Abstract
This paper estimates global logistic regression and logistic geographically weighted regression (GWR) models of urban growth in the adjacent border cities of Laredo, Texas in the United States and Nuevo Laredo, Tamaulipas in Mexico, for two time periods from 1985 to 2014. Historical land use and land cover patterns were monitored through Landsat imagery from the United States Geological Survey to identify instances of urban growth through land type change. Data on socioeconomic variables related to urban growth were collected from various sources and used as independent variables. In both time periods, the logistic GWR was proven to be a complementary model for the global logistic regression to explore the urban growth effect. In addition, GWR outperformed the global logistic regression model with respect to goodness of fit. These results suggest that local models are complementary to global models to empirically analyze the determinants of urban growth in study areas that contain political borders, presumably because the relationships between socioeconomic factors and urban growth are characterized by spatial heterogeneity in such areas. The spatial variable of the relationship between urban growth and the neighborhood interactions and proximity effect present the idea of complexity and interconnections between the land use change and associated factors.
Highlights
Interest in spatial patterns and spatially based drivers of urban growth has been increasing steadily in the literature that engages with the geographical analysis of land use and land cover change (e.g., [1,2,3])
Research has shown that geographically weighted regression (GWR) can outperform global regression models with respect to residual spatial autocorrelation and model goodness of fit [13]. To see if these expectations are borne out in a selected border study area, this paper examines two specific questions for the U.S.–Mexico border cities of Laredo and Nuevo Laredo: (1) What variables were most significantly related to urban growth/land cover change between 1985 and 2014? (2) Do GWR models outperform global regression models of urban land use change in understanding the relationships between urban growth and these explanatory variables in our international study area? If the relationships between these measurable factors and land cover change manifest in different ways on opposite sides of an international border, we should be able to infer that socioeconomic and institutional variables play a prominent role in patterns of urban growth
The global logistic regression model resulted in moderate goodness of fit, with −2 Log likelihood values of 2433.01 and 2369.87 for the time periods 1985–2000 and 2000–2014, respectively (Table 2)
Summary
School of Geographical Sciences, Northeast Normal University, Changchun 130024, China Key Laboratory of Geographical Processes and Ecological Security of Changbai Mountains, Ministry of Education, Northeast Normal University, Changchun 130024, China Received: 16 August 2020; Accepted: 22 September 2020; Published: 24 September 2020
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
