14TSB_ESAP: Tru-Nject: Proximal soil sensing based variable rate application of subsurface fertiliser liquid injection in vegetable/combinable crops

克兰菲尔德的研究小组将负责在选定的实验场进行在线可见光和近红外（vi-NIR）测量。在进行在线土壤测量之前和测量过程中，将采集土壤样本，用于升级克兰菲尔德大学已经开发的一般校准功能。这些升级后的校正函数将通过实验室测量的土壤有机碳（OC）、总氮（TN）、pH、磷（P）、镁（Mg）、钙（Ca）、水分含量（MC）和粘粒含量（CC）来验证其估计精度。这些土壤性质的实验室分析将在克兰菲尔德大学的土壤实验室进行，从实验田收集300个校准和验证土壤样品。克兰菲尔德校准模型的升级将基于化学计量学工具，该工具由主成分分析（PCA）和偏最小二乘回归（PLSR）分析的组合组成，使用Unscrambler 7.8软件（Camo Inc.;奥斯陆，挪威）。克兰菲尔德还将使用ArcGIS ArcMap（ESRI ArcGISTM版本10，CA，USA）开发两种类型的土壤图，例如全点图和比较图。生态地图旨在比较在线预测和实验室测量的土壤特性。全点图将用于理解和评估土壤肥力的田间变化，以及这些变化与作物生长和产量的关系。克兰菲尔德团队还将负责数据融合和地统计分析。数据融合将包括人工神经网络、支持向量机和其他必要的工具。数据融合将使用STATISTICA 10（StatSoft，Inc. USA）软件。将采用非监督分类算法来识别空间相似的类。聚类输出将导入ArcGIS ArcMap（ESRI ArcGISTM版本10，CA，USA），以辅助可视化和空间分析。将通过使用移动窗口过滤器来导出一组同质的管理区域，以最大限度地减少较小集群的出现。随后将绘制试验田的肥力图，每一类的管理区具有类似的产量潜力。在生育力地图绘制完成后，将与STC有限公司合作，使用RB 209（SEFRA）生成N推荐地图，克兰菲尔德也将协助STC有限公司进行成本效益分析，以评估该系统的经济可行性。该分析将基于从试验地块获得的输入数据，比较地下均匀施肥和变量施肥。克兰菲尔德团队将与Manterra一起评估技术集成，以便对系统进行最终评估，并确定是否需要进一步的技术和软件开发。克兰菲尔德还将在开发和传播所取得的成果方面发挥重要作用。他们将领导在精准农业领域的顶级期刊上发表科学论文，如精准农业，生物系统工程，农业计算机和电子，以及顶级国际和国家会议，如欧洲精准农业会议，HGCA精准农业研讨会和EurAgEng 2016。该小组将在谷物等贸易车间推广基于在线地下变量氮施用系统。他们还将为Manterra组织的培训研讨会提供所需的投入。

The research team at Cranfield will be responsible to carry out the on-line visible and near infrared (vi-NIR) measurment inselected experimental fields. Prior and during the on-line soil measurement, soil samples will be collected for upgrading thegeneral calibration functions already developed at Cranfield University. These upgraded calibration functions will bevalidated for accuracy estimation against laboratory measured soil organic carbon (OC), total nitrogenv(TN), pH,phosphorous (P), magnesium (Mg), calcium (ca), moisture content (MC) and clay content (CC). The laboratory analyses ofthese soil properties will be carried out in soil laboratory of Cranfield University for 300 calibration and validation soilsamples collected from the experimental fields. The upgrade of Cranfield calibration models will be based on chemometrictools consisting of a combination of principal component analyses (PCA) and partial least squares regression (PLSR)analysis to be carried out with Unscrambler 7.8 software (Camo Inc.; Oslo, Norway). Cranfield will also develop two typesof soil maps e.g. full-point and comparison maps using ArcGIS ArcMap (ESRI ArcGISTM version 10, CA, USA). Thecomparison maps will aim to compare between on-line predicted and laboratory measured soil properties. The full-pointmaps will be used for understanding and evaluation of within field variation in soil fertility, and how these are correlated withcrop growth and yield. By this, it is hoped to identify crop yield limiting factors.Cranfield team will be also responsible for data fusion and geostatistical analysis. Data fusion will include artificial neuralnetwork, support vector machine and other tools, as necessary. Data fusion will be carried out with STATISTICA 10(StatSoft, Inc. USA) software. An unsupervised classification algorithm will be employed to identify spatially similar classes.The clustering output will be imported into ArcGIS ArcMap (ESRI ArcGISTM version 10, CA, USA) to assist visualisationand spatial analysis. A homogenous set of management zones will be derived by using a moving-window filter to minimisethe occurrence of smaller clusters. This will be followed by the development of fertility maps for the experimental fields, withmanagement zones of each class having similar yield potential. After the fertility maps have been developed, Nrecommendation maps will be generated using RB209 (SEFRA), in collaboration with STC ltd., and agronomist Mr. PhilipEffingham.Cranfield will also assist STC Ltd. in the cost benefit analysis to evaluate the economic viability of the system. This analysiswill be based on input data obtained from the trial plots, comparing between subsurface homogeneous and variable rate Nfertilisation. Together with Manterra, Cranfield team will assess the technology integration, to allow for final evaluation ofthe system, and identify further technical and software development needed if any.Cranfield will also play an important role in the exploitation and dissemination of the results achieved. They will lead onpublishing scientific papers in top journals on precision agriculture e.g. Precision Agriculture, Biosyatems Engineering,Computers and Electronics in Agriculture, and top international and national conferences e.g. European Conference onPrecision Agriculture, HGCA Precision Agriculture Workshop and EurAgEng2016. The team will promote the on-line basedsubsurface variable rate N application system in trade workshops e.g. Cereals. They will also provide the needed input fortraining workshops organised by Manterra.