Forecasting land-based environmental variables using similarity analysis and temporal graph convolutional neural networks

A computer-based system comprising a memory and a processor system communicatively coupled to the memory, wherein the processor system is operable to perform processor system operations operable to analyze a land region that has been decomposed into a plurality of sub-regions that capture spatial heterogeneity of environmental conditions across the land region, the processor systems operations comprising:

• applying a feature extraction process that extracts a set of sub-region environmental descriptors for each of the plurality of sub-regions;
• wherein the plurality of sub-regions comprise catchments that are irregularly-shaped;
• wherein the sub-region environmental descriptors comprise non-linear relationships with one another;
• wherein the set of sub-region environmental descriptors include dynamic descriptors that change over time, the dynamic descriptors comprising time series of daily values of the dynamic descriptors;
• applying a similarity analysis to the set of sub-region environmental descriptors to generate groups of the plurality of sub-regions;
• creating a plurality of group-based graphs by encoding each of the groups into a corresponding group-based graph;
• and
• using a spatio-temporal neural network to train a model based at least in part on the plurality of group-based graphs.

A computer program product comprising a computer readable program stored on a computer readable storage medium, wherein the computer readable program, when executed on a processor system, causes the processor system to perform processor system operations operable to analyze a land region that has been decomposed into a plurality of sub-regions that capture spatial heterogeneity of environmental conditions across the land region, the processor system operations comprising:

• applying a feature extraction process that extracts a set of sub-region environmental descriptors for each of the plurality of sub-regions;
• wherein the plurality of sub-regions comprise catchments that are irregularly-shaped;
• wherein the sub-region environmental descriptors comprise non-linear relationships with one another;
• wherein the set of sub-region environmental descriptors include dynamic descriptors that change over time, the dynamic descriptors comprising time series of daily values of the dynamic descriptors;
• applying a similarity analysis to the set of sub-region environmental descriptors to generate groups of the plurality of sub-regions;
• creating a plurality of group-based graphs by encoding each of the groups into a corresponding group-based graph;
• and
• using a spatio-temporal neural network to train a model based at least in part on the plurality of group-based graphs.