AgriClimateBD: A Satellite Enriched Precision Agriculture Dataset for Bangladesh

Description

AgriClimateBD is a comprehensive precision agriculture dataset developed for Bangladesh by integrating agricultural production statistics, crop phenological information, and satellite-derived climate variables. The dataset contains 4,607 records and 33 features representing 73 crop types cultivated across 64 districts of Bangladesh. The dataset was developed by extending the previously published SPAS-Dataset-BD through the incorporation of climate information obtained from the NASA POWER (Prediction of Worldwide Energy Resources) platform. Agricultural production, area, temperature, and humidity information were collected from the Bangladesh Bureau of Statistics (BBS) Statistical Yearbook 2022, while crop lifecycle information, including transplanting, growth, harvesting periods, and seasonal classifications, was obtained through field surveys involving 223 farmers from diverse agroecological regions of Bangladesh. A Python-based automated data enrichment pipeline was employed to retrieve daily climate observations from the NASA POWER API for all 64 districts. Four climate parameters—precipitation, solar radiation, wind speed, and evapotranspiration—were aggregated into phenologically meaningful temporal windows. These include full-season summaries, pre-transplant month, transplant month, and post-transplant month conditions. The resulting climate features were integrated with agricultural and phenological attributes to create a unified dataset suitable for precision agriculture research. The dataset includes agricultural variables such as crop name, district, cultivated area, production, season, transplanting period, growth period, and harvest period; meteorological variables including average, maximum, and minimum temperature and humidity; and sixteen NASA POWER-derived climate variables representing rainfall, solar radiation, wind speed, and evapotranspiration across multiple crop-development stages. AgriClimateBD supports a wide range of applications including crop yield prediction, crop classification, climate-resilient agriculture, agricultural decision support systems, precision irrigation management, machine learning, deep learning, and agricultural policy analysis. The dataset is particularly valuable because it provides phenologically anchored climate information for Bangladesh’s agricultural systems, including numerous underrepresented crop species that are rarely available in public agricultural datasets. The dataset is provided in CSV format and is intended to facilitate reproducible research in agricultural informatics, climate-smart agriculture, remote sensing, and data-driven farming systems.

Steps to reproduce

1. Collect Agricultural Statistics -Obtain crop production, cultivated area, temperature, and humidity data from the "Bangladesh Bureau of Statistics (BBS) Statistical Yearbook 2022". -Extract district-wise records for all available crops across the 64 districts of Bangladesh. 2. Collect Crop Phenology Information -Conduct field surveys using structured questionnaires to gather crop-specific information, including transplanting period, growth period, harvesting period, and cropping season (Kharif-1, Kharif-2, and Rabi). -Standardize crop names and lifecycle information across surveyed regions. 3. Data Cleaning and Harmonization -Remove duplicate entries and correct inconsistencies in crop names, district names, and seasonal labels. -Standardize measurement units for area, production, temperature, and humidity. -Handle missing values using appropriate imputation techniques and remove records with excessive missing information. -Merge agricultural statistics and survey-derived phenological information using crop and district identifiers. 4. Generate Derived Features -Calculate the Area-to-Production (AP) Ratio. -Create seasonal mapping variables that associate each crop season with corresponding calendar months. 5. Retrieve Climate Data from NASA POWER -Use district latitude and longitude coordinates to query the NASA POWER Daily API for the year 2022. -Download the following climate variables: >PRECTOTCORR (precipitation) >ALLSKY_SFC_SW_DWN (solar radiation) >WS10M (wind speed) >EVLAND (land evapotranspiration) 6. Create Seasonal Climate Features -Aggregate daily NASA POWER observations into crop-season windows. -Compute total seasonal rainfall and average seasonal solar radiation, wind speed, and evapotranspiration. 7. Create Phenology-Based Climate Features -Identify the primary transplanting month for each crop record. -Aggregate climate variables for: >the month before transplanting, >the transplanting month, >the month after transplanting. -Generate twelve transplant-window climate features (four climate variables × three temporal windows). 8. Merge Climate and Agricultural Data -Join the NASA POWER-derived climate variables with the agricultural and phenological dataset using district and season information. -Verify record completeness and consistency after merging. 9. Quality Assurance -Perform range validation, consistency checks, and descriptive statistical analysis. -Confirm that all records contain valid climate information and plausible agricultural values. 10. Export Final Dataset -Save the final integrated dataset as a CSV file containing 4,607 records and 33 attributes. -The resulting dataset constitutes the AgriClimateBD dataset.

Institutions

• Dhaka International University

  Dhaka Division, Dhaka

Categories

Licence