ClimKrig - Ordinary Kriging of precipitation and air temperature records for Portuguese weather stations

This repository is part of the paper: Indicator-based assessment of post-fire recovery dynamics using satellite NDVI time-series (submitted to Ecological Indicators)
By Torres J., Gonçalves J., Marcos B. and Honrado J.

Description

This script reads a table with three initial columns holding the UID codes [1] and the XY point coordinates for weather stations [2:3] followed by a series of columns {4,...n} with annual records for air temperature (with _TMP_ in the middle of the column names) or precipitation (_PREC_) data and performs Ordinary Kriging interpolation for several different semi-variogram models, nugget values, ranges and estimation types (partial-sill is kept fixed and equal to variance).

The nugget component values to test can be set differently by temperature or precipitation variables. Model types include the Exponential, Spherical, Gaussian and Matern. Estimation types are Ordinary Least Squares (OLS) or Restricted Maximum Likelihood (REML).

10-fold cross-validation is used to assess performance and determine the best model, i.e., the combination maximizing the R². A map is produced and saved using the best combination for each variable. A predefined raster mask is used for setting the spatial resolution, CRS and extent of the interpolation process. In this case, we used a mask coincident with MODIS data at 250m, CRS: WGS 1984 UTM-29N for north Portugal.

Folder structure

• DATA - sample data directory

  • clim_data.csv: sample data including station IDs, XY coordinates and records for 2015 total annual precipitation and average annual temperature;
  • mask.tif: a raster layer use as a reference for interpolation.

• RES - interpolation results including:

  • kriging performance (10-fold CV),
  • 'best' variogram and
  • interpolated variables in raster format (GeoTIFF).