A comparison of the economic benefits of urban green spaces estimated with NDVI and with high-resolution land cover data

Many cities around the world have undertaken greening programs (e.g., planting urban trees, adding or enhancing parks, providing incentives for green roofs) to benefit from the amenities of urban green spaces. To evaluate the economic benefits of these programs, hedonic pricing is the approach of choice but its application depends on the availability of adequate land cover data. In this context, this paper contrasts results from Cliff-Ord spatial hedonic models applied to single family houses in Los Angeles (California, USA) with land use data at two spatial resolutions: (1) high resolution (0.6m) but relatively expensive classified land cover (CLC) data; and (2) moderate-resolution (30 m) but low-cost Normalized Difference Vegetation Index (NDVI) data from satellite imagery. Our results show that elasticities of price with respect to NDVI and with respect to tree canopy cover (TCC) are weakly correlated and often give conflicting signals for prioritizing tree planting efforts. In addition, correlations between NDVI and CLC grass measures are low, which suggest that hedonic models based on NDVI would miss the benefits of grassy areas. Our spatial lag-Tobit models that predict tree canopy cover from NDVI, structural characteristics, neighborhood amenities, and socio-demographic variables highlight the relative importance of the latter in explaining tree canopy cover in the vicinity of single family houses in Los Angeles; these characteristics are much less important to explain parcel TCC. Overall, our results highlight the advantages of using high-resolution CLC data over moderate resolution NDVI data to estimate the economic benefits of urban greening programs. (C) 2014 Elsevier B.V. All rights reserved.