基于PSO的DSSAT水稻品种参数优化

农业技术转移决策支持系统（DSSAT）在农业领域的应用越来越广泛，应用DSSAT的首要工作就是估计作物品种参数。GLUE参数估计器是DSSAT自带的参数估计工具，但GLUE参数估计器所估计的品种参数并不总有效，其估计参数的DSSAT模拟精度往往不高。本文利用4个品种水稻的田间实测产量数据，采用对比分析方法，以DSSAT自带的GLUE参数估计器运行结果为参照，将粒子群优化（PSO）的每个粒子视为一组水稻品种参数，在运行PSO算法过程中调用DSSAT模拟水稻产量，依据产量模拟误差和PSO的运行机制修改粒子，从而验证PSO优化DSSAT水稻品种参数的有效性及可行性。研究结果表明：两种算法均能较好识别DSSAT水稻品种参数，但GLUE参数估计器估计参数无效的频次较高；与GLUE参数估计器相比，PSO识别的参数均为有效参数，其优化参数的DSSAT模拟水稻产量的精度更高，标准化均方根误差（NRMSE）处于5.98%~8.78%之间，明显低于GLUE参数估计器的6.89%~18.06%，所模拟的水稻产量也更接近于实测产量。     

基于双参数和粒子滤波同化算法的夏玉米单产估测

为了提高河北省中部平原夏玉米的估产精度和进一步验证粒子滤波同化算法对农业作物估产的适用性,采用粒子滤波算法同化CERES-Maize模型模拟和MODIS数据反演的叶面积指数(Leaf area index,LAI)、条件植被温度指数(Vegetation temperature condition index,VTCI),应用随机森林回归算法确定夏玉米不同生育时期LAI和VTCI的权重,构建单产估测模型。结果表明,无论是单点尺度还是区域尺度,同化的LAI和VTCI均能较好地响应外部观测数据,同化LAI可减缓CERES-Maize模型模拟LAI的剧烈变化;同化VTCI结合模型模拟和遥感观测,更能反映夏玉米对水分胁迫的敏感性。利用2015年河北省中部平原各县(区)夏玉米产量对较优估产模型进行精度验证,结果表明,同化前后夏玉米产量模拟结果与统计产量间的归一化均方根误差由12.71%下降到10.50%,平均相对误差由12.57%下降到8.43%,说明基于同化LAI和VTCI构建的双参数单产估产模型可用于区域夏玉米单产估测。     

基于PEST的RZWQM2模型参数优化与验证

根据糯玉米-冬小麦田间喷灌试验不同处理结果,利用独立的自动参数估计软件PEST对RZWQM2模型进行参数优化,并分析了24个模型参数的综合敏感度。通过控制不同观测变量(土壤含水率、土壤氮素含量、作物叶面积指数、产量)模拟差异函数值在目标函数中的比重,优化目标方程,确定模型参数,并用田间试验数据对模型进行验证。结果表明,在不同观测变量的模拟差异函数值最接近条件下,冬小麦出叶间隔特性参数、冬小麦春化作用敏感特性参数及糯玉米出叶间隔特性参数等3个参数对模型整体模拟效果影响最大。相比试错法而言,基于PEST优化的RZWQM2模型能够更准确地模拟糯玉米-冬小麦轮作系统中水分、氮素及作物生长情况。     

Estimating cotton evapotranspiration crop coefficients with a multispectral vegetation index

Crop coefficients are a widely used and universally accepted method for estimating the crop evapotranspiration (ET_c) component in irrigation scheduling programs. However, uncertainties of generalized basal crop coefficient (K_cb) curves can contribute to ET_c estimates that are substantially different from actual ET_c. Limited research with corn has shown improvements to irrigation scheduling due to better water-use estimation and more appropriate timing of irrigations when K_cb estimates derived from remotely sensed multispectral vegetation indices (VIs) were incorporated into irrigation-scheduling algorithms. The purpose of this article was to develop and evaluate a K_cb estimation model based on observations of the normalized difference vegetation index (NDVI) for a full-season cotton grown in the desert southwestern USA. The K_cb data used in developing the relationship with NDVI were derived from back-calculations of the FAO-56 dual crop coefficient procedures using field data obtained during two cotton experiments conducted during 1990 and 1991 at a site in central Arizona. The estimation model consisted of two regression relations: a linear function of K_cb versus NDVI (r²=0.97, n=68) used to estimate K_cb from early vegetative growth to effective full cover, and a multiple regression of K_cb as a function of NDVI and cumulative growing-degree-days (GDD) (r²=0.82, n=64) used to estimate K_cb after effective full cover was attained. The NDVI for cotton at effective full cover was ~0.80; this value was used to mark the point at which the model transferred from the linear to the multiple regression function. An initial evaluation of the performance of the model was made by incorporating K_cb estimates, based on NDVI measurements and the developed regression functions, within the FAO-56 dual procedures and comparing the estimated ET_c with field observations from two cotton plots collected during an experiment in central Arizona in 1998. Preliminary results indicate that the ET_c based on the NDVI-K_cb model provided close estimates of actual ET_c.Communicated by R. Evans     

Evaluation of the RZWQM-CERES-Maize hybrid model for maize production

The root zone water quality model (RZWQM) was developed primarily for water quality research with a generic plant growth module primarily serving as a sink for plant nitrogen and water uptake. In this study, we coupled the CERES-Maize Version 3.5 crop growth model with RZWQM to provide RZWQM users with the option for selecting a more comprehensive plant growth model. In the hybrid model, RZWQM supplied CERES with daily soil water and nitrogen contents, soil temperature, and potential evapotranspiration, in addition to daily weather data. CERES-Maize supplied RZWQM with daily water and nitrogen uptake, and other plant growth variables (e.g., root distribution and leaf area index). The RZWQM-CERES hybrid model was evaluated with two well-documented experimental datasets distributed with DSSAT (Decision Support System for Agrotechnology Transfer) Version 3.5, which had various nitrogen and irrigation treatments. Simulation results were compared to the original DSSAT-CERES-Maize model. Both models used the same plant cultivar coefficients and the same soil parameters as distributed with DSSAT Version 3.5. The hybrid model provided similar maize prediction in terms of yield, biomass and leaf area index, as the DSSAT-CERES model when the same soil and crop parameters were used. No overall differences were found between the two models based on the paired t test, suggesting successful coupling of the two models. The hybrid model offers RZWQM users access to a rigorous new plant growth model and provides CERES-Maize users with a tool to address soil and water quality issues under different cropping systems.     

高斯-牛顿法及其在作物水分生产函数模型参数求解中的应用

对于作物水分生产函数的非线性模型,现有参数求解方法是先采用变量代换使模型线性化后再用最小二乘法求解。用这种方法得到的参数估计值并不是无偏估计,因而回归方程也不是最优的,相应相关系数也不能很好地反映拟合的显著性水平。介绍了高斯-牛顿法,并把它用于作物水分生产函数Jensen模型参数求解,该方法可使拟合结果逼近无偏估计,从而提高拟合的精度。同时,给出的未经线性化模型的相关系数更能反映拟合的显著性水平。     

融合无人机光谱信息与纹理特征的棉花叶面积指数估测

农田尺度下作物叶面积指数(Leaf area index, LAI)的精准监测,对于研究群体结构对产量和管理措施的响应具有重要意义。目前普遍采用无人机光谱特征反演作物的LAI指数,作为长势和冠层结构诊断的重要依据,其估测精度的准确性是否可以提高仍有待研究。作物表面特征,如灰度和颜色,在不同生育阶段会发生变化。为此,本研究考虑到LAI的影响因素,设置不同的种植密度和氮素水平营造差异化的冠层结构,利用搭载多光谱传感器的无人机获取主要生育时期棉花的冠层图像得到植被指数(Vegetation indexs, VIs),基于二阶概率统计滤波(Co-occurrence measures)方法获取均值(MEA)、方差(VAR)、协同性(HOM)、对比度(CON)、相异性(DIS)、信息熵(ENT)、二阶矩(SEM)和相关性(COR)等8个纹理特征值(Texture features, TFs)。最后,采用支持向量机回归(SVR)、偏最小二乘法(PLSR)、深度神经网络(DNN)分别建立基于光谱特征、纹理特征以及二者结合的棉花LAI的估算模型,并比较差异。试验结果表明：VI₍(...     

基于SPOT遥感数据的甘蔗叶面积指数反演和产量估算

利用SPOT遥感数据进行甘蔗叶面积指数LAI反演,建立最佳NDVI-LAI反演模型,同时结合不同生育期甘蔗叶面积指数的时序变化规律,建立各生育期甘蔗叶面积指数LAI与产量的相关关系,得到甘蔗叶面积指数LAI 产量最佳估产模型.在验证甘蔗叶面积指数LAI的基础上,利用遥感反演的甘蔗叶面积指数LAI进行甘蔗单产估算.结果表明:甘蔗叶面积指数LAI与NDVI之间存在显著的正相关关系,全生育期二者的相关性最高,以二次函数模型拟合效果最佳,决定系数R2为0.8429.将遥感数据反演得到的平均叶面积指数LAI数据代入甘蔗叶面积LAI-产量模型得到全县平均单产,与统计数据相比,相对误差仅为2.6％.说明该模型具有较好的估产效果,可以为甘蔗区域估产提供重要参考.     

耦合模拟退火优化最小二乘支持向量机的日参照蒸散量模拟计算

针对传统最小二乘支持向量机模型的训练速度慢、不易在线训练、计算量大及参数选择困难等缺陷,提出采用耦合模拟退火优化最小二乘支持向量机算法,建立基于耦合模拟退火优化最小二乘支持向量机的参照作物蒸散量预测模型。选取陕西省的榆林、安康和西安气象站监测的1971-2014年气象资料进行模型训练、测试与验证,研究气象监测获取原始数据作为网络输入,参照蒸散量ET0为输出,构建CSA-LSSVM预测模型,并将CSA-LS-SVM预测结果与LSSVM模型及经典ET0模型模拟计算结果进行比较。结果表明,CSA-LS-SVM模型模拟计算精度和总ET0模拟模型都优于LSSVM模型及其他经典模型模拟结果。该研究CSA-LS-SVM模型为陕西地区气象资料缺乏情况下ET0精确计算提供科学依据,为作物需水量的智能决策提供参考。     

基于MRE-PointNet+AE的绿萝叶片外形参数估测算法

为了准确、高效、自动获取植物叶片外形参数,提出一种基于多分辨率编码点云深度学习网络(MRE-PointNet)和自编码器模型的绿萝叶片外形参数估测算法。使用Kinect V2相机以垂直姿态获取绿萝叶片点云数据,采用直通滤波、分割、点云精简算法对数据进行预处理,通过测定的叶片外形参数反演绿萝叶片几何模型,并计算几何模型的叶长、叶宽、叶面积。将不同参数组合构建的几何模型离散成点云数据输入MRE-PointNet网络,得到几何模型叶片外形参数估测的预训练模型。针对拍摄过程中存在的叶片部分遮挡和噪声问题,采用自编码器网络对点云数据进行二次处理,以几何模型离散的点云数据作为输入,经过编码解码运算得到自编码器的预训练模型,提升了MRE-PointNet网络在遮挡情况下对叶片外形参数估测的鲁棒性。试验共采集300片绿萝叶片点云数据,按照2∶1比例进行划分,以其中200片点云数据作为训练集,对预训练模型MRE-PointNet做模型迁移的参数微调,以剩下的100片点云数据作为测试集,评估模型对绿萝叶片外形参数的估测能力。采用本文算法将外形参数估测值和真实值进行数学统计与线性回归分析,得出叶长、叶宽和叶面积估测的R^2和RMSE分别为0.9005和0.4170 cm、0.9131和0.3164 cm、0.9447和3.8834 cm^2。试验表明,基于MRE-PointNet和自编码器模型的绿萝叶片外形参数估测算法具有较高的精确度和实用性。     

黔中地区玉米需水量的计算与灌溉预报

作物需水量的计算和灌溉预报是农田水分管理的主要参数,实时灌溉预报是制定动态灌溉用水计划的基础,对区域节水,增加作物产量和提高经济效益起着重要作用。以黔中地区全生育期玉米为研究对象,结合玉米不同生育阶段的各项参数,进行作物需水量的计算和灌溉预报。     

The accuracy of evapotranspiration estimated with the FAO modified penman equation

Summary The FAO modified Penman equation has gained acceptance as a standard method of estimating reference crop evapotranspiration. Although theoretically sound the Penman equation becomes increasingly empirical when parameters or variables have to be estimated. When evapotranspiration estimates are being used for practical purposes the uncertainties introduced by these empirical factors and relationships should not be neglected. Evapotranspiration estimates for north-east Sri Lanka are used to illustrate the importance of the empiricisms in the FAO modified Penman equation. It is shown that the different empirical relationships used to estimate net radiation and the wind function in the FAO modified Penman equation and in the Penman (1963) equation produce a 23% difference in the estimate of annual reference crop evapotranspiration.     

基于蒸渗仪的冬小麦-夏玉米ET估算模型特征参数研究

为快速准确估算农田蒸散量,利用24个群集式蒸渗仪,在国家节水灌溉北京工程技术研究中心大兴节水灌溉试验站进行了两年的灌溉试验,获得冬小麦-夏玉米生育期的日内冠气温差和实际日蒸散量(ET_a)等数据,对不同水分处理下的S-I蒸散量估算模型进行率定及验证,并分析模型特征参数a、b的变化规律及两者的差异。结果表明:冬小麦的S-I模型特征参数a在日间随时间变化先增大、后减小,在严重水分胁迫处理时a为负值、且数值较小,其余灌溉处理时参数a由正值逐渐变化至负值;不同灌水处理b均为负值,充分灌溉处理时b在日间随时间变化逐渐增大,严重水分胁迫处理时b相对较大,日间变化趋势不稳定。水分胁迫对夏玉米模型参数的影响程度低于冬小麦,特征参数a均为正值,参数b均为负值,且随时间变化逐渐增大;水分胁迫处理时b变化范围明显小于其他两个处理,干旱处理特征参数日间变化较大。冬小麦与夏玉米不同处理之间模型参数a、b变化差异较大,但冠层温度和空气温度差T_c-T_a与日蒸散量和日净辐射量差ET_d-Rn_d间拟合精度都在13:00时最高,此时充分灌溉冬小麦和夏玉米的模型参数a、b分别为1.082、-1.127和1.588、-1.363。利用率定的S-I模型计算冬小麦和夏玉米主要生育期ET_d与实测ET_a之间的决定系数R~2均在0.7以上,均方根误差RMSE均小于0.89 mm/d,一致性系数d均在0.9以上。尤其是充分灌溉处理的数据间R~2和d均较高,RMSE小于其他处理,说明水分胁迫影响模型的估算精度,S-I模型能够更准确地估算水分胁迫较少农田的蒸散量。     

江苏地区不同参考作物蒸发蒸腾量估算模型

为了研究不同参考作物蒸发蒸腾量ET₀估算方法在江苏地区的适用性,收集了江苏省徐州市、高邮市和昆山市1957年1月至2019年12月的气象数据,采用12种不同模型估算了各站点的ET₀,其中模型Priestly-Taylor,Hansen,Jensen-Haise,Makkink是基于辐射数据的模型;MC-Cloud,1985 Hargreaves,Thornthwaite是基于温度数据的;Copais,Valiantzas 1和Valiantzas 2是综合法模型;XGBoost和SVM是机器学习模型.12种ET₀的估算模型计算值分别与Penman-Monteith模型(PM)计算值进行比较,结果表明:各站点的综合评价指数GPI最高的为机器学习模型中的SVM模型;在输入参数相同的情况下,机器学习模型模拟精度优于综合法和温度法以及辐射法中的Pristley-Taylor和Makkink模型;机器学习模型随着输入参数减少,模拟精度依次降低.研究结果可以为江苏地区气象数据不完善时估算ET₀提供科学依据.     

Optimization model of an irrigation reservoir for water allocation and crop planning under various weather conditions

This paper develops a non-linear programming optimization model with an integrated soil water balance, to determine the optimal reservoir release policies, the irrigation allocation to multiple crops and the optimal cropping pattern in irrigated agriculture. Decision variables are the cultivated area and the water allocated to each crop. The objective function of the model maximizes the total farm income, which is based on crop–water production functions, production cost and crop prices. The proposed model is solved using the simulated annealing (SA) global optimization stochastic search algorithm in combination with the stochastic gradient descent algorithm. The rainfall, evapotranspiration and inflow are considered to be stochastic and the model is run for expected values of the above parameters corresponding to different probability of exceedence. By combining various probability levels of rainfall, evapotranspiration and inflow, four weather conditions are distinguished. The model takes into account an irrigation time interval in each growth stage and gives the optimal distribution of area, the water to each crop and the total farm income. The outputs of this model were compared with the results obtained from the model in which the only decision variables are cultivated areas. The model was applied on data from a planned reservoir on the Havrias River in Northern Greece, is sufficiently general and has great potential to be applicable as a decision support tool for cropping patterns of an irrigated area and irrigation scheduling.     

气候变化条件下旱作玉米用水效率与单产变化趋势分析

以渭北旱塬合阳和长武2个试验站点为研究区域,通过多年的玉米田间试验数据评估CERES-Maize模型的适用性,再利用区域气候模式Reg CM4.0输出的气象数据对2050年前玉米单产及生产水足迹进行预测。结果表明:CERES-Maize模型可以很好地模拟雨养玉米产量和物候期,多数年份二者的绝对相对误差(Absolute relative error,ARE)在10%以内,CERES-Maize模型在渭北旱塬旱作农业区有很好的适用性。应用CERES-Maize模型模拟玉米生产水足迹,较传统水足迹计算方法得到的结果更为精确可靠。在RCP2.6气候情景下,随着温度升高和生育期有效降水量的增加,玉米产量呈上升趋势;在RCP8.5气候情景下,随着温度升高和生育期有效降水的减少,玉米产量呈下降趋势。气温上升幅度过大对玉米单产有明显的负面影响,降水与玉米用水效率呈正相关。为有效应对气候变化对旱作作物产量造成的负面影响,应采取减少温室气体排放量、增强土壤蓄水保墒能力、发展集雨补灌、筛选和培育节水抗旱新品种等措施。     

Evaluation of the CERES-Maize water and nitrogen balances under tile-drained conditions

The CERES-Maize model was developed to investigate how variations in environmental conditions, management decisions, and genetics interact to affect crop development and growth. A tile drainage subroutine was incorporated into CERES-Maize to improve soil-water and nitrogen leaching under subsurface tile drainage conditions. The purpose of this work was to evaluate the soil-water, soil-nitrogen, tile drainage, and tile-nitrogen loss routines of CERES-Maize for tile-drained fields in Iowa. An analysis was conducted based on information collected from a study of 36 plots consisting of five management systems during a 4-year period from 1993 to 1996, at Nashua, IA. The model was calibrated for each plot using data from 1994 and 1995, and validated using data from 1993 and 1996. Temporal soil-water contents and water flow from tile drains were calibrated to an average root mean square error (RMSE) of 0.036 cm³ cm⁻³ and 2.62 cm, respectively, compared to measured values. Validation trials gave an average RMSE for soil-water and tile drainage of 0.046 cm³ cm⁻³ and 5.3 cm, respectively. Soil-nitrate and tile-nitrogen flows were calibrated, with an RMSE of 6.27 μg NO₃ g⁻¹ soil⁻¹ and 3.21 kg N ha⁻¹ soil⁻¹, respectively. For the validation trials, the RMSE for soil-nitrate content and cumulative tile-nitrate flow was 6.82 μg NO₃ g⁻¹ soil⁻¹ and 8.8 kg N ha⁻¹, respectively. These results indicate that the new tile drainage algorithms describe water and nitrate movement reasonably well, which will improve the performance of CERES-Maize for artificially drained fields.     

基于EFAST的CROPGRO-Tomato模型参数全局敏感性分析

为了定量讨论番茄生长模型(DSSAT-CROPGRO-Tomato)中各参数对模拟输出结果的影响,运用扩展傅里叶幅度检验(EFAST)法,对影响番茄物候期、生长及生产等3类模型输出的各参数敏感性进行了研究,重点探讨了模型输入中品种参数、气象及土壤参数变化对模型输出结果的影响。结果表明:影响番茄开花期、坐果期和成熟期等物候期最敏感的品种参数和气象参数分别为初花期积温、日最低温度,而土壤参数对番茄物候期的影响可以忽略。影响番茄叶面积指数和冠层高度最敏感的品种参数为初花期积温,而影响番茄根、茎生长最敏感的品种参数为干物质分配比例,日最高温度和土壤田间持水率分别是影响番茄根、茎、叶生长最敏感的气象和土壤参数。影响番茄干物质量和总产量最敏感的品种参数为初花期积温,而影响番茄收获数量和单果质量的品种参数为最大单籽粒质量,日最高温度和土壤田间持水率也分别是影响番茄生产最敏感的气象和土壤参数。     

基于双参数和Morlet多时间尺度特性的冬小麦单产估测

为进一步研究冬小麦在不同时间尺度下长势及产量变化情况,以陕西省关中平原为研究区域,选择与作物长势密切相关的条件植被温度指数（VTCI）和叶面积指数（LAI）作为研究指数,Morlet小波作为函数,利用小波变换和交叉小波变换分别分析不同时间尺度下冬小麦各生育时期VTCI和LAI与单产时间序列间的主振荡周期和共振周期。通过计算小波互相关度,确定各生育时期VTCI和LAI的权重,从而分别构建基于加权VTCI、加权LAI的单参数和双参数估产模型。结果表明,不同生育时期VTCI和LAI与单产间存在不同的主振荡周期和共振周期;通过小波变换构建的基于加权VTCI、加权LAI单产估测模型的归一化均方根误差（NRMSE）分别为16.88%、13.58%,决定系数（R2）分别为0.259、0.520,基于双参数的估产模型NRMSE为13.52%, R2为0.531,表明基于双参数估产模型精度更高。通过交叉小波变换构建的基于加权VTCI、加权LAI单产估测模型的NRMSE分别为16.83%、13.56%,R2分别为0.263、0.522,基于双参数的估产模型NRMSE为13.40%,R2为0.533,表明基于交叉小波构建的估产模型比基于小波变换的估产模型精度均有所提高。利用共振周期构建的双参数估产模型对关中平原2011—2018年冬小麦的单产进行估测,结果显示,产量分布呈现西部高东部低的空间分布特征。     

AquaCrop模型的适用性及应用初探

AquaCrop模型是FAO新推出的以水分为驱动的作物生长模型。为了评价其在华北地区的适用性,于2009-2010年在中国水利水电科学研究院大兴试验站进行了夏玉米水分处理试验,其中2010年的试验数据用于参数率定,2009年的试验教据用于模型验证,并在此基础上对模型参数进行敏感性分析。结果表明,AquaCrop模型能够...