Some techniques for indexing and analysis of spatiotemporal data, as seen in some sources.
Spatiotemporal Data Indexing Technology
HR-tree
HR-tree stores an independent R-tree for each timestamp. After that, between two consecutive R-trees, if the same node is used, only one node is reserved to improve utilization. The query efficiency is higher.3DR-tree
3DR-tree creates a spatiotemporal index based on the R tree, and regards time information as another dimension of general space. Two-dimensional space objects are represented by two-dimensional space enclosing rectangles, and three-dimensional space-time objects are represented by the smallest enclosing rectangular cylinder of three-dimensional space. .The Q+R tree
is composed of two trees, an R * tree and a quad tree. Using the R* tree to index stationary objects and the quad tree to index moving objects, it can index relatively stationary objects and fast-moving objects.
spatiotemporal analysis technology
spatiotemporal change detection
Changes are generally interpreted as changes in spatial statistics over time. There are the following methods, geometric center, semivariation coefficient, spatial regression coefficient,
spatiotemporal pattern recognition
It refers to the spatiotemporal regularity of the attributes of things, and the main methods include SOM spatiotemporal clustering, EOF spatiotemporal decomposition, and various hot spot detection methods.
space-time regression
The purpose of spatiotemporal regression is to find the relationship between variables, which is generally extended on the classic regression models, including spatiotemporal panel model, spatiotemporal BHM, Bayesian network model, spatiotemporal T-GWR, spatiotemporal GAM, etc.
Modeling of spatiotemporal processes
It is somewhat similar to the modeling of infectious diseases, social relationship analysis, etc. There are cellular automata model, agent model, reflection diffusion equation, etc.
spatiotemporal evolutionary tree
Space-time data is the product of space-time process, and data defined artificially in one-dimensional, two-dimensional or high-dimensional coordinate systems may not be able to fully express the evolution process. Dimensional constraints. The core is: individual state changes form the evolution path of the state space, and the evolution paths of multiple individuals generate the hierarchical structure of the state space, which is described by state variables. The specific idea is: determine the state variable -> determine the state space ( Tree structure) --> project the spatiotemporal data of attribute variables into the state space --> individual evolution paths --> summarize the evolution laws of different types of groups --> individual states develop, grow, evolve, and compile along the evolution path, according to which Perform state prediction and analysis.
The data structures mentioned above will be analyzed in more detail later. (R tree, etc.)
This work is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 China Mainland License Agreement .