Vacant parcel assessment

To assess the thousands of demolitions that have occurred across Buffalo, NY (USA; Supplementary Fig. ¹), an Urban Vacant Land Assessment protocol was developed and applied to vacant parcels across the city^63,64. This protocol is a per lot assessment of general site information, parcel size, topography, vegetation, wastes, soil type, and hydrologic and land cover surveys of pervious and impervious surfaces (Supplementary Fig. ⁹; Supplementary Note ⁴). As part of the extended protocol, measurements of surface infiltration, soil type, land cover, and topography were made across each lot across 718 vacant parcels (Supplementary Table ¹²) within 15 sewersheds (Supplementary Table ¹). Topography and land cover were assessed for nine station points across each lot (Supplementary Fig. ⁹). The percentage of each lot covered by vegetation, taken to represent the fraction of pervious surface, was assessed on a scale from 0 to 100% with 25% increments (Supplementary Table ⁵). Soil compaction was assessed using a single mass penetrometer at a minimum of four locations per lot, with the assessor recording the number of blows to 5.1 cm (one increment), 10.2 cm (two increments), and 15.2 cm (three increments) at selected stations or until refusal. Soil texture and infiltration rates were assessed at two locations per lot, the lot center (capturing the area impacted by demolition) and lot rear (outside the demolition envelope), and averaged for 520 vacant parcels. Soil texture was assessed by feel⁶⁵ (Supplementary Table ¹³). Maximum infiltration rates were calculated from measured volumes of water infiltrated through time recorded with a Mini Disc Infiltrometer (© METER Group Inc), which enables a rapid approach to determine infiltration rates as unsaturated hydraulic conductivity⁶⁶. Maximum infiltration rates were estimated as unsaturated hydraulic conductivity⁶⁷ by fitting rates of cumulative infiltration as water volume through time using a simple infiltration model incorporating van Genuchten parameters for a given soil type⁶⁸, the radius of the infiltrometer, and the suction at the disk surface (set during each field experiment to -2 cm). Further details are included in Supplementary Note ⁴. Like all models, this simple approximation has limits, and generates negative rates for very slow infiltration rates. As this represents a physical impossibility, negative rates were converted to the smallest value measured in the dataset, 0.001 cm h⁻¹. Assessments were paired with existing information on the date of demolition, to determine whether infiltration rates were related to demolition type (Supplementary Fig. ²), and parcel geometry, to extract land surface slope (Fig. 3). Infiltration rates were subject to an assessment of uncertainty, with reported confidence intervals (99%) shown in Supplementary Figs. ¹0 and ¹¹.