Geoscience Reference
In-Depth Information
5. Statistical significance of the parameter is required for
regression equation.
6. Different weather variables, namely, T
max
, T
min
, relative
humidity (morning and evening) and rainfall having sig-
nificant value are considered to develop the regression
equation for prediction of yield. A regression model is
required to isolate the yield dependent on weather only,
instead of composite weather and agricultural package
fertiliser, irrigation, seed and so on. Thus, a regression
model, though economic in time consumption, might
contain large error percentages in yield prediction. A
case study of a wheat crop yield prediction of 14 districts
of eastern Uttar Pradesh during 2012-2013 at pre-harvest
stage (on 15 March, 2013) has been shown in Table 3.1,
compared with a yield predicted during 2011-2012 on the
same date. This shows the error variability in the range of
0-10% on either side in the model.
Mechani stic
models
These models explain not only the relationship between weather
parameters and yield but also the mechanism of these models
(explains the relationship of influencing dependent variables).
These models are based on physical selection.
Deter ministic
models
These models estimate the exact value of the yield or dependent
variable. Usually, these are developed by mathematical tech-
niques and have well-defined coefficients.
Stochastic
models
For this model, the probability element is attached to each out-
put. For each set of inputs, different outputs are given along
with probabilities. These models define the yield or state of
dependent variable at a given rate.
Dynamic
models
Time is included as a variable in this model. Both dependent
and independent variables have values that remain constant
over a given period of time.
Static models
In a static model, time is included not as a variable. Dependent
and independent variables having values remain constant over
a given period of time.
Descrip tive
model
A descriptive model defines the behaviour of a system in a sim-
ple manner. The model reflects little or none of the mechanisms
that are the causes of the phenomena, but consists of one or
more mathematical equations. An example of such an equa-
tion is the one derived from successively measured weights of
