the regions connective and made the administrative regions integrity and, then, could meet the requirements of rice yield estimation by RS, showing that the results were quite satisfying.

REMOTE SENSING BASED ESTIMATION SYSTEM FOR WINTER WHEAT YIELD IN NORTH CHINA PLAIN

This paper presents the applications of Landsat Thematic Mapper (TM) data and Advanced Very High Resolution Radiometer (AVHRR) time series data for winter wheat production estimation in North China Plain. The keytechniques are described systematically about winter wheat yield estimation system, including automatically extractingwheat area, simulating and monitoring wheat growth situation, building wheat unit yield model of large area and forecasting wheat production. Pattern recognition technique was applied to extract sown area using TM data. Temporal NDVI(Normal Division Vegetation Index) profiles were produced from 8 - 12 times AVHRR data during wheat growth dynamically. A remote sensing yield model for large area was developed based on greenness accumulation, temperature andgreenness change rate. On the basis of the solution of key problems, an operational system for winter wheat yield estimation in North China Plain using remotely sensed data was established and has operated since 1993, which consists of 4 subsystems, namely databases management, image processing, models bank management and production prediction system.The accuracy of wheat production prediction exceeded 96 per cent compared with on the spot measurement.

Developing a process-based and remote sensing driven crop yield model for maize(PRYM–Maize) and its validation over the Northeast China Plain

Spatial dynamics of crop yield provide useful information for improving the production. High sensitivity of crop growth models to uncertainties in input factors and parameters and relatively coarse parameterizations in conventional remote sensing(RS) approaches limited their applications over broad regions. In this study, a process-based and remote sensing driven crop yield model for maize(PRYM–Maize) was developed to estimate regional maize yield, and it was implemented using eight data-model coupling strategies(DMCSs) over the Northeast China Plain(NECP). Simulations under eight DMCSs were validated against the prefecture-level statistics(2010–2012) reported by National Bureau of Statistics of China, and inter-compared. The 3-year averaged result could give more robust estimate than the yearly simulation for maize yield over space. A 3-year averaged validation showed that prefecture-level estimates by PRYM–Maize under DMCS8, which coupled with the development stage(DVS)-based grain-filling algorithm and RS phenology information and leaf area index(LAI), had higher correlation(R, 0.61) and smaller root mean standard error(RMSE, 1.33 t ha^(–1)) with the statistics than did PRYM–Maize under other DMCSs. The result also demonstrated that DVS-based grain-filling algorithm worked better for maize yield than did the harvest index(HI)-based method, and both RS phenology information and LAI worked for improving regional maize yield estimate. These results demonstrate that the developed PRYM–Maize under DMCS8 gives reasonable estimates for maize yield and provides scientific basis facilitating the understanding the spatial variations of maize yield over the NECP.

Remote Sensing and GIS Based Spectro-Agrometeorological Maize Yield Forecast Model for South Tigray Zone, Ethiopia

Remote-sensing data acquired by satellite imageries have a wide scope in agricultural applications owing to their synoptic and repetitive coverage. This study reports the development of an operational spectro-agrometereological yield model for maize crop derived from time series data of SPOT VEGETATION, actual and potential evapotranspiration and rainfall estimate satellite data for the years 2003-2012. Indices of these input data were utilized to validate their strength in explaining grain yield recorded by the Central Statistical Agency through correlation analyses. Crop masking at crop land area was applied and refined using agro-ecological zones suitable for maize. Rainfall estimates and average Normalized Difference Vegetation Index were found highly correlated to maize yield with the former accounting for 85% variation and the latter 80%, respectively. The developed spectro-agrometeorological yield model was successfully validated against the predicted Zone level yields estimated by Central Statistical Agency (r2 = 0.88, RMSE = 1.405 q·ha-1 and 21% coefficient of variation). Thus, remote sensing and geographical information system based maize yield forecast improved quality and timelines of the data besides distinguishing yield production levels/areas and making intervention very easy for the decision makers thereby proving the clear potential of spectro-agrometeorological factors for maize yield forecasting, particularly for Ethiopia.

Research advances of SAR remote sensing for agriculture applications: A review

Synthetic aperture radar(SAR) is an effective and important technique in monitoring crop and other agricultural targets because its quality does not depend on weather conditions. SAR is sensitive to the geometrical structures and dielectric properties of the targets and has a certain penetration ability to some agricultural targets. The capabilities of SAR for agriculture applications can be organized into three main categories: crop identification and crop planting area statistics, crop and cropland parameter extraction, and crop yield estimation. According to the above concepts, this paper systematically analyses the recent progresses, existing problems and future directions in SAR agricultural remote sensing. In recent years, with the remarkable progresses in SAR remote sensing systems, the available SAR data sources have been greatly enriched. The accuracies of the crop classification and parameter extraction by SAR data have been improved progressively. But the development of modern agriculture has put forwarded higher requirements for SAR remote sensing. For instance, the spatial resolution and revisiting cycle of the SAR sensors, the accuracy of crop classification, the whole phenological period monitoring of crop growth status, the soil moisture inversion under the condition of high vegetation coverage, the integrations of SAR remote sensing retrieval information with hydrological models and/or crop growth models, and so on, still need to be improved. In the future, the joint use of optical and SAR remote sensing data, the application of multi-band multi-dimensional SAR, the precise and high efficient modeling of electromagnetic scattering and parameter extraction of crop and farmland composite scene, the development of light and small SAR systems like those onboard unmanned aerial vehicles and their applications will be active research areas in agriculture remote sensing. This paper concludes that SAR remote sensing has great potential and will play a more significant role in the various fields of agricultural remote sensing.

REGIONALIZATION FOR LARGE AREA CROP ESTIMATES BY REMOTE SENSING——A Case Study of Chinese Wheat

Ｔｈｅｃｒｏｐｅｓｔｉｍａｔｅｓｂｙｒｅｍｏｔｅｓｅｎｓｉｎｇ，ｄｅｖｅｌｏｐｉｎｇｑｕｉｃｋｌｙｉｎｒｅｃｅｎｔｄｅｃａｄｅｓ，ｉｓａｕｐｔｏｄａｔｅｔｅｃｈｎｉｑｕｅ．Ｓｏｍｅｓｙｓｔｅｍｓｏｆｃｒｏｐｅｓｔｉｍａｔｅｓｂｙｒｅｍｏｔｅｓｅｎ...

Estimating Rice Yield by HJ-1A Satellite Images

As illustrated by the case of Xuyi County, Jinhu County and Hongze County in Jiangsu Province, China, monitoring and forecasting of rice production were carried out by using HJ-1A satellite remote sensing images. The handhold GPS machines were used to measure the geographical position and some other information of these samples such as area shape. The GPS data and the interpretation marks were used to correct H J-1 image, assist human-computer interactive interpretation, and other operations. The test data had been participated in the whole classification process. The accuracy of interpreted information on rice planting area was more than 90% By using the leaf area index from the normalized difference vegetation index inversion, the biomass from the ratio vegetation index inversion, and combined with the rice yield estimation model, the rice yield was estimated. Further, the thematic map of rice production classification was made based on the rice yield data. According to the comparison results between measured and fitted values of yields and areas of sampling sites, the accuracy of the yield estimation was more than 85%. The results suggest that HJ-A/B images could basically meet the demand of rice growth monitoring and yield forecasting, and could be widely applied to rice production monitoring.

Remote sensing-based estimation of rice yields using various models:A critical review

Reliable estimation of region-wide rice yield is vital for food security and agricultural management.Field-scale models have increased our understanding of rice yield and its estimation under theoretical environmental conditions.However,they offer little infor-mation on spatial variability effects on farm-scale yield.Remote Sensing(RS)is a useful tool to upscale yield estimates from farm scales to regional levels.Much research used RS with rice models for reliable yield estimation.As several countries start to operatio-nalize rice monitoring systems,it is needed to synthesize current literature to identify knowledge gaps,to improve estimation accuracies,and to optimize processing.This paper critically reviewed significant developments in using geospatial methods,imagery,and quantitative models to estimate rice yield.First,essential characteristics of rice were discussed as detected by optical and radar sensors,band selection,sensor configuration,spatial resolution,mapping methods,and biophysical variables of rice derivable from RS data.Second,various empirical,process-based,and semi-empirical models that used RS data for spatial estimation of yield were critically assessed-discussing how major types of models,RS platforms,data assimilation algorithms,canopy state variables,and RS variables can be integrated for yield estimation.Lastly,to overcome current constraints and to improve accuracies,several possibilities were suggested-adding new modeling modules,using alternative canopy variables,and adopting novel modeling approaches.As rice yields are expected to decrease due to global warming,geospatial rice yield estimation techniques are indispensable tools for climate change assessments.Future studies should focus on resolving the current limitations of estimation by precise delineation of rice cultivars,by incorporating dynamic harvesting indices based on climatic drivers,using innovative modeling approaches with machine learning.

基于Sentinel-2和MODIS数据的宣恩县水稻遥感估产研究

基于作物的波谱反射特征,利用公开的多源遥感数据和相关技术能够实现农作物种植面积提取和产量预估,为作物长势监测等农业需求提供科学决策依据。本文首先基于Sentinel-2卫星影像,结合基于人工目视解译的监督分类、基于规则的面向对象分类和基于专家知识的决策树分类3种影像分类方法综合确定县级研究区的水稻种植范围,再选取水稻生育物候期内的多时相多光谱MODIS13Q1影像产品,建立影像提取出的植被指数EVI与水稻年产量之间的多元统计回归模型并应用于年产量预估,估产结果精度均达94%以上,符合实际需求。该模型可用性较强,对县域农作物遥感估产应用具有一定的指导意义。

基于遥感多参数和CNN-Transformer的冬小麦单产估测

为了提高冬小麦单产估测精度,改善估产模型存在的高产低估和低产高估等现象,以陕西省关中平原为研究区域,选取旬尺度条件植被温度指数(VTCI)、叶面积指数(LAI)和光合有效辐射吸收比率(FPAR)为遥感特征参数,结合卷积神经网络(CNN)局部特征提取能力和基于自注意力机制的Transformer网络的全局信息提取能力,构建CNN-Transformer深度学习模型,用于估测关中平原冬小麦产量。与Transformer模型(R^(2)为0.64,RMSE为465.40 kg/hm^(2),MAPE为8.04%)相比,CNN-Transformer模型具有更高的冬小麦单产估测精度(R^(2)为0.70,RMSE为420.39 kg/hm^(2),MAPE为7.65%),能够从遥感多参数中提取更多与产量相关的信息,且对于Transformer模型存在的高产低估和低产高估现象均有所改善。基于5折交叉验证法和留一法进一步验证了CNN-Transformer模型的鲁棒性和泛化能力。此外,基于CNN-Transformer模型捕获冬小麦生长过程的累积效应,分析逐步累积旬尺度输入参数对产量估测的影响,评估模型对于冬小麦不同生长阶段的累积过程的表征能力。结果表明,模型能有效捕捉冬小麦生长的关键时期,3月下旬至5月上旬是冬小麦生长的关键时期。

基于遥感多参数和IPSO-WNN的冬小麦单产估测

冬小麦是我国的主要粮食作物之一。为进一步准确地估测冬小麦产量,以陕西省关中平原为研究区域,选取冬小麦主要生育期与水分胁迫和光合作用等密切相关的条件植被温度指数(VTCI)、叶面积指数(LAI)和光合有效辐射吸收比率(FPAR)作为遥感特征参数,采用改进的粒子群算法优化小波神经网络(IPSO-WNN)以改善梯度下降方法易陷入局部最优的缺陷,并构建冬小麦产量估测模型。结果表明,IPSO-WNN模型的决定系数R2为0.66,平均绝对百分比误差(MAPE)为7.59%,相比于BPNN(R2=0.46,MAPE为11.80%)与WNN(R2=0.52,MAPE为9.80%),IPSO-WNN能够进一步提高模型的精度、增强模型的鲁棒性。采用灵敏度分析的方法探究对冬小麦产量影响较大的输入参数,结果发现,抽穗-灌浆期的FPAR对冬小麦产量影响最大,其次拔节期的VTCI、抽穗-灌浆期和乳熟期的LAI以及返青期和拔节期的FPAR对冬小麦产量的影响较大。通过IPSO-WNN输出获取冬小麦综合监测指数I,构建I与统计单产之间的估产模型以估测关中平原冬小麦单产,结果显示,估测单产与统计单产之间的R2为0.63,均方根误差(RMSE)为505.50 kg/hm^(2),相比于前人的研究较好地解决了估产模型存在的“低产高估”的问题,因此,本文基于IPSO-WNN构建的估产模型能够较准确地估测关中平原冬小麦产量。

基于WOFOST模型与遥感数据同化的县级尺度玉米估产研究

区域尺度的作物生长动态监测和产量预测对于保障粮食安全和农业政策的制定具有重要参考依据。遥感数据同化应用极大提高了作物估产的时效性和精度。为及时、准确地实现县级尺度粮食产量的估测,以及提升产量估测的精度,以辽宁省铁岭县为研究区,采用WOFOST(world food studies)模型与遥感同化相结合的方法对铁岭县玉米进行估产研究。通过采用扩展傅里叶幅度敏感性检验算法(extened Fourier amplitude sensitivity test,EFAST)敏感性分析方法实现玉米估产敏感性参数的分析,以及本地化;通过采用参数自动率定程序PEST(parameter estimation)实现参数的优化,验证结果为采样点产量的平均误差为852.39 kg·hm^(-2),模型模拟的精度达到92.82%。为进一步提高和优化模型估产精度,将遥感反演得到的叶面积指数采用集合卡尔曼滤波算法与模型模拟的叶面积指数进行数据同化,平均误差从同化前的852.39 kg·hm^(-2)降低为435.01 kg·hm^(-2),估产精度从92.82%提高到96.33%,有效提高了WOFOST模型估产的精度。结果表明:水分对玉米的生长发育限制并不大,其产量形成主要受光温影响,对温度、光能利用效率和最大同化速率有关的参数具有较高的敏感性;优化后的模型能够较好模拟铁岭县玉米生长发育情况,产量验证表明优化后的模型模拟的效果较好,但仍存在一定的误差;比值植被指数与叶面积指数的相关性最高,反演模型精度较好,反演结果表明叶面积指数在抽雄吐丝期差距较大,而在成熟期的差距不大;经过作物模型与遥感数据同化之后,估产的精度得到明显提高,说明遥感与作物模型同化是一种有效地提高作物估产和产量预测精度的方法。

基于无人机高光谱遥感与机器学习的小麦品系产量估测研究

为快速、准确地估测小麦产量,有效提高育种工作效率,本文以小麦品系为研究对象,收集小麦灌浆期无人机高光谱数据和产量数据。首先基于递归特征消除法筛选出特征波长作为模型输入变量,然后利用岭回归(Ridge regression,RR)、偏最小二乘回归(Partial least squares regression,PLS)、多元线性回归(Multiple linear regression,MLR)3种线性算法和随机森林(Random forest,RF)、梯度提升回归(Gradient boosting regression,GBR)、极限梯度提升(eXtreme gradient boosting,XGB)、高斯过程回归(Gaussian process regression,GPR)、支持向量回归(Support vector regression,SVR)、K最邻近算法(K-nearest neighbor,KNN)6种非线性算法构建单一算法产量估测模型并进行精度比较,最后基于Stacking算法构建多模型集成组合,筛选最佳集成模型。结果表明,基于不同算法的产量估测模型精度差异显著,非线性模型优于线性模型,基于GBR的产量估测模型在单一模型中表现最优,训练集R^(2)为0.72,RMSE为534.49 kg/hm^(2),NRMSE为11.10%,测试集R^(2)为0.60,RMSE为628.73 kg/hm^(2),NRMSE为13.88%。基于Stacking算法构建的集成模型性能与初级模型和次级模型的选择密切相关,以KNN、RR、SVR为初级模型组合,GBR为次级模型的集成模型有效提高了估测精度,相比单一模型GBR,训练集R^(2)提高1.39%,测试集R^(2)提高3.33%。本研究可为基于高光谱技术的小麦品系产量估测提供应用参考。

基于遥感多参数和VMD-GRU的冬小麦单产估测

为充分挖掘时间序列遥感参数的时序信息和趋势信息,并进一步提升冬小麦估产精度,以陕西省关中平原为研究区域,选取与冬小麦长势密切相关的生育时期尺度的条件植被温度指数(VTCI)、叶面积指数(LAI)和光合有效辐射吸收比率(FPAR)作为遥感参数,构建耦合变分模态分解(VMD)与门控循环单元(GRU)神经网络的估产模型。应用VMD算法将各个时间序列遥感参数分解为多组平稳的本征模态函数(IMF)分量,选取与原始时间序列遥感参数高度相关的IMF分量进行特征重构,并将重构特征作为GRU网络的输入,以构建冬小麦组合估产模型。结果表明,VMD-GRU组合估产模型决定系数为0.63,均方根误差为448.80 kg/hm^(2),平均相对误差为8.14%,相关性达到极显著水平(P<0.01),其精度优于单一估产模型精度,表明该组合估产模型能够提取非平稳时间序列数据的多尺度、多层次特征,并充分挖掘冬小麦各生育时期遥感参数间的内在联系,获得准确单产估测结果的同时提升了估产模型的可解释性。

基于低空无人机遥感的水稻产量估测方法研究进展

水稻作为主要的粮食作物之一,其产量估测对国家政策宏观调控、地方农情实时指导以及优良品种的定向培育都起着至关重要的作用。随着作物学科及其交叉学科的不断进步,估产的方法和模式也逐渐多样化。同时,随着遥感技术的发展,尤其是低空无人机的出现及其应用的普及,水稻智能遥感估产方法不断创新,估测精度不断提升,但针对基于无人机遥感的智能稻作产量估测缺乏系统和科学的归纳总结。鉴于此,本文在梳理目前主流水稻估产方法及其优缺点的基础上,聚焦探讨低空智能遥感技术在水稻产量估测中的应用及未来发展方向。围绕当前利用低空遥感技术获取的主要特征信息,探讨实现智能遥感水稻估产的模型开发。此外,还探讨了智能遥感技术在水稻产量估测中面临的挑战和问题,旨在深化并完善对低空遥感水稻的产量估测方法的理解,进而为水稻产量智能估计提供系统全面的参考和指导。

基于无人机影像的井冈蜜柚果树树形信息提取及产量估测

为实现基于无人机影像的井冈蜜柚果树树形信息(冠幅、树高)快速、准确提取及产量预测,通过基于无人机影像生成数字正射影像(DOM),计算4个植被指数,分析4个植被指数阈值分割提取冠幅的精度,确定敏感植被指数及其最佳分类阈值完成植被区域的提取,实现冠幅提取;再基于无人机影像生成的数字高程模型(DEM),提取果树树高;运用“冠幅、树高及冠幅+树高”三种模式对产量进行预测。结果表明,利用归一化差值指数(Normalized Difference Index, NDI)提取冠幅时精度最高,提取的东西冠幅与实测值之间决定系数R^(2)达0.917 2,南北冠幅与实测值之间的R^(2)达0.823 6,冠幅均值与实测值均值之间的R^(2)达0.892 8;基于DEM提取树高时,也具有较好的效果,提取的树高与实测值之间的R^(2)达0.863 3,均方根误差RMSE为0.148 m。进一步运用“冠幅、树高及冠幅+树高”三种模式对挂果数进行预测,运用“冠幅+树高”预测挂果数的R^(2)为0.676,调整R^(2)为0.638,预测效果最好。

融合光谱和纹理特征的玉米产量预测研究

本研究采集夏玉米拔节、抽雄、灌浆和成熟4个时期的无人机可见光和多光谱影像,提取和筛选植被指数与纹理特征参数,构建植被指数与纹理特征融合变量,采用反向传播神经网络、随机森林和支持向量机3种机器学习方法构建玉米产量预测模型。结果表明:相较于单一类型参数,融合植被指数和纹理特征进行产量预测模型精度更高;3种机器学习方法中以随机森林构建的玉米产量预测模型效果最好,且最佳预测时期为灌浆期(籽粒水泡期);综合评价建模和验证结果,基于多光谱影像植被指数与纹理特征融合变量和随机森林方法构建的模型玉米产量预测效果最佳。遥感信息的多特征融合与机器学习方法的搭配能够挖掘和利用更多信息并提高玉米产量预测的精度和鲁棒性。