A Weekly Multi-Modal Dataset of Soil Physicochemical Parameters, Environmental Conditions, Irrigation Records, Crop Growth Observations, Harvest Parameters, and Field Images from Six Cowpea and Maize Varieties in Sumbrungu, Bolgatanga, Ghana
Year: 2026
Type: dataset
Abstract: This dataset presents weekly multi-modal field trial data collected at the Sumbrungu Experimental Farm, Bolgatanga, Upper East Region, Ghana, under the LSACROFT Project (Development of a Low-cost Soil Analysis Device for Real-time Soil Monitoring, Crop Recommendation, and Fertilizer Optimization). Data were collected over a complete 15-week growing season from 14 March 2026 to 26 June 2026, covering six crop varieties: three cowpea (IT-17K-1367-2-3, SARI-TUGA, TUZIE-VALLENG) and three maize (Sanzal-sima, Wang-daata, Bahilifa), under a Randomised Complete Block Design with three replications and one control treatment. The dataset integrates seven complementary data streams. First, weekly soil physicochemical parameters measured using a low-cost customized soil sensor, including electrical conductivity, pH, nitrogen, phosphorus, potassium, moisture, temperature, salinity, and total dissolved solids, each retained at three processing stages: raw, unit-converted, and laboratory-calibrated (2,713 readings). Second, weekly atmospheric environmental parameters including mean air temperature, relative humidity, and total rainfall. Third, weekly sprinkler irrigation records including number of events, volume per event, total volume applied, and water source. Fourth, all farm management activities per plot per week including sowing, fertilizer application, weeding, thinning, and harvesting. Fifth, weekly cowpea crop growth and harvest observations including plant height, leaf count, stem diameter, growth stage, pod count, pod length, fresh weight, dry weight, harvest index, 100-seed weight, and stover yield (116 records). Sixth, weekly maize crop growth and harvest observations including plant height, leaf count, stem diameter, growth stage, ear length, rows per ear, grains per row, and 100-grain weight (144 records). Seventh, field photographs of all plots from Week 3 to Week 12. The dataset is intended for machine learning model development, TinyML embedded AI deployment, crop yield prediction, soil sensor calibration research, and AI-driven agricultural advisory system design. It is particularly relevant for researchers working on precision agriculture and climate-resilient food systems in the Guinea Savanna and Sudan Savanna zones of sub-Saharan Africa. All files are in Microsoft Excel format. Images are provided as a zipped archive. The dataset is released under CC BY 4.0.
Source: Mendeley Data
DOI: 10.17632/3wzp9nz3fh
Institutions: Institute for Scientific and Technological Information, Bolgatanga Technical University, Ghana Communication Technology University, Dr. Hilla Limann Technical University, Council for Scientific and Industrial Research, Council of Scientific and Industrial Research, Simon Diedong Dombo University of Business and Integrated Development Studies, University of Zambia
Language: English
Topic: Smart Agriculture and AI
Subfield: Plant Science
Domain: Life Sciences
Keywords: cowpea, maize, soil physicochemical properties, soil monitoring, crop yield prediction, meteorological observations, crop varieties, precision agriculture, TinyML, agricultural advisory system, crop recommendation, fertilizer optimization, field photographs, Guinea savanna, Sudan savanna, machine learning, sensor calibration, climate-resilient food systems
SDG: Zero hunger
Open Access status: green
OpenAlex ID: W7214025750
