Envirotyping and Digitalization
This project will develop the next-generation agricultural crop yield prediction framework that incorporates the effects of climate change. Yield forecast is critical for decision-makers at farm, state, regional, and national levels for quick and well-informed decision-making. An accurate, early yield prediction can help farmers to decide on what to grow and when to grow. It can also help governments decide on crop insurance payouts, export planning, etc. The new framework will provide more accurate yield predictions for the current crop season, as well as forecasts based on climate projections for the near future.
Crop yields are intricately linked to the availability of adequate soil water during certain critical stages of the crop's growth timeline. In the absence of the required quantity of easily accessible water in the soil during these critical periods, significant drops in yield may be expected. Plants experiencing water stress cannot build adequate biomass, which in turn can lead to weak flowering and production. The proposed research project will use a process-informed data science approach aimed towards quantifying this relationship between available water resource, growth stage, and yield for different crops and sub-types. Such quantification, based on past data, would provide a path to forecast yield estimates as a function of the expected weather/climate scenarios in the near future. The system will incorporate, among others:
•Historical yield and environmental data
•Weather prediction models to generate probable future climate scenarios over 5-10 years
•Crop phenology and genetic traits specific to Russia
•Site-specific information such as soil health, water dynamics, fertilizer application, etc.
•Advanced techniques such as Spatio-Temporally Weighted Deep Neural Networks to overcome the shortcomings of existing algorithms that only consider spatial non-stationarity for yield estimation.
The individual products, and their advantages, are listed below:
• Yield predictions for current crop season
- Helps understand probable return on investment
• Yield forecasts for future years based on climate projections
- Helps plan future planting operations – variety, timing, extent
- Helps prioritize crop breeding for future seasons
• Field-scale (not regional) predictions for each crop type
• Identification of drought scenarios in near-future
- Helps planning for effective resource (water, fertilizer, etc.) utilization and scheduling
- Allows planning of operations based on potential drought severity
• Access to enhanced research products in future from Skoltech Agro
- Improved environmental models, envirotyping, …
Intelligent forecast framework for crop yields incorporating effects of environmental factors such as drought and climate change