# A Compact Representation of Raster Time Series

## TL;DR

This paper introduces a novel compact data structure for raster time series that exploits temporal locality to reduce storage space while maintaining efficient query performance in GIS applications.

## Contribution

It presents a new approach that leverages temporal locality in raster time series to improve space efficiency without sacrificing query speed.

## Key findings

- Significant space reduction compared to naive methods
- Maintains competitive query times for various spatial queries
- Effectively exploits temporal locality in raster data

## Abstract

The raster model is widely used in Geographic Information Systems to represent data that vary continuously in space, such as temperatures, precipitations, elevation, among other spatial attributes. In applications like weather forecast systems, not just a single raster, but a sequence of rasters covering the same region at different timestamps, known as a raster time series, needs to be stored and queried. Compact data structures have proven successful to provide space-efficient representations of rasters with query capabilities. Hence, a naive approach to save space is to use such a representation for each raster in a time series. However, in this paper we show that it is possible to take advantage of the temporal locality that exists in a raster time series to reduce the space necessary to store it while keeping competitive query times for several types of queries.