Stress Modelling

The intervention that we have considered is the NPK(Nitrogen, Phosphorus and Potassium) amount in the soil in the region or Agro-climatic zone considered. Our intervention was to increase the NPK content in all Taluks in the agro-climatic zone to the highest value present in the KAG 2020-21 dataset. The target variable we hope to affect is the Crop Yields for Rice and Wheat.

After the intervention, our model conducts another step called stress modeling. In stress modeling, we consider the influences that logical neighbors have on a particular geographic region and vice-versa. For example, if in a particular geographical region, people have increased the production of a particular crop or started usage of a particular fertilizer, its neighboring and connected regions will develop stress. The region may decide to either imitate its neighbor or regulate itself so as to reach the state of MINIMUM STRESS.

So through our model, we plan to study whether a particular intervention in a region will have a ripple effect on its neighborhood regions and/or influence the changes that the intervention brings to that particular region.

There are two graphs that are mainly considered to analyze the impact created. The first is the bar graph which has both yellow and green bars for each agro-climatic zone. The yellow bar is the value of the target variable i.e., ( rice and wheat production) before any policy. The green bar is just after the intervention.

The second graph has a curve corresponding to each agro-climatic zone. This curve shows how the target variable( rice and wheat production) changes with time due to the effect of stress incurred by its neighbors.

The following link shows our analysis of different agro-climatic zones in Karnataka after intervening in those zones and using stress modeling.


(a): Effect of Fertilizers on Rice in Coastal regions

In Coastal regions, the intervention we have caused has resulted in a negative impact just after the intervention. After the stress modeling, we can see that there is not much difference which implies that the intervention has not caused any change to the target variable under consideration.

This holds true if we consider the situation realistically. As we know coastal regions have an abundant supply of nutrients, hence the usage of fertilizers is redundant which is clearly implied by the constant curve obtained.

(b) Wheat yield is in northern regions:

The second key observation was in wheat production, the northern agroclimatic zones have higher values for wheat production when compared to other zones. This implies that wheat is majorly cultivated in those regions and other regions cultivate wheat in comparatively low amounts which is true.

(c) Spike seen during stress modeling:

The third key observation is regarding the curve emerging during stress modeling of wheat production in the northern zones. We see that in the time steps following the policy intervention, there is an increase in production, till it reaches a peak at time step t3 and then it comes down and gradually settles down at a yield score. This kind of analysis shows that, though a policy may show immediate impact, lateral impact and the percolation of the policy intervention on neighboring areas cause the target variables to settle down to another equilibrium over a few time steps.