2026-10-12T15:38:05Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33952572 2026-09-21T12:41:29Z category_272 category_39 category_734 category_64 portal_87 item_type_3 month_year_09_2026

Estimating seasonal urban greenery with street view and satellite imagery: a cross-season weight regression approach Jin Zhu (164147) Fan Zhang (46132) Yuan Ding (617135) Kai Cao (252808) Ziyue Cao (25083501) Xiangnan Gao (25083504) Environmental Sciences not elsewhere classified Ecology Biological Sciences not elsewhere classified Cancer Street view imagery Sentinel-2 imagery green view index normalized difference vegetation index (NDVI) seasonality local regression model The green view index (GVI) derived from street view imagery (SVI) has become a key metric for assessing urban greenery at the streetscape level. However, the resource-intensive nature of SVI acquisition leads to substantially lower temporal resolution and irregular sampling intervals. This results in significant temporal gaps in historical SVI, rendering them unsuitable for extracting and analyzing seasonal GVI variations. This may potentially introduce biases into related research. In this study, the seasonal GVI gaps are estimated using the cross-season weight regression (CSWR) model, which significantly improves the accuracy of GVI estimation for missing seasons through an innovative modification of the classical local regression (LR) model. For intra-season and cross-season auxiliary data points, CSWR constructs local and seasonal weights using Gaussian kernel functions and Pearson distance metrics, respectively. The two types of weights are then integrated, and the model coefficients are estimated via weighted least squares to reconstruct missing-season GVI values. Using Nanjing, China, as a case study, our results show that CSWR achieves R2 values of 0.513 (spring), 0.663 (summer), 0.557 (autumn), and 0.510 (winter), representing improvements of 9.8, 19.4, 14.9, and 10.9% points, respectively, over ordinary least squares (OLS). Moreover, the estimation accuracy of CSWR improves as the number of cross-season auxiliary data points increases. This study provides, for the first time, a comprehensive seasonal mapping of streetscape greenery across an entire urban road network. CSWR offers broad utility for cities and regions exhibiting seasonal greenery dynamics. 2026-09-21T12:41:29Z Dataset Dataset 10.6084/m9.figshare.33952572.v1 https://figshare.com/articles/dataset/Estimating_seasonal_urban_greenery_with_street_view_and_satellite_imagery_a_cross-season_weight_regression_approach/33952572 CC BY 4.0