# High-resolution microclimatic grids for the Bohemian Forest Ecosystem based on in situ measurements

## TL;DR

This paper creates high-resolution temperature maps for a Central European forest using detailed measurements and models to better understand local climates.

## Contribution

The study introduces high-resolution microclimatic grids capturing canopy and near-ground temperatures in the Bohemian Forest Ecosystem.

## Key findings

- The developed grids have spatially cross-validated RMSE values from 0.41 °C to 2.34 °C for different temperature layers.
- The grids capture more spatial microclimatic variability compared to existing datasets like SoilTemp and ForestTemp.
- The method combines in situ measurements with LiDAR data and boosted spatial models for improved accuracy.

## Abstract

High-resolution microclimatic temperature grids are needed for ecosystem modelling, biodiversity conservation, and forest management. However, existing climatic grids are coarse and do not capture microclimatic temperatures beneath tree canopies or near the ground. To address this, we established a dense network of microclimatic loggers continuously measuring air, near-ground, and soil temperatures. We combined one year of these in situ temperature measurements with high-resolution LiDAR-derived land surface topography and forest structure using boosted spatial generalized additive models to develop 5-m resolution microclimatic grids for the largest forest wilderness area in Central Europe, the Bohemian Forest Ecosystem (BFE). The resulting grids provide reasonable estimates of local annual maximum, mean, and minimum temperatures and growing degree days at different heights with spatially cross-validated RMSE values ranging from 0.41 °C for annual mean soil temperature to 2.34 °C for maximum near-ground temperature. Compared to SoilTemp (soil temperature), ForestTemp (near-ground temperature), and downscaled ERA5-Land (air temperature), the BFE microclimatic grids provide the most accurate local temperature estimates and capture substantially more spatial microclimatic variability.