Satellite Data To Evaluate Crop Bioproductivity

Challenge: The Need To Evaluate The Bioproductivity Of Crops Grown On Large Territories

In Kazakhstan, large land territories (22.9 million hectares — almost one-tenth of its area) are used for crop production. Local farmers usually manage 1–10 thousand hectares; that’s why owners seek technical approaches that make field monitoring of significant areas at once possible.

Satellite-based earth observation can provide such coverage. Kazakhstan’s space company needed to evaluate crops’ current vegetation characteristics and chose EOSDA to solve the task.

Solution: Defining Crop Bioproductivity With A Biophysical Model Using DMP Data

The project’s main goal was to assess the characteristics of the bioproductivity of five crops (barley, spring wheat, legumes, sunflower, and rapeseed) on an administrative level. The final report contained the current estimate of the total biomass (total biomass includes a plant’s roots, stems, leaves, and generative organs) and biomass of reproductive organs, for instance, grain in wheat or barley. The biomass of reproductive organs indicates yield.

Results of bioproductivity assessment can be used for yield prediction.

It’s one of the methods of solving this problem. Using bioproductivity data also allows assessing crop state and agrometeorological conditions for a certain area. It’s very good for understanding the factors affecting crop growth in the current and previous years and taking specific measures to improve the situation in the fields if needed.

Aleksey Kryvobok

Chief Science Officer at EOS Data Analytics

For a region-level analysis, the scientists used open data from the Copernicus Global Land Service program for 2018 and 2019.

Our idea was to use the Dry matter Productivity (DMP) product to define biomass characteristics of crops (i.e., how much green mass of vegetation has grown on a field). Relatively large field areas in Kazakhstan also drove our choice.

Aleksey Kryvobok

Chief Science Officer at EOS Data Analytics

Dry matter Productivity (DMP) represents the vegetation’s overall growth rate or dry biomass increase with units of measure used for solving agro-statistical tasks (kg/ha/day). The DMP data is generated every 10 days with a spatial resolution of 300 m.

Weather data from NASA Power and soil data from the FAO Soils Portal were also obtained and used for modeling.

After model calibration, the team compared this data with actual data and model calculations to obtain the total biomass for the growing season. EOSDA specialists determined the sowing and harvest dates based on the actual data. And the model was used to define the coefficient to transit from total biomass to the biomass of productive organs (yield) . The latter parameter is also known as the harvest index.

The accuracy of solving the task differed from region to region. For instance, the relative error of spring wheat biomass calculation was from 0.6–1% to 49% and 33.8% for barley.

For more accurate results, we would need to assess a significantly larger number of fields and crops.

Aleksey Kryvobok

Chief Science Officer at EOS Data Analytics

Outcome: Biomass Salutation For Crops Across Kazakhstan’s Regions

EOSDA specialists made a report on the total biomass and biomass of plant’s reproductive organs — yield. The team finished the project in two months. Data collection and preparation for model development took a significant amount of time.