Big Data Processing Methods in GIS

The article discusses methods for processing big data in geographic information systems (GIS) with an emphasis on the use of recurrent neural networks (RNN) for forecasting geospatial processes. Modern approaches are described, including distributed computing on clusters (Hadoop, Spark) and cloud platforms (Google Earth Engine), providing efficient processing of spatial data. Particular attention is paid to RNN architectures, such as LSTM, their application in temporal forecasting problems (weather, transport, land use) and comparison with traditional methods. The article provides a numerical example illustrating the use of RNN for time series forecasting, with an accuracy analysis and visualization of the results.