Supercomputer will help to increase yields in the Black Earth region
Researches at the Skolkovo Institute of Science and Technology (Skoltech) conducted a sensitivity analysis and identified the most important factors, affecting the yield of various crops in the Black Earth region using the Zhores supercomputer.
To predict the productivity of agricultural crops, digital models of crop formation are used, which describe the processes occurring in the soil, climatic conditions and properties of crops. To improve the quality of the forecasting, various input data is used, including information on the environment and management features in agriculture. At the same time, users of such models in some countries don’t have access to agrochemical data, which significantly increases the cost and labor intensity of the prediction calibration.
The researchers at Skoltech used one of the popular publicly available Monica models and found a way to highlight the most important factors affecting yield, based on historical data and the results of simulations of technological processes. With the use of the best-in-class supercomputer Zhores, the scientists managed to increase the efficiency of calculations and conduct not one simulation experiment per day but up to 500 thousand experiments per hour.
Such an enormous amount of computation is required to conduct a qualitative sensitivity analysis, which helps to determine how individual changes (such as soil or fertilizer parameters) affect crop yield.
“Obtaining information about soils is an extremely difficult problem in our country. Unfortunately, data on soil properties and yield is not published. Nevertheless, we found a way out of the situation using the Zhores supercomputer to solve this problem. Now we can simulate all possible options and determine the most important yield parameters, without performing time-consuming and expensive operations. We hope that our achievements will contribute to the further implementation of digital technologies in Russian agriculture,” said Senior Lecturer at Skoltech Maria Pukalchik.
