Abstract:Quantifying the spatial association between ecological indicators (e.g., chlorophyll-a) and environmental parameters is crucial for explaining the ecological status in coastal ecosystems. Although global and local regression models have been widely used to estimate spatial relationships in marine environmental processes, spatial anisotropy caused by strong coastal-inland environmental gradients has not been investigated. This is very likely to result in incomprehensive characterization of the coastal ecological status. To better quantify the spatially anisotropic nonstationary relationship in coastal environments, a spatial proximity neural network (SPNN) was proposed in this paper to address the nonlinear effects of spatial anisotropy. A directional geographically neural network weighted regression (DGNNWR) model was accordingly developed by combining a geographically neural network weighted regression (GNNWR) with SPNN to incorporate anisotropic impacts into spatial nonstationarity. Modeling of chlorophyll-a in Zhejiang coastal areas of China in the spring over 2015-2017 was conducted to examine its performance. The results demonstrated that DGNNWR achieved a better fitting accuracy and a more adequate prediction ability than ordinary linear regression (OLR), geographically weighted regression (GWR), GNNWR, and anisotropic-based GWR models. Notably, compared to the best comparison model, the fitting error indicators were declined for more than 30% and the fitted R2 was considerably increased from 0.83 to 0.92 using our proposed DGNNWR. The spatial mapping of parameter estimates confirmed that DGNNWR successfully handled the anisotropic nonstationarity in coastal environments and quantified the main driven parameters of Chl-a. Based on the spatially refined relationship between Chl-a and environmental parameters, we further characterized the spatial and temporal distributions of Chl-a in Zhejiang coastal areas in the spring of 2015-2017, and then investigated the impacts of riverine discharges and ocean currents on the spatiotemporal variations of Chl-a. The findings are crucial to formulate appropriate mitigation strategies for eutrophication and are meaningful for the management of coastal ecosystems.

Spatial and Attribute Neural Network Weighted Regression for the Accurate Estimation of Spatial Non-Stationarity

Geographically neural network weighted regression for the accurate estimation of spatial non-stationarity

Geographically and temporally neural network weighted regression for modeling spatiotemporal non-stationary relationships

Geographically convolutional neural network weighted regression: a method for modeling spatially non-stationary relationships based on a global spatial proximity grid

GNNWR: An Open-Source Package of Spatiotemporal Intelligent Regression Methods for Modeling Spatial and Temporal Non-Stationarity

GNNWR: an open-source package of spatiotemporal intelligent regression methods for modeling spatial and temporal nonstationarity

A Generalized Spatial Autoregressive Neural Network Method for Three-Dimensional Spatial Interpolation

Modeling Spatially Anisotropic Nonstationary Processes in Coastal Environments Based on a Directional Geographically Neural Network Weighted Regression.

Geographically and temporally weighted neural networks for satellite-based mapping of ground-level PM2.5

A neural network model to optimize the measure of spatial proximity in geographically weighted regression approach: a case study on house price in Wuhan

Improving Geographically Weighted Regression Considering Directional Nonstationary for Ground-Level PM2.5 Estimation

A geographically weighted artificial neural network

Integration of satellite observations and station measurements for the mapping of ground-level PM2.5 using geographically and temporally weighted neural networks.

Geographically weighted regression based on a network weight matrix: a case study using urbanization driving force data in China

The estimation of hourly PM2.5 concentrations across China based on a Spatial and Temporal Weighted Continuous Deep Neural Network (STWC-DNN)

Deep Learning for the Spatial Additive Autoregressive Model with Nonparametric Endogenous Effect

A Novel Flexible Geographically Weighted Neural Network for High-Precision PM2.5 Mapping across the Contiguous United States

A Spatiotemporal Autoregressive Neural Network Interpolation Method for Discrete Environmental Factors

Geographically weighted regression with the integration of machine learning for spatial prediction

A geographically weighted deep neural network model for research on the spatial distribution of the down dead wood volume in Liangshui National Nature Reserve (China)

Neural Networks for Geospatial Data