stages. Phenological stage dates are also dependent and predictable (Üstündağ
2017).
Accuracy is the main performance measure for all yield efficiency estimation
methods. Tolerance is directly dependent on accuracy within the chosen statistical
confidence factor (Zc). Total tolerance of yield estimation depends on the summation
of the tolerances in crop area estimation and crop yield efficiency estimation since
their product determines the yield for each region.
Since it is usually not possible to monitor all the harvesting data at a chosen
region, an important aspect is having a reliable method that enables interpolationbased mappings concerning reference data from sampling points. Although the error
rate in the spatial distribution of meteorological parameters can be reduced and
managed by using data fusion methods techniques (Bagis et al. 2012), crop status
monitoring systems may even require the change of input parameter sets besides the
adaption of model parameters.
The soil-adjusted vegetation index (SAVI), modified SAVI (MSAVI), NDVI,
and Global Environmental Monitoring Index (GEMI) are indices in correlation with
surface vegetation. They are computed from monitoring data of remote sensing
satellites or aerial platforms (Herndl 2008). The vapor pressure deficit (VPD),
growing degree days (GDD), photo-thermal unit (PTU), helio-thermal unit (HTU),
reference evapotranspiration (ET0), crop evapotranspiration (ETc), minimum temperature (Tmin), and precipitation (P) are some of the parameters known to have
correlation with plant behavior characteristics. They are computed by using temporal
data of terrestrial monitoring systems (Allen et al. 1998; Amrawat et al. 2013;
Bazgeera et al. 2007). On the other hand, they are spatially interpolatable by using
inverse distance weighting with elevation correction (IDWEC) or kriging methods
since their input variables are basic physical measurements as humidity, air pressure,
air temperature, etc.
The model is initiated by using the statistical agrometeorological data in order to
provide decision support during the planning phase before sowing. Initially, statistical yield prediction (Odoh and Chinedum 2014) is expressed together with the
tolerance as (Eq. 7.1)
YE ¼ μ YE Ç Zc ∙
σ YE ffiffi ffi
n
p
ð7:1Þ
where Zc is the statistical confidence factor and μ and σ are the average and standard
deviation of past yield efficiency within the “n” amount of data sampling at a
specified location or coordinate (x, y). It should be considered that there are different
scales for location, and it differs in the way of construction of the data. If the location
specifies a town or basin, then the past yield efficiency records can be used in the
computation of average or expected trend value (trend-estimated yield). Basins
consist of towns having similar climatic and terrain characteristics within the
geographical neighborhoods in Turkey. Yield efficiency estimation tolerance is at
maximum at the statistical estimation phase.
7 Data Fusion in Agricultural Information Systems
111
2017).
Accuracy is the main performance measure for all yield efficiency estimation
methods. Tolerance is directly dependent on accuracy within the chosen statistical
confidence factor (Zc). Total tolerance of yield estimation depends on the summation
of the tolerances in crop area estimation and crop yield efficiency estimation since
their product determines the yield for each region.
Since it is usually not possible to monitor all the harvesting data at a chosen
region, an important aspect is having a reliable method that enables interpolationbased mappings concerning reference data from sampling points. Although the error
rate in the spatial distribution of meteorological parameters can be reduced and
managed by using data fusion methods techniques (Bagis et al. 2012), crop status
monitoring systems may even require the change of input parameter sets besides the
adaption of model parameters.
The soil-adjusted vegetation index (SAVI), modified SAVI (MSAVI), NDVI,
and Global Environmental Monitoring Index (GEMI) are indices in correlation with
surface vegetation. They are computed from monitoring data of remote sensing
satellites or aerial platforms (Herndl 2008). The vapor pressure deficit (VPD),
growing degree days (GDD), photo-thermal unit (PTU), helio-thermal unit (HTU),
reference evapotranspiration (ET0), crop evapotranspiration (ETc), minimum temperature (Tmin), and precipitation (P) are some of the parameters known to have
correlation with plant behavior characteristics. They are computed by using temporal
data of terrestrial monitoring systems (Allen et al. 1998; Amrawat et al. 2013;
Bazgeera et al. 2007). On the other hand, they are spatially interpolatable by using
inverse distance weighting with elevation correction (IDWEC) or kriging methods
since their input variables are basic physical measurements as humidity, air pressure,
air temperature, etc.
The model is initiated by using the statistical agrometeorological data in order to
provide decision support during the planning phase before sowing. Initially, statistical yield prediction (Odoh and Chinedum 2014) is expressed together with the
tolerance as (Eq. 7.1)
YE ¼ μ YE Ç Zc ∙
σ YE ffiffi ffi
n
p
ð7:1Þ
where Zc is the statistical confidence factor and μ and σ are the average and standard
deviation of past yield efficiency within the “n” amount of data sampling at a
specified location or coordinate (x, y). It should be considered that there are different
scales for location, and it differs in the way of construction of the data. If the location
specifies a town or basin, then the past yield efficiency records can be used in the
computation of average or expected trend value (trend-estimated yield). Basins
consist of towns having similar climatic and terrain characteristics within the
geographical neighborhoods in Turkey. Yield efficiency estimation tolerance is at
maximum at the statistical estimation phase.
7 Data Fusion in Agricultural Information Systems
111