Abstract
Timely griddedgross domestic product (GDP) data is a fundamental indicator in many applications. It is critical to characterize the complex relationship between GDP and its auxiliary information for accurately estimating gridded GDP. However, few knowledge is available about the performance of deep learning approaches for learning this complex relationship. This article develops a novel convolutional neural network based GDP downscaling approach (GDPnet) to transform the statistical GDP data into GDP grids by integrating various geospatial big data. An existing autoencoder-based downscaling approach (Resautonet) is employed to compare with GDPnet. The latest county-level GDP data of China and the multiple geospatial big data are adopted to generate the 1-km gridded GDP data in 2019. Due to the different related auxiliary data of each GDP sector, the two downscaling approaches are first separately built for each GDP sector and then the results are merged to the gridded total GDP data. Experimental results show that the two deep learning approaches had good predictive power with <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</i> <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> over 0.8, 0.9, and 0.92 for the three sectors tested by county-level GDP data. Meanwhile, the proposed GDPnet outperformed the existing Resautonet. The average <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</i> <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> of GDPnet was 0.034 higher than that of Resautonet in terms of county-level GDP test data. Furthermore, GDPnet had higher accuracy ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</i> <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> = 0.739) than Resautonet ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</i> <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> = 0.704) assessed by town-level GDP data. In addition, the proposed GDPnet is faster (about 78% running time) than the Resautonet. Hence, the proposed approach provides a valuable option for generating gridded GDP data.
Highlights
G ROSS domestic product (GDP) is a monetary metric that measures the market value of all final goods and services within a given region during a period of time, often seasonally and annually [1,2,3]
This paper presents a novel downscaling approach using deep learning, GDPnet, for estimating gridded GDP data from traditional statistical GDP data by integrating a variety of geospatial big datasets
It aimed to take advantage of convolutional neural network and residual connection to characterize the complex relationship between GDP and auxiliary data
Summary
G ROSS domestic product (GDP) is a monetary metric that measures the market value of all final goods and services within a given region during a period of time, often seasonally and annually [1,2,3]. The road network density was adopted by Murakami and Yamagata [22] to represent the industry activities and it was combined with land cover as auxiliary variables to downscale the national-level GDP to generate global GDP grids at the spatial resolution of 0.5-degree in 1980, 1990, 2000 and 2010. To fill the abovementioned research gaps, we first develop a novel GDP downscaling approach using one-dimension convolutional neural network and residual connection (GDPnet) Both the proposed GDPnet and the existing Resautonet are employed to generate the gridded GDP data of China using the multiple geospatial big data and the latest county-level statistical GDP data from the 2020 yearbook.
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
