EOSDA

Cotton, often called the country’s “white gold,” has a long tradition of cultivation. Its production peaked during the Soviet period. In 1981, Azerbaijan’s share in the USSR cotton farming reached almost 10%, which amounted to 1.5% of global production.

Following the collapse of the Soviet Union, it dropped (Azerbaijan’s focus on the petroleum industry was a concomitant factor, too.) Searching for promising sources of income after the 2014–2016 global surge in oil prices, the Republic of Azerbaijan resumed its interest in the cotton sector.

Compared to the 2015/2016 market year, the cotton area in 2022/2023 has quadrupled, while yield per hectare increased by about 1.6 times.

Field-scale, regional, and country-wide crop yield prediction is indispensable for agricultural research, farm management, and import and export decision-making.

Challenge: Finding The Optimal Method Of Predicting Yields For Cotton And Tobacco On A District Level.
The project was large-scale and included the classification of several crops, besides cotton and tobacco yield prediction, which we carried out. The forecast had to be made for fields in two districts in Azerbaijan — Shaki and Sabirabad.

Solution: Analysing Statistical, Meteorological, Soil, And Phenology Data With The WOFOST Model

A customer provided statistical information on cotton and tobacco yields at the district level from 2000 until 2017.

The team, in turn, collected the remaining data. It included:

• Statistics indicating the features of crop cultivation necessary for a model to make predictions or calculations
• Geographic location of fields and their areas
• Dates of transplantation of tobacco plants from greenhouses or seedbeds.
• Meteorological data (humidity, precipitation, wind speed, maximum and minimum temperatures) from the Global Telecommunication System (GTS) and some specific models estimating of evapotranspiration and global radiation.
• Soil (soil moisture content and rooting depth) data.

In addition, the project team found the optimal sowing/transplantation dates for tobacco and cotton for 2010–2018.

Historical weather data is also indispensable for such tasks. Specialists sourced it from NASA POWER datasets. Meteorological parameters in NASA POWER datasets are taken from NASA’s Modern Era Retro-analysis for Research and Applications (MERRA-2) and the GEOS version 5.12.4 assimilation models. MERRA-2 is a new version of the Goddard Earth Observing System (GEOS) developed at the Global Modeling and Assimilation Office (GMAO) at NASA’s Goddard Space Flight Center.