Professional Documents
Culture Documents
20
i
1 HEADING 1
3. Area of Boro cultivation in the block estimated from Small Area Survey.
4. Area of (Non Boro) Rabi cultivation in the block estimated from Small Area Survey.
5. Average (on the blocks) annually replenishable groundwater level in each of the
districts, i.e. level of groundwater that have declined in the district over the annum.
Standard units for Area, groundwater are square meter (m2 ) and cubic meter (m3 )
respectively (ref. http://www.cgwb.gov.in). These standard units are used through-
out this text and in our analysis.
2 Objective
Our objective is to explain the groundwater depletion or variation in groundwater de-
pletion by lean season irrigation. It is important to notice that the data in our hand
is strictly limited to West Bengal only. Therefore the results and conclusions will be
limited to this particular state of India. There is no reason at all to believe that we will
have same results on different parts of the world, because groundwater depletion is surely
affected by geographic characteristics of the part.
Thus we are actually trying to find out any kind of relationship between lean season
irrigation and groundwater depletion and then explain effects of different independent
factors involved.
3 Method of Study
3.1 Assumptions
We make two necessary and relevant assumptions for our study, these are,
Uniform distribution of groundwater within each of the districts Since we have
only the average annually replenishable groundwater level in the districts(but not
for the blocks), the whole information on blockwise annually replenishable ground-
water levels are not available to us. Thus this assumption helps us to analysis our
data.
Lean season irrigation in some area affects only the groundwater in that area
It says that irrigation in other blocks do not affect groundwater level in a block.
This assumption is quite reasonable since block size is generally much bigger than
the total area of cultivation. This assumption is technically needed for assuring the
independence of the observations from each of the blocks.
3 METHOD OF STUDY 2
∂R.G.W/R.G.W ∂ log(R.G.W )
= (1)
∂Area Irri/Area Irri ∂ log(Area Irri)
Where R.G.W denotes annual replenishable ground water in the blocks and AreaIrri
denotes Area of irrigation in the blocks. Thus, we shall try to model log(R.G.W ) as a
linear function of log(Area Irri).
By our first assumption, we can find the amount replenishable groundwater for a block
by multiplying the replenishable groundwater level by total area of the block. Thus the
amount of annually groundwater depletion is available to us for the blocks.
+βAreaIrri(Rabi) log(AreaIrri(Rabi)) +
with normally distributed with mean zero and equal variance. Relevant sources of
error for the model would be 1) formula error, 2) estimation error for estimating Area of
Boro and Rabi irrigation from SA, 3) Absence of other relevant explanatory variables.
NOTE: Our Goodness of Fit of the model would be based on R¯2 (for nested models)
and AIC(for non nested models). Results are summarized as follows
The last result in the trial model is an evidence of the heteroscedasticity of the model
residuals. We suspect that the groundwater depletion may not be equally affected by a
unit change in amount of irrigation(in Area) for different levels of Irrigation.
Modification 1 We assume V (|T otalIrrigation) = σ 2 (T otalIrri)−2
Modification 2 Two more blocking factors along with their interactions are assumed.
They are 1) Whether proportion of Area of Boro in a block CB or not, 2) Whether
proportion of Area of Rabi in a block CR or not. On the basis of AIC comparison
the choice of CB and CR turns out to be CB = CR = 0.3. These factorial effects
can be graphically seen in Figure 2.
Figure 3 shows some outlining points on the data. We exclude those points to do our
analysis.
Figure 4 gives diagrammatical view of the model residuals. Plots support our as-
sumptions on the errors in the model.
5.2 Residuals
Min 1Q Median 3Q Max
-90.0824 -24.115 -0.8836 24.4081 87.6247
6 CONCLUSION 5
5.3 Coefficients
Factor/Regressor Estimate Std. Error t value Pr(> |t|) Signif
(Intercept) 5.87511 0.24841 23.651 < 2e-16 ***
(side.of.Bhagirathi)1 0.24812 0.0533 4.655 4.79e-06 ***
(side.of.Bhagirathi)2 0.28652 0.06116 4.685 4.18e-06 ***
log(Area.of.Boro.from.SA) 0.1552 0.02639 5.881 1.04e-08 ***
log(Area.of.Rabi.from.SA) 0.27971 0.02486 11.251 < 2e-16 ***
(Prop.Boro > 0.3)TRUE -0.68454 0.06905 -9.913 < 2e-16 ***
(Prop.Rabi > 0.3)TRUE -0.62339 0.07499 -8.313 2.85e-15 ***
(Prop.Boro > 0.3)TRUE : 0.65773 0.09557 6.882 3.21e-11 ***
(Prop.Rabi > 0.3)TRUE
Signif. codes: 0 *** 0.001 ** 0.01 * 0.05 . 0.1 1
6 Conclusion
• The study shows positive effect of relative change in Boro or Rabi irrigation on
relative change in replenishable groundwater. For 1 unit relative change in Boro
irrigation there is a15.52% relative change in replenishable groundwater, while for
Rabi thos value is 27.97%.
• The areas on the east or west of Bhagirathi are more affected in terms of relative
change in replenishable groundwater than that in areas in north of Bhagirathi. This
difference is 24.8% for the east and 28.6% for the west.
• Blocks with more irrigation of both Boro and Rabi are affected more, in terms of
relative change in groundwater depletion, than others (nearly 68% more).
7 Further
The Model only explains 57.41% of the variation. There are necessarily other factors
or independent variables affecting the relative change in R.G.W. With more information
on irrigation in the blocks and detailed information on falling trend in rainy season
(which has been ignored in our study due to large amount of missing data) more specific
conclusions may be reached.