Dataset for: A Typology-Based Scalable Framework for Rooftop Rainwater Harvesting to Improve Industrial Water Sustainability in Bangladesh's Textile Industry
Description
This dataset supports a study investigating whether industrial-scale rooftop rainwater harvesting (RWH) feasibility can be screened using a simplified structural typology, rather than requiring detailed facility-by-facility audits. The underlying hypothesis is that two readily observable structural variables, workforce scale and usable rooftop catchment area, are sufficient to predict RWH demand coverage and emission-reduction potential across heterogeneous industrial facilities. Data collection: 67 textile factories were selected via stratified random sampling (90% confidence, finite population correction, N=6,447 registered units) from Bangladesh's two dominant textile industrial belts (Dhaka-Gazipur-Narayanganj-Narsingdi and Chattogram). For each factory, workforce size was obtained from DIFE registration categories; usable rooftop area was digitised from high-resolution satellite imagery and classified into Small/Medium/Large classes using Jenks Natural Breaks; annual harvestable rainwater volume was estimated from local rainfall and rooftop area; non-potable water demand was estimated from workforce size; and CO2e emission reductions were calculated from avoided groundwater pumping energy. Each factory was assigned to one of nine workforce-rooftop typologies (T1-T9), a 3x3 matrix crossing three workforce bands with three rooftop size classes. Key findings: demand coverage varied systematically by typology, from 6% (High Workforce-Small Roof, T7) to 499% (Medium Workforce-Large Roof, T6). Three structural feasibility zones emerged: Surplus (>100% coverage: T1, T2, T6), Balanced (50-100%: T5, T9), and Deficit (<30%: T4, T7, T8). One typology (T3, Low Workforce-Large Roof) had zero sampled factories. Emission reductions ranged from 0.15 to 2.82 t CO2e/yr per factory and scaled primarily with rooftop size rather than coverage percentage. Interpretation and use: each row represents one factory; columns report raw inputs (location, workforce, rooftop area) and derived outputs (typology, coverage %, emission savings). Researchers can use this dataset to replicate the typology classification, verify reported summary statistics, or extend the analysis to other industrial contexts or sectors.
Files
Steps to reproduce
1. Sample selection: Apply stratified random sampling to DIFE-registered textile factories in Dhaka and Chattogram Divisions, using a finite population correction formula (N=6,447, z=1.645 for 90% confidence, e=0.10, p=0.5) to derive the target sample size of 67 factories, proportionally allocated across industrial subsector and geographic distribution. 2. Rooftop digitisation: For each sampled factory, delineate the rooftop boundary manually using high-resolution satellite imagery (e.g., Google Earth Pro), applying a rooftop utilisation factor to convert digitised area to usable catchment area. 3. Size classification: Classify usable rooftop area into Small, Medium, and Large classes using Jenks Natural Breaks, applied across the full sample. 4. Typology assignment: Cross-classify each factory's workforce band (Low/Medium/High, from DIFE registration category) with its rooftop size class to assign one of nine typologies (T1–T9). 5. Harvest volume and demand: Estimate annual harvestable rainwater volume from local rainfall data and usable rooftop area; estimate non-potable water demand from workforce size. 6. Coverage and emissions: Calculate demand coverage (%) as harvested volume divided by demand; calculate CO2e emission reductions from the groundwater pumping volume displaced, using a regional emission factor. Full formulas and parameter values are reported in the accompanying manuscript's Methods section.