SPEI-based hierarchical clustering for climate zonation

Knowledge graph centered on SPEI-based hierarchical clustering for climate zonation with 12 nodes and 28 connections. Top connected: Crested Butte, Engelmann Spruce, Salix lucida subsp. lasiandra, Abies lasiocarpa, Statistical framework to assess long-term spatio-t.

Method synopsis

Novel application of hierarchical k-means clustering to cumulative distribution curves of Standard Precipitation-Evapotranspiration Index to identify climate zones. Uses temporal segmentation to detect changing zonation patterns over time.

Synthesized from method descriptions across 1 paper using this protocol.

Procedure as described in the canonical source

Steps below were extracted from the paper that introduces this protocol — Statistical framework to assess long-term spatio-temporal climate changes: East River mountainous watershed case study (2022), Stochastic environmental research and risk assessment. Implementations in other papers (listed below) may differ.

1. Perform hierarchical clustering analysis
   Conduct hierarchical clustering using Hartigan and Wong algorithm with application of hkmeans function in the R library factoextra. Apply clustering to cumulative distribution curves of SPEI for each of the three segments shown in Figure 13. Determine optimal number of clusters using the elbow method with the function fviz_nbclust of the R package factoextra. Results show optimal number of clusters is 5.
   Quantities: Clustering applied to 17 locations across 3 temporal segments with optimal 5 clustersDuration: Applied to segmented time series dataConditions: Standard clustering analysis conditions
   Equipment: R library factoextra, hkmeans function, fviz_nbclust function, Hartigan and Wong algorithm
2. Generate spatial zonation maps
   Conduct mapping using application of R libraries ggmap, ggrepel, and ggplot2 for visualization of results. Create hierarchical clustering trees-dendrograms and 2D PCA maps, and zonation maps for periods before and after the time series breaks. Map spatial distribution of meteorological stations showing different clustering patterns for the three temporal segments.
   Quantities: Maps created for all 17 meteorological stations across 3 temporal segmentsDuration: Applied to final clustering resultsConditions: Standard GIS mapping conditions
   Equipment: R library ggmap, R library ggrepel, R library ggplot2

Typical Equipment

• R library factoextra
• hkmeans function
• fviz_nbclust function
• Hartigan and Wong algorithm

Output Measurements

• cluster assignments
• climate zones
• temporal zonation changes

Papers Using This Protocol (1)

Species Studied With This Protocol

Places Where Used