strategies are important tools for market regulation, food security, and sustainable
economic development.
Data fusion is an alternative to analytical models for economic evaluations too.
For example, oil price change has a lead/lag relationship on food prices, and it can be
used as one of the inputs of data fusion models for the prediction of market price and
decision support for optimal subsidy management. On the other hand, production
units (asset) x yield efficiency x pricing as depending on quality x market conditionbased unit prices are not enough for the optimization of the integrated system. The
main restriction is the balanced use of the natural resources. Water and soil are two
major components of the basin. The water level of the aquifers goes down if the crop
pattern-related irrigation regime is not balanced with respect to precipitation and
other water income of the basin hydrological system. Sustainability management
sets restriction to all four other components within the integrated agricultural
management.
7.3 Regression Model Example for Real-Time Yield
Efficiency Monitoring
The first example is a regression model-based data fusion for yield monitoring. It is
intended to be used for in-season real-time part of the four-step yield monitoring
system. Its accuracy increases due to the reduction of the remaining time to harvest
since observed values replace statistical agrometeorological estimation. Continuous
time crop-specific yield monitoring system consists of four processes:
(a) Phenological stage mapping.
(b) Data segmentation depending on the phenological stage together with correlation analysis.
(c) Yield prediction model depending on phenological stage segmented data.
(d) Model parameters adaptation after harvest by using the geo-statistical data.
The TARBIL system has 440 agrometeorological-phenological monitoring stations in Turkey (Fig. 7.3). Regression model-based yield estimation example here
uses 31 observation records from wheat parcels located in South Eastern Turkey.
They are residing in Mardin, Şanlıurfa, Diyarbakır, and Gaziantep Provinces.
Agrometeorological dataset shown in the Appendix is acquired from wheat parcels
with dry farming conditions. Province yield efficiency data is provided by
TURKSTAT (TÜİK, Turkish Statistics Institute). The total amount of sensors,
including atmospheric, soil, and phenological measurements, is 35 in each of the
monitoring stations. Meteorological sensors’ sampling time interval is 10 minutes.
Camera image capture period is 30 minutes.
Acquired data is processed at a high-performance computer (HPC) at Istanbul
Technical University. Data acquisition and processing operations are supported by
the command and control center responsible for the QoS (quality of service). This
7 Data Fusion in Agricultural Information Systems
109
economic development.
Data fusion is an alternative to analytical models for economic evaluations too.
For example, oil price change has a lead/lag relationship on food prices, and it can be
used as one of the inputs of data fusion models for the prediction of market price and
decision support for optimal subsidy management. On the other hand, production
units (asset) x yield efficiency x pricing as depending on quality x market conditionbased unit prices are not enough for the optimization of the integrated system. The
main restriction is the balanced use of the natural resources. Water and soil are two
major components of the basin. The water level of the aquifers goes down if the crop
pattern-related irrigation regime is not balanced with respect to precipitation and
other water income of the basin hydrological system. Sustainability management
sets restriction to all four other components within the integrated agricultural
management.
7.3 Regression Model Example for Real-Time Yield
Efficiency Monitoring
The first example is a regression model-based data fusion for yield monitoring. It is
intended to be used for in-season real-time part of the four-step yield monitoring
system. Its accuracy increases due to the reduction of the remaining time to harvest
since observed values replace statistical agrometeorological estimation. Continuous
time crop-specific yield monitoring system consists of four processes:
(a) Phenological stage mapping.
(b) Data segmentation depending on the phenological stage together with correlation analysis.
(c) Yield prediction model depending on phenological stage segmented data.
(d) Model parameters adaptation after harvest by using the geo-statistical data.
The TARBIL system has 440 agrometeorological-phenological monitoring stations in Turkey (Fig. 7.3). Regression model-based yield estimation example here
uses 31 observation records from wheat parcels located in South Eastern Turkey.
They are residing in Mardin, Şanlıurfa, Diyarbakır, and Gaziantep Provinces.
Agrometeorological dataset shown in the Appendix is acquired from wheat parcels
with dry farming conditions. Province yield efficiency data is provided by
TURKSTAT (TÜİK, Turkish Statistics Institute). The total amount of sensors,
including atmospheric, soil, and phenological measurements, is 35 in each of the
monitoring stations. Meteorological sensors’ sampling time interval is 10 minutes.
Camera image capture period is 30 minutes.
Acquired data is processed at a high-performance computer (HPC) at Istanbul
Technical University. Data acquisition and processing operations are supported by
the command and control center responsible for the QoS (quality of service). This
7 Data Fusion in Agricultural Information Systems
109