CN109359862B

Movatterモバイル変換

Info

Publication number: CN109359862B
Application number: CN201811206088.7A
Authority: CN
Inventors: 夏兴生; 王凯; 潘耀忠; 朱秀芳; 姬忠林
Original assignee: Beijing Normal University
Current assignee: Beijing Normal University
Priority date: 2018-10-17
Filing date: 2018-10-17
Publication date: 2020-09-18
Anticipated expiration: 2038-10-17
Also published as: CN109359862A

Abstract

Translated fromChinese

本发明公开了一种粮食作物实时估产方法及系统。该方法包括：获取历史数据；根据不同年份不同时期的增强型植被指数计算增强型植被指数的距平；以第i时期增强型植被指数的距平为自变量，以估产差值为因变量，构建线性回归方程，得到第i时期的估产模型，估产差值为统计产量与趋势产量的差值或统计产量与第i‑1时期的估产值的差值；将第i时期的增强型植被指数的距平代入第i时期的估产模型，得到第i时期的估产波动值；根据第i时期的估产波动值和第i‑1时期估产值计算第i时期的模型估产值；计算第i时期的估产模型的误差；采用第i时期的估产模型的误差对第i时期的模型估产值进行修正，得到第i时期的估产值。本发明能够实现准确有效、实时性强的粮食作物估产。

The invention discloses a real-time yield estimation method and system of grain crops. The method includes: obtaining historical data; calculating the anomaly of the enhanced vegetation index according to the enhanced vegetation index in different years and different periods; taking the anomaly of the enhanced vegetation index in the i-th period as the independent variable, and using the estimated yield difference as the dependent variable, A linear regression equation is constructed to obtain the estimated yield model of the i-th period, and the estimated yield difference is the difference between the statistical yield and the trend yield or the difference between the statistical yield and the estimated output value of the i-1 period; Substitute the anomaly into the estimated output model of the i-th period to obtain the estimated output fluctuation value of the i-th period; calculate the model-estimated output value of the i-th period according to the estimated production fluctuation value of the i-th period and the estimated output value of the i-1 period; calculate the estimated output value of the i-th period The error of the estimated production model; the estimated output value of the model in the i-th period is corrected by using the error of the production-estimated model of the i-th period to obtain the estimated output value of the i-th period. The invention can realize accurate, effective and real-time food crop yield estimation.

Description

Translated fromChinese

一种粮食作物实时估产方法及系统A method and system for real-time yield estimation of food crops

技术领域technical field

本发明涉及粮食作物估产领域，特别是涉及一种粮食作物实时估产方法及系统。The invention relates to the field of grain crop yield estimation, in particular to a grain crop real-time yield estimation method and system.

背景技术Background technique

及时、准确、有效的粮食作物估产对于应对国际粮价变动、稳定粮食安全及农业生产管理决策具有重要的意义。现代农业发展至今，一系列用于作物估产的技术方法在预测产量方面均取得了不同程度的成功，根据其采用的手段分为统计估产、气象估产、农学估产、遥感估产等。Timely, accurate and effective estimation of food crop production is of great significance for coping with international food price changes, stabilizing food security and making decisions on agricultural production management. Since the development of modern agriculture, a series of technical methods for crop yield estimation have achieved varying degrees of success in predicting yield.

统计估产是通过整体考虑影响作物产量的各种统计因子与作物产量之间的关系进行产量预测，常用的为趋势产量估测。该方法没有考虑作物产量形成的复杂机理，模型简单却没有明确反应作物的生长发育过程。Statistical yield estimation is to predict yield by considering the relationship between various statistical factors affecting crop yield and crop yield as a whole, and is commonly used for trend yield estimation. This method does not consider the complex mechanism of crop yield formation, and the model is simple but does not clearly reflect the growth and development process of crops.

农学估产具有良好的农学基础，以此为基础发展的作物模拟模型被广泛应用，但此类模型复杂，需要大量的输入数据进行校准和调优，实际应用中很难满足其完整而有效的数据需求，其次，不同区域地理环境、种植制度的差异性也使得各农学参数在区域间发生变化，因此，该方式目前仍然停留在实验调试阶段，很难进行大范围估产业务应用。Agronomic yield estimation has a good agronomic foundation, and crop simulation models developed on this basis are widely used, but such models are complex and require a large amount of input data for calibration and tuning, and it is difficult to meet their complete and effective data in practical applications. Second, the differences in geographical environment and planting systems in different regions also make various agronomic parameters change between regions. Therefore, this method is still in the stage of experiment and debugging, and it is difficult to carry out large-scale production estimation business applications.

气象估产是依据气象要素对于作物的生长状况具有明显影响的机理构建气象要素的产量系数来解释不同气象要素对产量的影响而进行的产量预测，例如，高金兰、马晓群等统计分析安徽省气候变化下降水量对粮食产量的影响，表明产量波动主要由水分条件不稳定导致。但该模型要素单一，未考虑其他因子对产量的影响，且气象要素以站点观测为主，空间覆盖范围有限。Meteorological yield estimation is a yield prediction based on the mechanism that meteorological factors have a significant impact on the growth of crops. The yield coefficient of meteorological factors is constructed to explain the impact of different meteorological factors on yield. For example, Gao Jinlan, Ma Xiaoqun, etc. Statistical analysis of climate change in Anhui Province The effect of falling water on grain yield indicated that yield fluctuation was mainly caused by unstable water conditions. However, the model has a single element, does not consider the impact of other factors on yield, and the meteorological elements are mainly station observations, with limited spatial coverage.

遥感估产则是基于现代遥感科学技术获得作物生长的光谱信息反演作物生长参数，建立其与产量的相关关系实现产量的预估。遥感反演的生长参数是反映作物在特定环境下生长状况的综合指标，该指标是气象因子、土壤因子、田间管理因子等影响的综合结果，被认为是估产的最优因子，因此，基于遥感的农业估产被广泛应用。例如，王恺宁等利用冬小麦灌浆期四种不同植被指数组合建立模型，弥补单植被指数的缺陷，提高估产精度。但是，遥感技术仍然受到天气、同物异谱、异物同谱等各种影响，无法实现理论上的及时有效性。Remote sensing yield estimation is based on the spectral information of crop growth obtained by modern remote sensing science and technology to invert crop growth parameters, and establish the correlation between them and yield to achieve yield estimation. The growth parameter retrieved by remote sensing is a comprehensive index reflecting the growth status of crops in a specific environment. This index is the comprehensive result of the influence of meteorological factors, soil factors, field management factors, etc., and is considered to be the optimal factor for estimating yield. Therefore, based on remote sensing The agricultural yield estimation is widely used. For example, Wang Kaining et al. used a combination of four different vegetation indices at the grain-filling stage of winter wheat to build a model to make up for the shortcomings of a single vegetation index and improve the yield estimation accuracy. However, remote sensing technology is still subject to various influences such as weather, different spectra of different objects, and the same spectrum of different objects, so it cannot achieve theoretical timeliness and effectiveness.

发明内容SUMMARY OF THE INVENTION

本发明的目的是提供一种粮食作物实时估产方法及系统，能够实现准确有效、实时性强的粮食作物估产。The purpose of the present invention is to provide a real-time grain crop yield estimation method and system, which can realize accurate, effective and real-time grain crop yield estimation.

为实现上述目的，本发明提供了如下方案：For achieving the above object, the present invention provides the following scheme:

一种粮食作物实时估产方法，所述方法包括：A method for estimating real-time yield of food crops, the method comprising:

获取历史数据，所述历史数据包括不同年份不同时期的增强型植被指数、不同年份粮食作物的统计产量和不同年份粮食作物的趋势产量；obtaining historical data, the historical data including the enhanced vegetation index in different years and different periods, the statistical yield of grain crops in different years and the trend yield of grain crops in different years;

根据不同年份不同时期的增强型植被指数计算增强型植被指数的距平；Calculate the anomaly of the enhanced vegetation index according to the enhanced vegetation index in different years and different periods;

以第i时期增强型植被指数的距平为自变量，以估产差值为因变量，构建线性回归方程，得到第i时期的估产模型，所述估产差值为统计产量与趋势产量的差值或统计产量与第i-1时期的估产值的差值，其中，i＝1,…,n；Taking the anomaly of the enhanced vegetation index in the ith period as the independent variable, and taking the estimated yield difference as the dependent variable, a linear regression equation is constructed to obtain the estimated yield model of the ith period, and the estimated yield difference is the difference between the statistical yield and the trend yield. Or the difference between the statistical output and the estimated output value of the i-1th period, where i=1,...,n;

将第i时期的增强型植被指数的距平代入第i时期的估产模型，得到第i时期的估产波动值；Substitute the anomaly of the enhanced vegetation index in the ith period into the estimated yield model of the ith period to obtain the estimated yield fluctuation value of the ith period;

根据第i时期的估产波动值和第i-1时期估产值计算第i时期的模型估产值；Calculate the model estimated output value of the i-th period according to the estimated output fluctuation value of the i-th period and the estimated output value of the i-1 period;

计算第i时期的估产模型的误差；Calculate the error of the estimated production model for the i-th period;

采用第i时期的估产模型的误差对第i时期的模型估产值进行修正，得到第i时期的估产值。The estimated output value of the model in the i-th period is corrected by using the error of the estimated output model of the i-th period to obtain the estimated output value of the i-th period.

可选的，所述以第i时期增强型植被指数的距平为自变量，以估产差值为因变量，构建线性回归方程，得到第i时期的估产模型，具体包括：Optionally, the anomaly of the enhanced vegetation index in the i-th period is used as the independent variable, and the estimated yield difference is used as the dependent variable, and a linear regression equation is constructed to obtain the i-th period. The estimated yield model specifically includes:

构建线性回归方程f(x_i)＝a×x_i+b，当i＝1时，f(x_i)为统计产量与趋势产量的差值，当i＞1时，f(x_i)为统计产量与第i-1时期估产值的差值，a和b均为线性回归方程的系数；Construct a linear regression equation f(x_i )=a×x_i +b, when i=1, f(x_i ) is the difference between statistical yield and trend yield, and when i>1, f(x_i ) is The difference between the statistical output and the estimated output value in the i-1 period, a and b are the coefficients of the linear regression equation;

确定线性回归方程的系数a和b，得到第i时期的估产模型。Determine the coefficients a and b of the linear regression equation, and obtain the estimated production model of the i-th period.

可选的，所述将第i时期的增强型植被指数的距平代入第i时期的估产模型，得到第i时期的估产波动值，具体包括：Optionally, the anomaly of the enhanced vegetation index in the i-th period is substituted into the yield estimation model in the i-th period to obtain the estimated yield fluctuation value in the i-th period, specifically including:

将第i时期的增强型植被指数的距平代入f'(x_i)＝a×x_i+b，其中，f'(x_i)为第i时期的估产波动值。Substitute the anomaly of the enhanced vegetation index in the i-th period into f'(x_i )=a×x_i +b, where f'(x_i ) is the estimated yield fluctuation value in the i-th period.

可选的，所述根据第i时期的估产波动值和第i-1时期估产值计算第i时期的估产值，具体包括：Optionally, calculating the estimated output value of the i-th period according to the estimated output fluctuation value of the i-th period and the estimated output value of the i-1 period, specifically including:

当i＝1时，根据E＝p+f'(x_i)计算第i时期的估产值，其中，p为趋势产量；When i=1, calculate the estimated output value of the i-th period according to E=p+f'(x_i ), where p is the trend output;

当i＞1时，根据E＝q+f'(x_i)计算第i时期的估产值，其中，q为第i-1时期的估产值。When i>1, the estimated output value of the i-th period is calculated according to E=q+f'(_xi ), where q is the estimated output value of the i-1-th period.

本发明还提供了一种粮食作物实时估产系统，所述系统包括：The present invention also provides a food crop real-time yield estimation system, the system includes:

数据获取模块，用于获取历史数据，所述历史数据包括不同年份不同时期的增强型植被指数、不同年份粮食作物的统计产量和不同年份粮食作物的趋势产量；a data acquisition module for acquiring historical data, the historical data including the enhanced vegetation index in different years and different periods, the statistical yield of grain crops in different years and the trend yield of grain crops in different years;

距平计算模块，用于根据不同年份不同时期的增强型植被指数计算增强型植被指数的距平；The anomaly calculation module is used to calculate the anomaly of the enhanced vegetation index according to the enhanced vegetation index of different years and different periods;

模型构建模块，用于以第i时期增强型植被指数的距平为自变量，以估产差值为因变量，构建线性回归方程，得到第i时期的估产模型，所述估产差值为统计产量与趋势产量的差值或统计产量与第i-1时期的估产值的差值，其中，i＝1,…,n；The model building module is used to use the anomaly of the enhanced vegetation index in the ith period as an independent variable and the estimated yield difference as a dependent variable to construct a linear regression equation to obtain an estimated yield model in the ith period, and the estimated yield difference is the statistical yield The difference from the trend yield or the difference between the statistical yield and the estimated output value of the i-1th period, where i=1,...,n;

波动值计算模块，用于将第i时期的增强型植被指数的距平代入第i时期的估产模型，得到第i时期的估产波动值；The fluctuation value calculation module is used to substitute the anomaly of the enhanced vegetation index in the i-th period into the yield estimation model in the i-th period to obtain the estimated yield fluctuation value in the i-th period;

模型估产值计算模块，用于根据第i时期的估产波动值和第i-1时期估产值计算第i时期的估产值；The model estimated output value calculation module is used to calculate the estimated output value of the ith period according to the estimated output fluctuation value of the ith period and the estimated output value of the i-1 period;

模型误差计算模块，用于计算第i时期的估产模型的误差；The model error calculation module is used to calculate the error of the production estimation model in the i-th period;

估产值计算模块，用于采用第i时期的估产模型的误差对第i时期的模型估产值进行修正，得到第i时期的估产值。The estimated output value calculation module is used to modify the estimated output value of the model in the ith period by using the error of the estimated output model of the ith period to obtain the estimated output value of the ith period.

可选的，所述模型构建模块具体包括：Optionally, the model building module specifically includes:

线性方程构建单元，用于构建线性回归方程f(x_i)＝a×x_i+b，当i＝1时，f(x_i)为统计产量与趋势单产的差值，当i＞1时，f(x_i)为统计产量与第i-1时期的估产值的差值，a和b均为线性回归方程的系数；Linear equation construction unit, used to construct a linear regression equation f(x_i )=a×x_i +b, when i=1, f(x_i ) is the difference between statistical yield and trend yield, when i>1 , f(x_i ) is the difference between the statistical output and the estimated output value of the i-1 period, a and b are the coefficients of the linear regression equation;

系数单元，用于确定线性回归方程的系数a和b，得到第i时期的估产模型。The coefficient unit is used to determine the coefficients a and b of the linear regression equation, and obtain the estimated production model of the i-th period.

可选的，所述波动值计算模块具体包括：Optionally, the fluctuation value calculation module specifically includes:

波动值计算单元，用于将第i时期的增强型植被指数的距平代入f'(x_i)＝a×x_i+b，其中，f'(x_i)为第i时期的估产波动值。The fluctuation value calculation unit is used to substitute the anomaly of the enhanced vegetation index in the i-th period into f'(x_i )=a×x_i +b, where f'(x_i ) is the estimated yield fluctuation value in the i-th period .

可选的，所述模型估产值计算模块具体包括：Optionally, the model estimated output value calculation module specifically includes:

估产值第一计算单元，用于当i＝1时，根据E＝p+f'(x_i)计算第i时期的估产值，其中，p为趋势产量；The first calculation unit of the estimated output value is used to calculate the estimated output value of the i-th period according to E=p+f'(x_i ) when i=1, where p is the trend output;

估产值第二计算单元，用于当i＞1时，根据E＝q+f'(x_i)计算第i时期的估产值，其中，q为第i-1时期的估产值。The second calculation unit of estimated output value is configured to calculate the estimated output value of the i-th period according to E=q+f'(_xi ) when i>1, where q is the estimated output value of the i-1-th period.

根据本发明提供的具体实施例，本发明公开了以下技术效果：本发明提供的粮食作物实时估产方法及系统以趋势产量为基数，定期计算由短期环境要素改变而引起的产量波动值，预估作物的产量。而且，本发明采用增强型植被指数作为环境影响因子，能够综合反应气象因子、土壤因子和田间管理因子的变化。进而，使得本发明提供的粮食作物实时估产方法及系统能够实现准确有效、实时性强的粮食作物估产。According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects: the method and system for estimating real-time yield of grain crops provided by the present invention take the trend yield as the base, and periodically calculate the yield fluctuation value caused by the change of short-term environmental factors, and estimate the yield fluctuation value. crop yield. Moreover, the present invention adopts the enhanced vegetation index as the environmental influence factor, which can comprehensively reflect the changes of meteorological factors, soil factors and field management factors. Furthermore, the real-time grain crop yield estimation method and system provided by the present invention can realize accurate, effective and real-time grain crop yield estimation.

附图说明Description of drawings

为了更清楚地说明本发明实施例或现有技术中的技术方案，下面将对实施例中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings required in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some of the present invention. In the embodiments, for those of ordinary skill in the art, other drawings can also be obtained according to these drawings without any creative effort.

图1为本发明实施例粮食作物实时估产方法的流程示意图；Fig. 1 is the schematic flow sheet of the real-time yield estimation method of grain crops according to the embodiment of the present invention;

图2为本发明实施例粮食作物实时估产系统的结构示意图。FIG. 2 is a schematic structural diagram of a real-time grain crop yield estimation system according to an embodiment of the present invention.

具体实施方式Detailed ways

下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

为使本发明的上述目的、特征和优点能够更加明显易懂，下面结合附图和具体实施方式对本发明作进一步详细的说明。In order to make the above objects, features and advantages of the present invention more clearly understood, the present invention will be described in further detail below with reference to the accompanying drawings and specific embodiments.

本发明基于历史数据求得趋势产量，计算每期估产的参数距平，进行模型估产，最后根据模型误差纠正：估产值＝模型估产值±模型估产值*模型平均绝对误差百分比。The present invention obtains the trend yield based on historical data, calculates the parameter anomaly of estimated yield in each period, carries out model yield estimation, and finally corrects according to the model error: estimated output value=model estimated output value±model estimated output value*model average absolute error percentage.

图1为本发明实施例粮食作物实时估产方法的流程示意图，如图1所示，本发明提供的粮食作物实时估产方法包括以下步骤：Fig. 1 is the schematic flow sheet of the method for estimating yield of grain crops in real time according to the embodiment of the present invention, and as shown in Fig. 1, the method for estimating yield in real time of grain crops provided by the present invention comprises the following steps:

步骤101：获取历史数据，历史数据包括不同年份不同时期的增强型植被指数、不同年份粮食作物的统计产量和不同年份粮食作物的趋势产量；Step 101: Obtain historical data, the historical data includes the enhanced vegetation index in different years and different periods, the statistical yield of grain crops in different years, and the trend yield of grain crops in different years;

步骤102：根据不同年份不同时期的增强型植被指数计算增强型植被指数的距平；Step 102: Calculate the anomaly of the enhanced vegetation index according to the enhanced vegetation index in different years and different periods;

步骤103：以第i时期增强型植被指数的距平为自变量，以估产差值为因变量，构建线性回归方程，得到第i时期的估产模型，所述估产差值为统计产量与趋势产量的差值或统计产量与第i-1时期的估产值的差值，其中，i＝1,…,n；Step 103: Taking the anomaly of the enhanced vegetation index in the ith period as the independent variable, and taking the estimated yield difference as the dependent variable, construct a linear regression equation to obtain the estimated yield model in the ith period, where the estimated yield difference is the statistical yield and the trend yield The difference of , or the difference between the statistical output and the estimated output value of the i-1th period, where i=1,...,n;

步骤104：将第i时期的增强型植被指数的距平代入第i时期的估产模型，得到第i时期的估产波动值；Step 104: Substitute the anomaly of the enhanced vegetation index in the i-th period into the yield estimation model in the i-th period to obtain the estimated yield fluctuation value in the i-th period;

步骤105：根据第i时期的估产波动值和第i-1时期估产值计算第i时期的估产值。Step 105: Calculate the estimated output value of the i-th period according to the estimated output fluctuation value of the i-th period and the estimated output value of the i-1-th period.

步骤106：计算第i时期的估产模型的误差；Step 106: Calculate the error of the production estimation model in the i-th period;

步骤107：采用第i时期的估产模型的误差对第i时期的模型估产值进行修正，得到第i时期的估产值。Step 107 : correcting the estimated output value of the model in the ith period by using the error of the estimated output model of the ith period, to obtain the estimated output value of the ith period.

其中，步骤103具体包括：Wherein, step 103 specifically includes:

构建线性回归方程f(x_i)＝a×x_i+b，当i＝1时，f(x_i)为统计产量与趋势产量的差值，当i＞1时，f(x_i)为统计产量与第i-1时期的估产值的差值，a和b均为线性回归方程的系数；Construct a linear regression equation f(x_i )=a×x_i +b, when i=1, f(x_i ) is the difference between statistical yield and trend yield, and when i>1, f(x_i ) is The difference between the statistical output and the estimated output value of the i-1 period, a and b are the coefficients of the linear regression equation;

步骤104具体包括：Step 104 specifically includes:

步骤105具体包括：Step 105 specifically includes:

具体的操作方案如下：The specific operation plan is as follows:

1)估产时空尺度的确定1) Determination of the temporal and spatial scale of production estimates

选定估产区域、估产对象、估产起始时间，估产结束时间，估产周期。比如，发明案例估产区为河北省邢台县，估产对象为冬小麦，估产起始时间为10月17日，估产结束时间为6月27日(冬小麦跨年生长，估产时间顺序递进)，估产周期为16天(决定环境影响参数数据递进的时间尺度，本例中10月17日第一次估产环境参数则为10月1日-10月16日的均值距平，在此之前的影响忽略不计)，因此，总共估产17次。模型构建数据时间尺度为2001-2015年，估产目标时间为2016年。Select the production estimation area, the production estimation object, the production estimation start time, the production estimation end time, and the production estimation cycle. For example, the estimated production area of the invention case is Xingtai County, Hebei Province, and the estimated production target is winter wheat. The estimated production start time is October 17, and the estimated production end time is June 27 (winter wheat grows across the year, and the estimated production time sequence is progressive), and the estimated production cycle It is 16 days (determining the time scale of the environmental impact parameter data progression, in this example, the first estimated production environment parameter on October 17th is the average deviation from October 1st to October 16th, and the impact before that is ignored. not counted), therefore, a total of 17 births are estimated. The model construction data time scale is 2001-2015, and the estimated production target time is 2016.

2)数据的获取及处理。2) Data acquisition and processing.

本案例数据的获取主要包括估产区不同年份的冬小麦统计产量和不同年份不同时期的增强型植被指数(EVI)。具体为1986年至2016年河北省邢台县的冬小麦单位面积产量，本案例由统计年鉴中的冬小麦总产量和播种面积计算得到，并以直线滑动平均方法计算得到2001-2016年的趋势产量(表1-表17第1列、第2列)；2001年至2016年河北省邢台县冬小麦种植区10月17日至第二年6月27日以16天为周期的遥感反演增强型植被指数(EVI)，本案例中由MODIS产品数据和由遥感数据提取的冬小麦分布区数据计算得到。The acquisition of data in this case mainly includes estimating the statistical yield of winter wheat in different years in the production area and the enhanced vegetation index (EVI) in different years and periods. Specifically, the yield per unit area of winter wheat in Xingtai County, Hebei Province from 1986 to 2016 was calculated from the total yield and sown area of winter wheat in the Statistical Yearbook, and the trend yield from 2001 to 2016 was calculated by the linear moving average method (Table 1). 1-Table 17, column 1 and column 2); Remote sensing inversion enhanced vegetation index with a cycle of 16 days from October 17 to June 27 of the next year in the winter wheat planting area of Xingtai County, Hebei Province from 2001 to 2016 (EVI), calculated in this case from MODIS product data and winter wheat distribution data extracted from remote sensing data.

3)不同年份不同时期的增强型植被指数距平的计算(EVI距平)3) Calculation of enhanced vegetation index anomalies in different years and periods (EVI anomalies)

以2001年至2015年冬小麦种植区不同时期EVI计算得到各自的EVI距平(表1-表17第4列)。The respective EVI anomalies were calculated based on the EVI of winter wheat planting areas in different periods from 2001 to 2015 (Table 1-Table 17, column 4).

4)基于历史数据的估产模型构建4) Construction of production estimation model based on historical data

(1)第一次估产模型构建(1) Construction of the first production estimation model

计算2001年至2015年实际产量(表1第1列)和趋势产量(表1第2列)的实际差值(表1第3列)；Calculate the actual difference (column 3 of table 1) between actual production (column 1 of table 1) and trend yield (column 2 of table 1) from 2001 to 2015;

实际差值(表1第3列)和第1个周期的EVI距平(表1第4列)进行线性回归得到第一次估产模型的回归系数a＝10.466，b＝-96.942，即得估产波动模型1＝10.466×EVI_10-16-96.942，由此可得估产波动值1(表1第5列)；估产模型1＝趋势单产+估产波动模型1＝趋势单产+10.466×EVI_10-16-96.942，由此可得模型估产值1(表1第6列)；Perform linear regression on the actual difference (column 3 of Table 1) and the EVI anomaly of the first cycle (column 4 of Table 1) to obtain the regression coefficients a=10.466 and b=-96.942 of the first estimated yield model, that is, the estimated yield Fluctuation model 1=10.466×EVI_10-16_-96.942 , from which the estimated output fluctuation value 1 (the fifth column of Table 1) can be obtained; -96.942, from which the model estimated output value 1 (column 6 of Table 1) can be obtained;

由实际产量(表1第1列)和模型估产值1(表1第6列)求差得到估产差值1(表1第7列)，进而由估产差值1(表1第7列)与实际产量(表1第1列)求绝对比值得到模型误差1(表1第8列)。本阶段模型平均误差为2％。Calculate the difference between the actual output (column 1 of Table 1) and the model estimated output value 1 (column 6 of Table 1) to obtain the estimated yield difference 1 (column 7 of Table 1), and then calculate the estimated yield difference 1 (column 7 of Table 1) The absolute ratio with the actual yield (column 1 of Table 1) yields a model error of 1 (column 8 of Table 1). The average error of the model at this stage is 2%.

综上，第一次估产模型＝估产模型1×(1±2％)＝(趋势单产+10.466×EVI_10-16-96.942)×(1±2％)。To sum up, the first estimated yield model=estimated yield model 1×(1±2%)=(trend yield+10.466×EVI_10-16 -96.942)×(1±2%).

表1Table 1

(2)第二次估产模型构建(2) Construction of the second production estimation model

计算2001年至2015年实际产量(表2第1列)和模型估产值1(表2第2列)的估产差值1(表2第3列)；Calculate the estimated yield difference 1 (column 3 of Table 2) between actual production (column 1 of Table 2) and model estimated output 1 (column 2 of Table 2) from 2001 to 2015;

估产差值1(表2第3列)和第2个周期的EVI距平(表2第4列)进行线性回归得到第二次估产模型的回归系数a＝10.41，b＝0.4364，即得估产波动模型2＝10.41×EVI_11-1+0.4364，由此可得估产波动值2(表2第5列)；估产模型2＝模型估产值1+估产波动模型2＝模型估产值1+10.41×EVI_11-1+0.4364，由此可得模型估产值2(表2第6列)；Perform linear regression on the estimated yield difference 1 (column 3 of Table 2) and the EVI anomaly of the second cycle (column 4 of Table 2) to obtain the regression coefficients of the second yield estimation model a=10.41, b=0.4364, that is, the estimated yield Fluctuation model 2=10.41×EVI_11-1 +0.4364, from which the estimated output fluctuation value 2 (the fifth column of Table 2) can be obtained; EVI_11-1 +0.4364, from which the estimated output value of the model 2 (column 6 of Table 2) can be obtained;

由实际产量(表2第1列)和模型估产值2(表2第6列)求差得到估产差值2(表2第7列)，进而由估产差值2(表2第7列)与实际产量(表2第1列)求绝对比值得到模型误差2(表2第8列)。本阶段模型平均误差为2％。Calculate the difference between the actual output (column 1 of Table 2) and the model estimated output value 2 (column 6 of Table 2) to obtain the estimated yield difference 2 (column 7 of Table 2), and then calculate the estimated yield difference 2 (column 7 of Table 2) Calculate the absolute ratio with the actual production (column 1 of Table 2) to obtain the model error 2 (column 8 of Table 2). The average error of the model at this stage is 2%.

综上，第二次估产模型＝估产模型2×(1±2％)＝(模型估产值1+10.41×EVI_11-1+0.4364)×(1±2％)。To sum up, the second production estimation model=estimated production model 2×(1±2%)=(model estimated production value 1+10.41×EVI_11-1 +0.4364)×(1±2%).

表2Table 2

(3)第三次估产模型构建(3) Construction of the third production estimation model

计算2001年至2015年实际产量(表3第1列)和模型估产值2(表3第2列)的估产差值2(表3第3列)；Calculate the estimated yield difference 2 (Table 3, column 3) between actual production (Table 3, column 1) and model estimated output 2 (Table 3, column 2) from 2001 to 2015;

估产差值3(表3第3列)和第3个周期的EVI距平(表3第4列)进行线性回归得到第二次估产模型的回归系数a＝10.324，b＝0.6597，即得估产波动模型3＝10.324×EVI_11-17+0.6597，由此可得估产波动值3(表3第5列)；估产模型3＝模型估产值2+估产波动模型3＝模型估产值2+10.324×EVI_11-17+0.6597，由此可得模型估产值3(表3第6列)；Perform linear regression on the estimated yield difference 3 (column 3 of Table 3) and the EVI anomaly of the third cycle (column 4 of Table 3) to obtain the regression coefficients of the second yield estimation model a=10.324, b=0.6597, that is, the estimated yield Fluctuation model 3=10.324×EVI_11-17 +0.6597, from which the estimated output fluctuation value 3 (the fifth column of Table 3) can be obtained; EVI_11-17 +0.6597, from which the model estimated output value 3 (column 6 of Table 3) can be obtained;

由实际产量(表3第1列)和模型估产值3(表3第6列)求差得到估产差值3(表3第7列)，进而由估产差值3(表3第7列)与实际产量(表3第1列)求绝对比值得到模型误差3(表3第8列)。本阶段模型平均误差为2％。Calculate the difference between the actual output (column 1 of Table 3) and the model estimated output value 3 (column 6 of Table 3) to obtain the estimated yield difference 3 (column 7 of Table 3), and then calculate the estimated yield difference 3 (column 7 of Table 3) Calculate the absolute ratio with the actual production (column 1 of Table 3) to obtain the model error 3 (column 8 of Table 3). The average error of the model at this stage is 2%.

综上，第三次估产模型＝估产模型2×(1±2％)＝(模型估产值2+10.324×EVI_11-17+0.6597)×(1±2％)。To sum up, the third production estimation model=estimated production model 2×(1±2%)=(model estimated production value 2+10.324×EVI_11-17 +0.6597)×(1±2%).

表3table 3

(4)第四次估产模型构建(4) Construction of the fourth production estimation model

计算2001年至2015年实际产量(表4第1列)和模型估产值3(表4第2列)的估产差值3(表4第3列)；Calculate the estimated yield difference3 (Table 4, column 3) between actual production (Table 4, column 1) and model estimated output3 (Table 4, column 2) from 2001 to 2015;

估产差值3(表4第3列)和第4个周期的EVI距平(表4第4列)进行线性回归得到第二次估产模型的回归系数a＝10.224，b＝0.767，即得估产波动模型4＝10.224×EVI_12-3+0.767，由此可得估产波动值4(表4第5列)；估产模型4＝模型估产值3+估产波动模型4＝模型估产值3+10.224×EVI_12-3+0.767，由此可得模型估产值4(表4第6列)；Perform linear regression on the estimated yield difference 3 (the third column of Table 4) and the EVI anomaly of the fourth cycle (the fourth column of Table 4) to obtain the regression coefficients of the second estimated yield model a=10.224, b=0.767, that is, the estimated yield Fluctuation model 4 = 10.224×EVI_12-3 +0.767, from which the estimated output fluctuation value 4 (the fifth column of Table 4) can be obtained; estimated yield model 4 = model estimated output value 3 + estimated yield fluctuation model 4 = model estimated output value 3 + 10.224× EVI_12-3 +0.767, from which the model estimated output value 4 (column 6 of Table 4) can be obtained;

由实际产量(表4第1列)和模型估产值4(表4第6列)求差得到估产差值4(表4第7列)，进而由估产差值4(表4第7列)与实际产量(表4第1列)求绝对比值得到模型误差4(表4第8列)。本阶段模型平均误差为2％。Calculate the difference between the actual output (column 1 of Table 4) and the model estimated output value 4 (column 6 of Table 4) to obtain the estimated yield difference 4 (column 7 of Table 4), and then calculate the estimated yield difference 4 (column 7 of Table 4) The absolute ratio with the actual yield (column 1 of Table 4) yields a model error of 4 (column 8 of Table 4). The average error of the model at this stage is 2%.

综上，第四次估产模型＝估产模型4×(1±2％)＝(模型估产值3+10.224×EVI_12-3+0.767)×(1±2％)。To sum up, the fourth production estimation model=estimated production model 4×(1±2%)=(model estimated production value 3+10.224×EVI_12-3 +0.767)×(1±2%).

表4Table 4

(5)第五次估产模型构建(5) Construction of the fifth production estimation model

计算2001年至2015年实际产量(表5第1列)和模型估产值4(表5第2列)的估产差值4(表5第3列)；Calculate the difference between the actual production (column 1 of Table 5) and the estimated output value of the model4 (column 2 of Table 5) between 2001 and 2015 (column 3 of Table 5);

估产差值4(表5第3列)和第5个周期的EVI距平(表5第4列)进行线性回归得到第二次估产模型的回归系数a＝10.123，b＝0.8014，即得估产波动模型5＝10.123×EVI_12-19+0.8014，由此可得估产波动值5(表5第5列)；估产模型5＝模型估产值4+估产波动模型5＝模型估产值4+10.123×EVI_12-19+0.8014，由此可得模型估产值5(表5第6列)；Perform linear regression on the estimated yield difference 4 (the third column of Table 5) and the EVI anomaly of the fifth cycle (the fourth column of Table 5) to obtain the regression coefficients of the second estimated yield model a=10.123, b=0.8014, that is, the estimated yield Fluctuation model 5=10.123×EVI_12-19 +0.8014, from which the estimated output fluctuation value 5 (the fifth column of Table 5) can be obtained; EVI_12-19 +0.8014, from which the model estimated output value 5 (column 6 of Table 5) can be obtained;

由实际产量(表5第1列)和模型估产值5(表5第6列)求差得到估产差值5(表5第7列)，进而由估产差值5(表5第7列)与实际产量(表5第1列)求绝对比值得到模型误差5(表5第8列)。本阶段模型平均误差为2％。Calculate the difference between the actual output (column 1 of Table 5) and the model estimated output value 5 (column 6 of Table 5) to obtain the estimated yield difference 5 (column 7 of Table 5), and then calculate the estimated yield difference 5 (column 7 of Table 5) The absolute ratio with the actual yield (column 1 of Table 5) yields a model error of 5 (column 8 of Table 5). The average error of the model at this stage is 2%.

综上，第五次估产模型＝估产模型5×(1±2％)＝(模型估产值4+10.123×EVI_12-19+0.8014)×(1±2％)。To sum up, the fifth production estimation model=estimated production model 5×(1±2%)=(model estimated production value 4+10.123×EVI_12-19 +0.8014)×(1±2%).

表5table 5

(6)第六次估产模型构建(6) Construction of the sixth production estimation model

计算2001年至2015年实际产量(表6第1列)和模型估产值5(表6第2列)的估产差值5(表6第3列)；Calculate the difference between the actual production (column 1 of table 6) and the estimated output value of the model 5 (column 2 of table 6) from 2001 to 2015 (column 3 of table 6);

估产差值5(表6第3列)和第6个周期的EVI距平(表6第4列)进行线性回归得到第二次估产模型的回归系数a＝10.023，b＝0.8081，即得估产波动模型6＝10.023×EVI_1-1+0.8081，由此可得估产波动值6(表6第5列)；估产模型6＝模型估产值5+估产波动模型6＝模型估产值5+10.023×EVI_1-1+0.8081，由此可得模型估产值6(表6第6列)；Perform linear regression on the estimated yield difference 5 (the third column of Table 6) and the EVI anomaly of the sixth cycle (the fourth column of Table 6) to obtain the regression coefficients of the second estimated yield model a=10.023, b=0.8081, that is, the estimated yield Fluctuation model 6=10.023×EVI_1-1 +0.8081, from which the estimated output fluctuation value 6 can be obtained (the fifth column of Table 6); EVI_1-1 +0.8081, from which the estimated output value 6 of the model can be obtained (column 6 of Table 6);

由实际产量(表6第1列)和模型估产值6(表6第6列)求差得到估产差值6(表6第7列)，进而由估产差值6(表6第7列)与实际产量(表6第1列)求绝对比值得到模型误差6(表6第8列)。本阶段模型平均误差为2％。Calculate the difference between the actual output (column 1 of Table 6) and the model estimated output value 6 (column 6 of Table 6) to obtain the estimated yield difference 6 (column 7 of Table 6), and then calculate the estimated yield difference 6 (column 7 of Table 6) The absolute ratio with the actual yield (column 1 of Table 6) yields a model error of 6 (column 8 of Table 6). The average error of the model at this stage is 2%.

综上，第六次估产模型＝估产模型6×(1±2％)＝(模型估产值5+10.023×EVI_1-1+0.8081)×(1±2％)。To sum up, the sixth estimated production model=estimated production model 6×(1±2%)=(model estimated production value 5+10.023×EVI_1-1 +0.8081)×(1±2%).

表6Table 6

(7)第七次估产模型构建(7) Construction of the seventh production estimation model

计算2001年至2015年实际产量(表7第1列)和模型估产值6(表7第2列)的估产差值6(表7第3列)；Calculate the estimated yield difference6 (Table 7, column 3) between actual production (Table 7, column 1) and model estimated output6 (Table 7, column 2) from 2001 to 2015;

估产差值6(表7第3列)和第7个周期的EVI距平(表7第4列)进行线性回归得到第二次估产模型的回归系数a＝9.9352，b＝0.6943，即得估产波动模型7＝9.9352×EVI_1-17+0.6943，由此可得估产波动值7(表7第5列)；估产模型7＝模型估产值6+估产波动模型7＝模型估产值6+9.9352×EVI_1-17+0.6943，由此可得模型估产值7(表7第6列)；Perform linear regression on the estimated yield difference 6 (the third column of Table 7) and the EVI anomaly of the seventh cycle (the fourth column of Table 7) to obtain the regression coefficients of the second estimated yield model a=9.9352, b=0.6943, that is, the estimated yield Fluctuation model 7=9.9352×EVI_1-17 +0.6943, from which the estimated output fluctuation value 7 (the fifth column of Table 7) can be obtained; EVI_1-17 +0.6943, from which the model estimated output value 7 (column 6 of Table 7) can be obtained;

由实际产量(表7第1列)和模型估产值7(表7第6列)求差得到估产差值7(表7第7列)，进而由估产差值7(表7第7列)与实际产量(表7第1列)求绝对比值得到模型误差7(表7第8列)。本阶段模型平均误差为2％。Calculate the difference between the actual output (column 1 of Table 7) and the model estimated output value 7 (column 6 of Table 7) to obtain the estimated yield difference 7 (column 7 of Table 7), and then calculate the estimated yield difference 7 (column 7 of Table 7) The absolute ratio of the actual yield (column 1 of Table 7) yields the model error of 7 (column 8 of Table 7). The average error of the model at this stage is 2%.

综上，第七次估产模型＝估产模型7×(1±2％)＝(模型估产值6+9.9352×EVI_1-17+0.6943)×(1±2％)。To sum up, the seventh production estimation model=estimated production model 7×(1±2%)=(model estimated production value 6+9.9352×EVI_1-17 +0.6943)×(1±2%).

表7Table 7

(8)第八次估产模型构建(8) Construction of the eighth production estimation model

计算2001年至2015年实际产量(表8第1列)和模型估产值7(表8第2列)的估产差值7(表8第3列)；Calculate the estimated production difference7 (Table 8, column 3) between the actual production (column 1 of Table 8) and the model estimated output7 (column 2 of Table 8) from 2001 to 2015;

估产差值7(表8第3列)和第8个周期的EVI距平(表8第4列)进行线性回归得到第二次估产模型的回归系数a＝9.8789，b＝0.4544，即得估产波动模型8＝9.8789×EVI_2-2+0.4544，由此可得估产波动值8(表8第5列)；估产模型8＝模型估产值7+估产波动模型8＝模型估产值7+9.8789×EVI_2-2+0.4544，由此可得模型估产值8(表8第6列)；Perform linear regression on the estimated yield difference 7 (the third column of Table 8) and the EVI anomaly of the 8th cycle (the fourth column of Table 8) to obtain the regression coefficients of the second estimated yield model a=9.8789, b=0.4544, that is, the estimated yield Fluctuation model 8 = 9.8789×EVI_2-2 +0.4544, from which the estimated output fluctuation value 8 (the fifth column of Table 8) can be obtained; estimated yield model 8 = model estimated output value 7 + estimated yield fluctuation model 8 = model estimated output value 7 + 9.8789× EVI_2-2 +0.4544, from which the estimated output value of the model 8 (column 6 of Table 8) can be obtained;

由实际产量(表8第1列)和模型估产值8(表8第6列)求差得到估产差值8(表8第7列)，进而由估产差值8(表8第7列)与实际产量(表8第1列)求绝对比值得到模型误差8(表8第8列)。本阶段模型平均误差为2％。Calculate the difference between the actual output (column 1 of Table 8) and the model estimated output value 8 (column 6 of Table 8) to obtain the estimated yield difference 8 (column 7 of Table 8), and then calculate the estimated yield difference 8 (column 7 of Table 8) The absolute ratio of the actual yield (column 1 of Table 8) yields a model error of 8 (column 8 of Table 8). The average error of the model at this stage is 2%.

综上，第八次估产模型＝估产模型8×(1±2％)＝(模型估产值7+9.8789×EVI_2-2+0.4544)×(1±2％)。To sum up, the eighth production estimation model=estimated production model 8×(1±2%)=(model estimated production value 7+9.8789×EVI_2-2 +0.4544)×(1±2%).

表8Table 8

(9)第九次估产模型构建(9) Construction of the ninth production estimation model

计算2001年至2015年实际产量(表9第1列)和模型估产值8(表9第2列)的估产差值8(表9第3列)；Calculate the difference between the actual production (column 1 of table 9) and the estimated output value of the model8 (column 2 of table 9) from 2001 to 2015 between the estimated production values8 (column 3 of table 9);

估产差值8(表9第3列)和第9个周期的EVI距平(表9第4列)进行线性回归得到第二次估产模型的回归系数a＝9.8522，b＝0.2286，即得估产波动模型8＝9.8522×EVI_2-18+0.2286，由此可得估产波动值9(表9第5列)；估产模型9＝模型估产值8+估产波动模型9＝模型估产值8+9.8522×EVI_2-18+0.2286，由此可得模型估产值9(表9第6列)；Perform linear regression on the estimated yield difference 8 (the third column of Table 9) and the EVI anomaly of the 9th cycle (the fourth column of Table 9) to obtain the regression coefficients of the second estimated yield model a=9.8522, b=0.2286, that is, the estimated yield Fluctuation model 8 = 9.8522×EVI_2-18 +0.2286, from which the estimated output fluctuation value 9 (the fifth column of Table 9) can be obtained; estimated output model 9 = model estimated output value 8 + estimated output fluctuation model 9 = model estimated output value 8 + 9.8522× EVI_2-18 +0.2286, from which the estimated output value of the model 9 (column 6 of Table 9) can be obtained;

由实际产量(表9第1列)和模型估产值9(表9第6列)求差得到估产差值9(表9第7列)，进而由估产差值9(表9第7列)与实际产量(表9第1列)求绝对比值得到模型误差9(表9第8列)。本阶段模型平均误差为2％。Calculate the difference between the actual output (column 1 of Table 9) and the estimated output value of the model 9 (column 6 of Table 9) to obtain the estimated yield difference 9 (column 7 of Table 9), and then calculate the estimated yield difference 9 (column 7 of Table 9) The absolute ratio with the actual yield (column 1 of Table 9) yields a model error of 9 (column 8 of Table 9). The average error of the model at this stage is 2%.

综上，第九次估产模型＝估产模型9×(1±2％)＝(模型估产值8+9.8522×EVI_2-18+0.2286)×(1±2％)。To sum up, the ninth production estimation model=estimated production model 9×(1±2%)=(model estimated production value 8+9.8522×EVI_2-18 +0.2286)×(1±2%).

表9Table 9

(10)第十次估产模型构建(10) Construction of the tenth production estimation model

计算2001年至2015年实际产量(表10第1列)和模型估产值9(表10第2列)的估产差值9(表10第3列)；Calculate the difference between the actual production (column 1 of table 10) and the estimated output value of the model 9 (column 2 of table 10) from 2001 to 2015;

估产差值9(表10第3列)和第10个周期的EVI距平(表10第4列)进行线性回归得到第二次估产模型的回归系数a＝9.8423，b＝0.0923，即得估产波动模型10＝9.8423×EVI_3-6+0.0923，由此可得估产波动值10(表10第5列)；估产模型10＝模型估产值9+估产波动模型10＝模型估产值9+9.8423×EVI_3-6+0.0923，由此可得模型估产值10(表10第6列)；Perform linear regression on the estimated yield difference 9 (the third column of Table 10) and the EVI anomaly of the 10th cycle (the fourth column of Table 10) to obtain the regression coefficients of the second estimated yield model a=9.8423, b=0.0923, that is, the estimated yield Fluctuation model 10=9.8423×EVI_3-6 +0.0923, from which the estimated output fluctuation value 10 (the fifth column of Table 10) can be obtained; estimated yield model 10=model estimated output value 9+ estimated yield fluctuation model 10=model estimated output value 9+9.8423× EVI_3-6 +0.0923, from which the estimated output value of the model is 10 (column 6 of Table 10);

由实际产量(表10第1列)和模型估产值10(表10第6列)求差得到估产差值10(表10第7列)，进而由估产差值10(表10第7列)与实际产量(表10第1列)求绝对比值得到模型误差10(表10第8列)。本阶段模型平均误差为2％。Calculate the difference between the actual output (column 1 of Table 10) and the model estimated output value 10 (column 6 of Table 10) to obtain the estimated yield difference 10 (column 7 of Table 10), and then calculate the estimated yield difference 10 (column 7 of Table 10) The absolute ratio of the actual yield (column 1 of Table 10) yields a model error of 10 (column 8 of Table 10). The average error of the model at this stage is 2%.

综上，第十次估产模型＝估产模型10×(1±2％)＝(模型估产值9+9.8423×EVI_3-6+0.0923)×(1±2％)。To sum up, the tenth production estimation model=estimated production model 10×(1±2%)=(model estimated production value 9+9.8423×EVI_3-6 +0.0923)×(1±2%).

表10Table 10

(11)第十一次估产模型构建(11) Construction of the eleventh production estimation model

计算2001年至2015年实际产量(表11第1列)和模型估产值10(表11第2列)的估产差值10(表11第3列)；Calculate the difference between the actual production (column 1 of Table 11) and the estimated output value of the model 10 (column 2 of Table 11) between 2001 and 2015 (column 3 of Table 11);

估产差值10(表11第3列)和第11个周期的EVI距平(表11第4列)进行线性回归得到第二次估产模型的回归系数a＝9.8386，b＝0.0342，即得估产波动模型11＝9.8386×EVI_3-22+0.0342，由此可得估产波动值11(表11第5列)；估产模型11＝模型估产值10+估产波动模型11＝模型估产值10+9.8386×EVI_3-22+0.0342，由此可得模型估产值11(表11第6列)；Perform linear regression on the estimated yield difference of 10 (the third column of Table 11) and the EVI anomaly of the 11th cycle (the fourth column of Table 11) to obtain the regression coefficients of the second estimated yield model a=9.8386, b=0.0342, that is, the estimated yield Fluctuation model 11=9.8386×EVI_3-22 +0.0342, from which the estimated output fluctuation value 11 can be obtained (the fifth column of Table 11); EVI_3-22 +0.0342, from which the estimated output value of the model 11 (column 6 of Table 11) can be obtained;

由实际产量(表11第1列)和模型估产值11(表11第6列)求差得到估产差值11(表11第7列)，进而由估产差值11(表11第7列)与实际产量(表11第1列)求绝对比值得到模型误差11(表11第8列)。本阶段模型平均误差为2％。Calculate the difference between the actual output (column 1 of Table 11) and the model estimated output value 11 (column 6 of Table 11) to obtain the estimated yield difference 11 (column 7 of Table 11), and then calculate the estimated yield difference 11 (column 7 of Table 11) The absolute ratio with the actual yield (column 1 of Table 11) yields the model error 11 (column 8 of Table 11). The average error of the model at this stage is 2%.

综上，第十一次估产模型＝估产模型11×(1±2％)＝(模型估产值10+9.8386×EVI_3-22+0.0342)×(1±2％)。To sum up, the eleventh production estimation model=estimated production model 11×(1±2%)=(model estimated production value 10+9.8386×EVI_3-22 +0.0342)×(1±2%).

表11Table 11

(12)第十二次估产模型构建(12) Construction of the Twelfth Production Estimation Model

计算2001年至2015年实际产量(表12第1列)和模型估产值11(表12第2列)的估产差值11(表12第3列)；Calculate the estimated yield difference between 2001 and 2015 (column 1 of table 12) and the estimated output value of the model11 (column 2 of table 12);

估产差值11(表12第3列)和第12个周期的EVI距平(表12第4列)进行线性回归得到第二次估产模型的回归系数a＝9.842，b＝-0.0164，即得估产波动模型12＝9.842×EVI_4-7-0.0164，由此可得估产波动值12(表12第5列)；估产模型12＝模型估产值11+估产波动模型12＝模型估产值11+9.842×EVI_4-7-0.0164，由此可得模型估产值12(表12第6列)；Perform linear regression on the estimated yield difference 11 (the third column of Table 12) and the EVI anomaly of the 12th cycle (the fourth column of Table 12) to obtain the regression coefficients of the second estimated yield model a=9.842, b=-0.0164, that is, Estimated output fluctuation model 12=9.842×EVI_4-7 -0.0164, from which the estimated output fluctuation value 12 can be obtained (the fifth column of Table 12); ×EVI_4-7 -0.0164, from which the estimated output value of the model can be obtained as 12 (column 6 of Table 12);

由实际产量(表12第1列)和模型估产值12(表12第6列)求差得到估产差值12(表12第7列)，进而由估产差值12(表12第7列)与实际产量(表12第1列)求绝对比值得到模型误差12(表12第8列)。本阶段模型平均误差为2％。Calculate the difference between the actual output (column 1 of Table 12) and the model estimated output value 12 (column 6 of Table 12) to obtain the estimated yield difference 12 (column 7 of Table 12), and then calculate the estimated yield difference 12 (column 7 of Table 12) The absolute ratio of the actual yield (column 1 of Table 12) yields the model error 12 (column 8 of Table 12). The average error of the model at this stage is 2%.

综上，第十二次估产模型＝估产模型12×(1±2％)＝(模型估产值11+9.842×EVI_4-7-0.0164)×(1±2％)。To sum up, the twelfth production estimation model=estimated production model 12×(1±2%)=(model estimated production value 11+9.842×EVI_4-7 -0.0164)×(1±2%).

表12Table 12

(13)第十三次估产模型构建(13) Construction of the Thirteenth Production Estimation Model

计算2001年至2015年实际产量(表13第1列)和模型估产值12(表13第2列)的估产差值12(表13第3列)；Calculate the estimated yield difference12 (Table 13, column 3) between actual production (Table 13, column 1) and model estimated output12 (Table 13, column 2) from 2001 to 2015;

估产差值12(表13第3列)和第13个周期的EVI距平(表13第4列)进行线性回归得到第二次估产模型的回归系数a＝9.848，b＝-0.0294，即得估产波动模型13＝9.848×EVI_4-23-0.0294，由此可得估产波动值13(表13第5列)；估产模型13＝模型估产值12+估产波动模型13＝模型估产值12+9.848×EVI_4-23-0.0294，由此可得模型估产值13(表13第6列)；Perform linear regression on the estimated yield difference 12 (the third column of Table 13) and the EVI anomaly of the 13th cycle (the fourth column of Table 13) to obtain the regression coefficients of the second estimated yield model a=9.848, b=-0.0294, that is, Estimated yield fluctuation model 13=9.848×EVI_4-23 -0.0294, from which the estimated yield fluctuation value 13 (the fifth column of Table 13) can be obtained; ×EVI_4-23 -0.0294, from which the estimated output value of the model 13 (the sixth column of Table 13) can be obtained;

由实际产量(表13第1列)和模型估产值13(表13第6列)求差得到估产差值13(表13第7列)，进而由估产差值13(表13第7列)与实际产量(表13第1列)求绝对比值得到模型误差13(表13第8列)。本阶段模型平均误差为2％。Calculate the difference between the actual output (column 1 of Table 13) and the model estimated output value 13 (column 6 of Table 13) to obtain the estimated yield difference 13 (column 7 of Table 13), and then calculate the estimated yield difference 13 (column 7 of Table 13) The absolute ratio of the actual yield (column 1 of Table 13) yields the model error 13 (column 8 of Table 13). The average error of the model at this stage is 2%.

综上，第十三次估产模型＝估产模型13×(1±2％)＝(模型估产值12+9.848×EVI_4-23-0.0294)×(1±2％)。To sum up, the thirteenth production estimation model=estimated production model 13×(1±2%)=(model estimated production value 12+9.848×EVI_4-23 -0.0294)×(1±2%).

表13Table 13

(14)第十四次估产模型构建(14) Construction of the fourteenth production estimation model

计算2001年至2015年实际产量(表14第1列)和模型估产值13(表14第2列)的估产差值13(表14第3列)；Calculate the estimated yield difference13 (column 3 of table 14) between actual production (column 1 of table 14) and model estimated output13 (column 2 of table 14) from 2001 to 2015;

估产差值13(表14第3列)和第14个周期的EVI距平(表14第4列)进行线性回归得到第二次估产模型的回归系数a＝9.8232，b＝0.1986，即得估产波动模型14＝9.8232×EVI_5-9+0.1986，由此可得估产波动值14(表14第5列)；估产模型14＝模型估产值13+估产波动模型14＝模型估产值13+9.8232×EVI_5-9+0.1986，由此可得模型估产值14(表14第6列)；Perform linear regression on the estimated yield difference 13 (the third column of Table 14) and the EVI anomaly of the 14th cycle (the fourth column of Table 14) to obtain the regression coefficients of the second estimated yield model a=9.8232, b=0.1986, that is, the estimated yield Fluctuation model 14=9.8232×EVI_5-9 +0.1986, from which the estimated output fluctuation value 14 (5th column of Table 14) can be obtained; EVI_5-9 +0.1986, from which the model estimated output value 14 (column 6 of Table 14) can be obtained;

由实际产量(表14第1列)和模型估产值14(表14第6列)求差得到估产差值14(表14第7列)，进而由估产差值14(表14第7列)与实际产量(表14第1列)求绝对比值得到模型误差14(表14第8列)。本阶段模型平均误差为2％。Calculate the difference between the actual output (column 1 of Table 14) and the model estimated output value 14 (column 6 of Table 14) to obtain the estimated yield difference 14 (column 7 of Table 14), and then calculate the estimated yield difference 14 (column 7 of Table 14) The absolute ratio of the actual yield (column 1 of Table 14) yields the model error 14 (column 8 of Table 14). The average error of the model at this stage is 2%.

综上，第十四次估产模型＝估产模型14×(1±2％)＝(模型估产值13+9.8232×EVI_5-9+0.1986)×(1±2％)。To sum up, the fourteenth production estimation model=estimated production model 14×(1±2%)=(model estimated production value 13+9.8232×EVI_5-9 +0.1986)×(1±2%).

表14Table 14

(15)第十五次估产模型构建(15) Construction of the fifteenth production estimation model

计算2001年至2015年实际产量(表15第1列)和模型估产值14(表15第2列)的估产差值14(表15第3列)；Calculate the difference between actual production (column 1 of Table 15) and model estimated output14 (column 2 of Table 15) from 2001 to 2015 between the estimated yields14 (column 3 of Table 15);

估产差值14(表15第3列)和第15个周期的EVI距平(表15第4列)进行线性回归得到第二次估产模型的回归系数a＝9.7529，b＝0.5682，即得估产波动模型15＝9.7529×EVI_5-25+0.5682，由此可得估产波动值15(表15第5列)；估产模型15＝模型估产值14+估产波动模型15＝模型估产值14+9.7529×EVI_5-25+0.5682，由此可得模型估产值15(表15第6列)；Perform linear regression on the estimated yield difference 14 (3rd column of Table 15) and the EVI anomaly of the 15th cycle (4th column of Table 15) to obtain the regression coefficients of the second estimated yield model a=9.7529, b=0.5682, that is, the estimated yield Fluctuation model 15=9.7529×EVI_5-25 +0.5682, from which the estimated output fluctuation value 15 (5th column of Table 15) can be obtained; EVI_5-25 +0.5682, from which the model estimated output value of 15 can be obtained (column 6 of Table 15);

由实际产量(表15第1列)和模型估产值15(表15第6列)求差得到估产差值15(表15第7列)，进而由估产差值15(表15第7列)与实际产量(表15第1列)求绝对比值得到模型误差15(表15第8列)。本阶段模型平均误差为2％。Calculate the difference between the actual output (column 1 of Table 15) and the model estimated output value 15 (column 6 of Table 15) to obtain the estimated yield difference 15 (column 7 of Table 15), and then calculate the estimated yield difference 15 (column 7 of Table 15) The absolute ratio of the actual yield (column 1 of Table 15) yields the model error of 15 (column 8 of Table 15). The average error of the model at this stage is 2%.

综上，第十五次估产模型＝估产模型15×(1±2％)＝(模型估产值14+9.7529×EVI_5-25+0.5682)×(1±2％)。To sum up, the fifteenth production estimation model=estimated production model 15×(1±2%)=(estimated production value of the model 14+9.7529×EVI_5-25 +0.5682)×(1±2%).

表15Table 15

(16)第十六次估产模型构建(16) Construction of the sixteenth production estimation model

计算2001年至2015年实际产量(表16第1列)和模型估产值15(表16第2列)的估产差值15(表16第3列)；Calculate the estimated yield difference 15 (column 3 of Table 16) between actual production (column 1 of Table 16) and model estimated output 15 (column 2 of Table 16) from 2001 to 2015;

估产差值15(表16第3列)和第16个周期的EVI距平(表16第4列)进行线性回归得到第二次估产模型的回归系数a＝9.69，b＝0.5265，即得估产波动模型16＝9.69×EVI_6-10+0.5265，由此可得估产波动值16(表16第5列)；估产模型16＝模型估产值15+估产波动模型16＝模型估产值15+9.69×EVI_6-10+0.5265，由此可得模型估产值16(表16第6列)；Perform linear regression on the estimated yield difference 15 (the third column of Table 16) and the EVI anomaly of the 16th cycle (the fourth column of Table 16) to obtain the regression coefficients of the second estimated yield model a=9.69, b=0.5265, that is, the estimated yield Fluctuation model 16=9.69×EVI_6-10 +0.5265, from which the estimated output fluctuation value 16 can be obtained (the fifth column of Table 16); EVI_6-10 +0.5265, from which the estimated output value of the model 16 (column 6 of Table 16) can be obtained;

由实际产量(表16第1列)和模型估产值16(表16第6列)求差得到估产差值16(表16第7列)，进而由估产差值16(表16第7列)与实际产量(表16第1列)求绝对比值得到模型误差16(表16第8列)。本阶段模型平均误差为2％。Calculate the difference between the actual output (column 1 of Table 16) and the model estimated output value 16 (column 6 of Table 16) to obtain the estimated yield difference 16 (column 7 of Table 16), and then calculate the estimated yield difference 16 (column 7 of Table 16) The absolute ratio to the actual yield (column 1 of Table 16) yields the model error 16 (column 8 of Table 16). The average error of the model at this stage is 2%.

综上，第十六次估产模型＝估产模型16×(1±2％)＝(模型估产值16+9.69×EVI_6-10+0.5265)×(1±2％)。To sum up, the 16th production estimation model=estimated production model 16×(1±2%)=(model estimated production value 16+9.69×EVI_6-10 +0.5265)×(1±2%).

表16Table 16

(17)第十七次估产模型构建(17) Construction of the seventeenth production estimation model

计算2001年至2015年实际产量(表17第1列)和模型估产值16(表17第2列)的估产差值16(表17第3列)；Calculate the difference between actual production (column 1 of table 17) and model estimated production value16 (column 2 of table 17) from 2001 to 2015, the estimated yield difference16 (column 3 of table 17);

估产差值16(表17第3列)和第17个周期的EVI距平(表17第4列)进行线性回归得到第二次估产模型的回归系数a＝9.6533，b＝0.3035，即得估产波动模型16＝9.6533×EVI_6-26+0.3035，由此可得估产波动值17(表17第5列)；估产模型17＝模型估产值16+估产波动模型17＝模型估产值16+9.6533×EVI_6-26+0.3035，由此可得模型估产值17(表17第6列)；Linear regression is performed between the estimated yield difference 16 (the third column of Table 17) and the EVI anomaly of the 17th cycle (the fourth column of Table 17) to obtain the regression coefficients of the second estimated yield model a=9.6533, b=0.3035, that is, the estimated yield Fluctuation model 16=9.6533×EVI_6-26 +0.3035, from which the estimated output fluctuation value 17 can be obtained (the fifth column of table 17); EVI_6-26 +0.3035, from which the model estimated output value 17 (column 6 of Table 17) can be obtained;

由实际产量(表17第1列)和模型估产值17(表17第6列)求差得到估产差值17(表17第7列)，进而由估产差值17(表17第7列)与实际产量(表17第1列)求绝对比值得到模型误差17(表17第8列)。本阶段模型平均误差为2％。Calculate the difference between the actual output (column 1 of Table 17) and the model estimated output value 17 (column 6 of Table 17) to obtain the estimated yield difference 17 (column 7 of Table 17), and then calculate the estimated yield difference 17 (column 7 of Table 17) The absolute ratio of the actual yield (column 1 of Table 17) yields the model error 17 (column 8 of Table 17). The average error of the model at this stage is 2%.

综上，第十七次估产模型＝估产模型17×(1±2％)＝(模型估产值16+9.6533×EVI_6-26+0.3035)×(1±2％)。To sum up, the 17th production estimation model=estimated production model 17×(1±2%)=(model estimated production value 16+9.6533×EVI_6-26 +0.3035)×(1±2%).

表17Table 17

3)目标年份的估产：基于上述历史数据构建的周期性估产模型，进行目标年份的周期性实时估产(估产周期必须与所建立的估产模型周期一致，否则重新构建模型)。3) Production estimation in the target year: Based on the periodic production estimation model constructed based on the above historical data, perform periodic real-time production estimation in the target year (the production estimation period must be consistent with the established production estimation model period, otherwise the model will be rebuilt).

本发明展示中的目标估产为2016年河北邢台的冬小麦实时估产。The target estimated yield in the present invention is the real-time estimated yield of winter wheat in Xingtai, Hebei in 2016.

由步骤101可得2016年河北邢台的冬小麦趋势产量为5379.3494kg/ha。因此，采用上述2)构建的估产模型逐步估产结果如表18所示,2016年统计公布产量为5385.184kg/ha,在本案最终估产结果5288.571±105.771kg/ha范围内，证明本技术方案具有可行性。Fromstep 101, the trend yield of winter wheat in Xingtai, Hebei in 2016 is 5379.3494kg/ha. Therefore, the step-by-step production estimation results using the production estimation model constructed in the above 2) are shown in Table 18. In 2016, the output was 5385.184kg/ha. The final production estimation result in this case was within the range of 5288.571±105.771kg/ha, which proves that this technical solution is feasible. sex.

表18Table 18

本发明的主要优点是：The main advantages of the present invention are:

(1)能够实现作物生长过程中的准实时估产。(1) It can realize quasi-real-time yield estimation in the process of crop growth.

(2)以环境影响参数构建的模型体现了短期环境要素变化对产量的影响，具有一定的机理。(2) The model constructed with environmental impact parameters reflects the impact of short-term environmental factor changes on yield, and has a certain mechanism.

(3)灵活性高。在实施过程中可以灵活选择环境影响参数构建波动模型，即可根据不同估产区域、估产对象数据的获取情况、模型的精度要求、业务生产的周期要求等，灵活选择单因子或多因子组合参数，构建简单模型或复杂模型实施估产。(3) High flexibility. In the implementation process, the environmental impact parameters can be flexibly selected to construct a fluctuation model, and the single-factor or multi-factor combination parameters can be flexibly selected according to different production estimation areas, the acquisition of production target data, the accuracy requirements of the model, and the cycle requirements of business production, etc. Build simple or complex models to estimate production.

(3)具有业务化实现的可行性。(3) It has the feasibility of business realization.

本发明还提供了一种粮食作物实时估产系统，如图2所示，该系统包括：The present invention also provides a food crop real-time yield estimation system, as shown in Figure 2, the system includes:

数据获取模块201，用于获取历史数据，历史数据包括不同年份不同时期的增强型植被指数、不同年份粮食作物的统计产量和不同年份粮食作物的趋势产量；Thedata acquisition module 201 is used for acquiring historical data, the historical data includes the enhanced vegetation index in different years and different periods, the statistical yield of grain crops in different years and the trend yield of grain crops in different years;

距平计算模块202，用于根据不同年份不同时期的增强型植被指数计算增强型植被指数的距平；Theanomaly calculation module 202 is used for calculating the anomaly of the enhanced vegetation index according to the enhanced vegetation index of different years and different periods;

模型构建模块203，用于以第i时期增强型植被指数的距平为自变量，以估产差值为因变量，构建线性回归方程，得到第i时期的估产模型，所述估产差值为统计产量与趋势产量的差值或统计产量与第i-1时期的估产值的差值，其中，i＝1,…,n；Themodel building module 203 is used for taking the anomaly of the enhanced vegetation index in the i-th period as an independent variable, and using the estimated yield difference as a dependent variable to construct a linear regression equation to obtain an estimated yield model in the i-th period, and the estimated yield difference is a statistical value. The difference between the output and the trend output or the difference between the statistical output and the estimated output value of the i-1 period, where i=1,...,n;

波动值计算模块204，用于将第i时期的增强型植被指数的距平代入第i时期的估产模型，得到第i时期的估产波动值；The fluctuationvalue calculation module 204 is used for substituting the anomaly of the enhanced vegetation index of the ith period into the estimated yield model of the ith period to obtain the estimated yield fluctuation value of the ith period;

模型估产值计算模块205，用于根据第i时期的估产波动值和第i-1时期估产值计算第i时期的估产值；The model estimated outputvalue calculation module 205 is used to calculate the estimated output value of the ith period according to the estimated output fluctuation value of the ith period and the estimated output value of the i-1th period;

模型误差计算模块206，用于计算第i时期的估产模型的误差；The modelerror calculation module 206 is used to calculate the error of the estimated production model in the i-th period;

估产值计算模块207，用于采用第i时期的估产模型的误差对第i时期的模型估产值进行修正，得到第i时期的估产值。The estimated outputvalue calculation module 207 is configured to correct the model estimated output value of the ith period by using the error of the estimated output model of the ith period to obtain the estimated output value of the ith period.

其中，模型构建模块203具体包括：Wherein, themodel building module 203 specifically includes:

波动值计算模块204具体包括：The fluctuationvalue calculation module 204 specifically includes:

模型估产值计算模块205具体包括：The model estimated outputvalue calculation module 205 specifically includes:

本发明提供的粮食作物实时估产系统以趋势产量为基数，定期计算由短期环境要素改变而引起的产量波动值，预估作物的产量。而且，本发明采用增强型植被指数作为环境影响因子，能够综合反应气象因子、土壤因子和田间管理因子的变化。进而，使得本发明提供的粮食作物实时估产方法及系统能够实现准确有效、实时性强的粮食作物估产。The grain crop real-time yield estimation system provided by the invention takes the trend yield as the base, calculates the yield fluctuation value caused by the change of short-term environmental elements on a regular basis, and estimates the crop yield. Moreover, the present invention adopts the enhanced vegetation index as the environmental influence factor, which can comprehensively reflect the changes of meteorological factors, soil factors and field management factors. Furthermore, the real-time grain crop yield estimation method and system provided by the present invention can realize accurate, effective and real-time grain crop yield estimation.

本说明书中各个实施例采用递进的方式描述，每个实施例重点说明的都是与其他实施例的不同之处，各个实施例之间相同相似部分互相参见即可。对于实施例公开的系统而言，由于其与实施例公开的方法相对应，所以描述的比较简单，相关之处参见方法部分说明即可。The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same and similar parts between the various embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant part can be referred to the description of the method.

本文中应用了具体个例对本发明的原理及实施方式进行了阐述，以上实施例的说明只是用于帮助理解本发明的方法及其核心思想；同时，对于本领域的一般技术人员，依据本发明的思想，在具体实施方式及应用范围上均会有改变之处。综上所述，本说明书内容不应理解为对本发明的限制。In this paper, specific examples are used to illustrate the principles and implementations of the present invention. The descriptions of the above embodiments are only used to help understand the methods and core ideas of the present invention; meanwhile, for those skilled in the art, according to the present invention There will be changes in the specific implementation and application scope. In conclusion, the contents of this specification should not be construed as limiting the present invention.