Tuesday , June 22 2021

Algorithms and large data sets insure against climate threats – Radio 102.9

l When starting a new campaign, the producer encounters two large variables that he does not control: climate and markets.

To give you some predictability, you can predict sales (forward) or use tools such as futures contracts and options available on the markets. But when it comes to climate, it is only insurance against hail: in the face of drought and flood the producer is not always insured.

Hail policy offered for decades is the only thing that a producer must manage the growing and complex climate risk.


S4 is an Argentine technology company applying to agriculture, with offices in Argentina, Brazil and the United States. Developed a technology that accurately reflects the climate impact on arable crops, from satellite images provided by NASA and the European Space Agency "Today the market reacted and saw the opportunity to innovate in this field." Satellite databases, as well as Big Data processes in the cloud, helped develop the index The index is reflected in the Rofex market and is used by risk takers, such as Munich Re, to offer services to farmers, "said Santiago González Venzano, founder of S4.


The indexes are built from a database of satellite images from across the planet and have a history that began in 2000. Thanks to machine learning processes, algorithms can be developed that recognize crops: each crop has its own "spectral character" as if it were a "bar code". In addition to knowing what crop is (soy, maize, wheat, etc.), another algorithm allows us to "take an impulse" and know how it develops: at what stage of development is and what is its strength. In this way, indicators are constructed that reflect the growth of each crop in each town. Then, knowing the history of the index in each locality, a mathematical risk model is developed.

This risk calculation allows the "risk taker" to charge a bonus in exchange for compensation to farmers when the Index falls below a set value.

Source link