Urban vacant lands impart hydrological benefits across city landscapes

The study investigates whether urban vacant parcels, created primarily through demolition in shrinking cities, provide hydrologic benefits by detaining rainfall and reducing stormwater runoff into aging wastewater systems. With vacant land comprising over 15% of land area in many U.S. cities and cities facing mandates to reduce combined sewer overflows, the authors hypothesize that vacant lots have high rainfall detention capacity (RDT), partitioning more precipitation to infiltration than runoff. Using Buffalo, NY—a city with extensive demolition and vacancy—as a case study, the research aims to quantify parcel-scale infiltration, assess how land cover and topography influence detention, and scale parcel findings to estimate citywide impacts on hydrologic partitioning across varying precipitation regimes.

The paper situates vacant parcels within the broader context of urban green infrastructure (GI) as decentralized stormwater controls known to reduce runoff volumes and CSO frequency. Prior work documents GI effectiveness (e.g., rain gardens, green roofs, street trees), the role of lawns in disconnection of impervious areas, and the importance of patchwork green spaces in urban hydrology. Urban soils are often disturbed, compacted, and heterogeneous, with evidence of lost intermediate soil horizons and altered permeability compared to reference soils, making city-specific field data critical. Management frameworks have proposed leveraging vacant land for stormwater control in shrinking cities, but empirical, citywide quantification of vacant lots’ detention capacity has been limited. The study builds on this literature by providing direct measurements of infiltration and integrating land cover/topography to estimate citywide detention benefits from vacant land portfolios.

• Study area and sampling: The Urban Vacant Land Assessment protocol was applied across Buffalo, NY. Detailed field measurements were collected on 718 vacant parcels across 15 sewersheds; infiltration rates and soil texture were measured and averaged for 520 parcels. Topography and land cover were assessed at nine stations per lot; soil compaction was measured with a single-mass penetrometer at a minimum of four locations per lot.
• Infiltration measurement: At two locations per lot (center and rear), maximum infiltration rates were determined using a Mini Disk Infiltrometer under unsaturated conditions. Rates were estimated as unsaturated hydraulic conductivity by fitting cumulative infiltration versus time using an infiltration model with van Genuchten parameters (based on field-assessed soil texture), infiltrometer radius, and a set suction head of -2 cm. Negative fitted rates (physical impossibility) were set to the dataset minimum (0.001 cm h−1). Measurement uncertainty was quantified, with 99% confidence intervals reported for a subset (n=426).
• Land cover and topography: Vegetation cover (as a proxy for pervious area) was assessed in 25% increments. Parcel slopes and drainage orientation were derived from field surveys and parcel geometry, noting most parcels drain toward the street with low gradients (<1.5%).
• Pre-demolition and redevelopment states: Former building footprints were hand-delineated from 2002 orthoimagery for 2482 parcels to estimate pre-demolition impervious area. 2017 imagery captured redevelopment/incomplete demolition footprints. Pre-demolition footprints were considered a conservative estimate (not including sidewalks/driveways/garages).
• Rainfall data and design storms: Hourly warm-season (April–October) precipitation (1997–2017) from Buffalo Niagara International Airport was used to characterize rainfall intensities and frequencies. Hourly data were cross-checked against daily datasets for consistency.
• RDT modeling: A simple infiltration-excess framework estimated rainfall detention (RDT) and overland flow (OF) at the parcel scale for 1-hour events. For a parcel with average maximum infiltration rate Imax, pervious fraction V, parcel area A, and hourly rainfall intensity P: • If P ≤ Imax: RDT = P*(V/100)A; OF = P(1 − V/100)A. • If P > Imax: RDT = Imax(V/100)A; OF = (P − Imax)(V/100)A + P(1 − V/100)*A. Assumptions include treating unvegetated/bare/impervious surfaces as non-contributing to RDT and assuming free drainage at depth.
• Citywide scaling: The distribution of measured parcel Imax (n=520) was randomly resampled to 2482 parcels (with known areas and building footprints) 1000 times to estimate cumulative RDT across three states: pre-demolition (circa 2002), post-demolition (entire lot pervious, upper bound), and redevelopment (circa 2017). Warm-season totals were computed for each year (1997–2017). Uncertainty from infiltration measurement propagated to RDT estimates (envelopes shown for subsets).

• Infiltration rates: Parcel-average maximum infiltration rates ranged from 0.001 to 5.39 cm h−1 (n=520), with weak spatial structure and variability spanning three orders of magnitude within sewersheds. No clear link was found between demolition technique and measured infiltration rates.
• Rainfall-climate context: Over 1997–2017 warm seasons, 93% of hourly rainfall intensities were <0.5 cm h−1. Approximately 80% of parcels had infiltration rates ≥ the median hourly intensity (0.08 cm h−1), and ~60% ≥ the 75th percentile (0.18 cm h−1).
• Event-scale detention: For common low-intensity 1-hour storms (0.01–0.05 cm h−1), the ~500 lots were estimated to infiltrate about 15.5 to 83.9 m³ per event; for hourly events exceeding 1 cm depth, >546 m³ per event. At rainfall ≤0.01 cm h−1, vacant lands infiltrate an estimated 91–94% of precipitation volume. Runoff generation typically initiates when rainfall rates exceed ~0.5 cm h−1, but such high-intensity hours are rare (<2%). Only ~4% of parcels had minimum infiltration rates near 0.01 cm h−1.
• Parcel characteristics supporting detention: 77% of parcels were entirely vegetated; parcels had low slopes (<1.5%) and predominantly drained toward streets (~88%). Compaction was greater at street-facing fronts; most lot areas had less compact soils and vegetation that promote infiltration and surface storage.
• Citywide impacts: Demolition across ~2482 properties increased cumulative rainfall detention rate by an average of 52% (range 51–54%) compared to pre-demolition. Historical imagery suggests ~0.29 km² of pervious area gained across these lots, with ~0.0028 km² impervious area present post-2017 from redevelopment/incomplete demolition. Warm-season infiltration volumes during a wetter year were ~436,000 m³ (range 420,000–452,000 m³) and during a drier year ~225,000 m³ (range 217,000–232,000 m³). Normalized gains in infiltration and reductions in runoff were relatively consistent across years, indicating persistent benefits across precipitation regimes. Vacant land disconnects/eliminates a substantial amount of connected impervious surface (~34% of vacant land area), creating a distributed hydrologic sink.

Findings demonstrate that vacant parcels function as decentralized green spaces with substantial rainfall detention capacity, particularly for frequent, low-intensity storms that dominate the warm season. The patchwork of pervious vacant lots amidst impervious surfaces enhances disconnections within the drainage network, reducing direct runoff inputs to combined/separate sewer systems and potentially lowering CSO frequency and volume. Despite disturbances from demolition, most parcels exhibit vegetation, low slopes, and limited high-compaction zones, enabling infiltration to exceed typical rainfall intensities on most lots. The citywide scaling reveals that demolition-driven conversion from impervious to pervious surfaces yields sizable cumulative infiltration gains (≈51–54%), supporting consent-order objectives for overflow reduction. The study situates vacant lands alongside GI, lawns, and urban trees as complementary components of a portfolio approach to stormwater management. While the simple infiltration-excess model focuses on surface partitioning, processes like interception and evapotranspiration likely augment detention, suggesting estimates are conservative. The broader implication is that socio-ecological dynamics (development, vacancy, redevelopment) substantially regulate urban hydrologic partitioning and can be leveraged to improve stormwater outcomes.

Urban vacant land constitutes a tangible stormwater resource capable of retaining substantial rainfall volumes on seasonal and annual bases, with greatest benefits for frequent, low-intensity events. Even under projections of more intense storms, vacant lots are expected to provide net benefits by detaining precipitation up to threshold depths and reducing/delaying runoff delivery to sewer systems. Integrating vacant lands with retrofit green infrastructure can enhance resilience and scale benefits citywide. Cities experiencing similar socio-ecological cycles may realize comparable hydrologic gains by intentionally incorporating vacant land into stormwater strategies, alongside residential lawns and other GI, and by prioritizing equitable redevelopment. Future work should refine parcel selection for stormwater function, align redevelopment with social equity, and expand standardized field assessments to better constrain subsurface controls and inform city-scale planning.

• Subsurface uncertainty: The model assumes free drainage; limited data on urban subsurface architecture, macroporosity, and water table dynamics introduce uncertainty (e.g., potential for saturation-excess runoff in localized high water table areas).
• Simplified surface processes: The model does not explicitly include interception, evapotranspiration, surface depression storage, or lateral redistribution, likely making detention estimates conservative but uncertain.
• Impervious routing assumptions: Building footprints pre- and post-demolition are treated as fully impervious and directly connected to sewers. This may overestimate runoff in some cases and underestimate pre-demolition impervious contributions from ancillary features (driveways/sidewalks), making the net benefits likely conservative.
• Measurement/model limits: Infiltration estimated from Mini Disk Infiltrometer under set suction with texture-based van Genuchten parameters; very slow infiltration fits required truncation to 0.001 cm h−1. Spatial heterogeneity within lots and temporal variability are simplified by using average maximum infiltration rates.
• Generalizability: Results are specific to Buffalo’s climate (frequent low-intensity storms), soils, and demolition practices; applicability to other cities requires local assessment of rainfall regimes, soils, geology, and vacancy characteristics.
• Potential contamination: Infiltration on vacant lots could mobilize contaminants from prior land uses or road-derived pollutants; water quality impacts were not quantified.