A geospatial mapping pipeline for ecologists

Hoogen, Johan van den; Robmann,; Routh,; Lauber,; Tiel, van; Danylo,; Crowther, Thomas W.

doi:10.1101/2021.07.07.451145

Cited by 19 publications

(19 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To assess where our model is extrapolating -and thus possibly providing less reliable predictions -we calculated for each pixel the percentage of quantitative covariate layers for which the pixel value lies outside the range of data covered by the dataset. Finally, we used a spatial leave-oneout cross-validation analysis to test the effect of spatial autocorrelation in the dataset (Figure S5; Roberts et al, 2017;van den Hoogen et al, 2021). This approach each time validates a model on data from one distinct location and trains a model on the remaining data.…”

Section: Geospatial Modellingmentioning

confidence: 99%

ForestTemp – Sub‐canopy microclimate temperatures of European forests

Haesen

Lembrechts

Frenne

et al. 2021

Global Change Biology

Self Cite

View full text Add to dashboard Cite

Ecological research heavily relies on coarse-gridded climate data based on standardized temperature measurements recorded at 2 m height in open landscapes. However, many organisms experience environmental conditions that differ substantially from those captured by these macroclimatic (i.e. free air) temperature grids. In forests, the tree canopy functions as a thermal insulator and buffers sub-canopy microclimatic conditions, thereby affecting biological and ecological processes. To improve the assessment of climatic conditions and climate-change-related impacts on forest-floor biodiversity and functioning, high-resolution temperature grids reflecting forest microclimates are thus urgently needed. Combining more than 1200 time series of in situ near-surface forest temperature with topographical, biological and macroclimatic variables in a machine learning model, we predicted the mean monthly offset between sub-canopy temperature at 15 cm above the surface and free-air temperature over the period 2000-2020 at a spatial resolution of 25 m across Europe. This offset was used to evaluate the difference between microclimate and macroclimate across space and seasons and finally enabled us to calculate mean annual and monthly temperatures for European forest understories. We found that sub-canopy air temperatures differ substantially from free-air temperatures, being on average 2.1°C (standard deviation ± 1.6°C) lower in summer and 2.0°C higher (±0.7°C) in winter across Europe. Additionally, our high-resolution maps expose considerable microclimatic variation within landscapes, not captured by the gridded macroclimatic products. The provided forest sub-canopy temperature maps will enable future research to model below-canopy biological processes and patterns, as well as species distributions more accurately.

show abstract

Section: Geospatial Modellingmentioning

confidence: 99%

ForestTemp – Sub‐canopy microclimate temperatures of European forests

Haesen

Lembrechts

Frenne

et al. 2021

Global Change Biology

Self Cite

View full text Add to dashboard Cite

show abstract

“…To generate global maps of monthly temperature offsets (Figure 2 ), we trained Random Forest (RF) models for each month, using the temperature offsets as the response variables and the global variable layers as predictors (Breiman, 2001 ; Hengl et al, 2018 ). We used a geospatial RF modelling pipeline as developed by van den Hoogen et al ( 2021 ). RF models are machine learning models that combine many classification trees using randomized subsets of the data, with each tree iteratively dividing data into groups of most closely related data points (Hengl et al, 2018 ).…”

Section: Methodsmentioning

confidence: 99%

“…This variable importance adds up the decreases in the impurity criterion (i.e. the measure on which the local optimal condition is chosen) at each split of a node for each individual variable over all trees in the forest (van den Hoogen et al, 2021 ).…”

Section: Methodsmentioning

confidence: 99%

“…To assess the uncertainty in the monthly models, we performed a stratified bootstrapping procedure, with total size of the bootstrap samples equal to the original training data (van den Hoogen et al, 2021 ). Using biomes as a stratification category, we ensured the samples included in each of the bootstrap training collections were proportionally representative of each biome's total area.…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Global maps of soil temperature

Lembrechts

Hoogen

Aalto

et al. 2022

Global Change Biology

Self Cite

163

View full text Add to dashboard Cite

Research in global change ecology relies heavily on global climatic grids derived from estimates of air temperature in open areas at around 2 m above the ground. These climatic grids thus fail to reflect conditions below vegetation canopies and near the ground surface, where critical ecosystem functions are controlled and most terrestrial species reside. Here we provide global maps of soil temperature and bioclimatic variables at a 1-km² resolution for 0-5 and 5-15 cm depth. These maps were created by calculating the difference (i.e., offset) between in-situ soil temperature measurements, based on time series from over 1200 1-km² pixels (summarized from 8500 unique temperature sensors) across all of the world's major terrestrial biomes, and coarse-grained air temperature estimates from ERA5-Land (an atmospheric reanalysis by the European Centre for Medium-Range Weather Forecasts). We show that mean annual soil temperature differs markedly from the corresponding 2 m gridded air temperature, by up to 10°C (mean = 3.0 ± 2.1°C), with substantial variation across biomes and seasons. Over the year, soils in cold and/or dry biomes are substantially warmer (3.6 ± 2.3°C warmer than gridded air temperature), whereas soils in warm and humid environments are on average slightly cooler (0.7 ± 2.3°C cooler). The observed substantial and biome-specific offsets underpin that the projected impacts of climate and climate change on biodiversity and ecosystem functioning are inaccurately assessed when air rather than soil temperature is used, especially in cold environments. The global soil-related bioclimatic variables provided here are an important step forward for any application in ecology and related disciplines. Nevertheless, we highlight the need to fill remaining global gaps by collecting more in-situ measurements of microclimate conditions to further enhance the spatiotemporal resolution of global soil temperature products for ecological applications.

show abstract

“…, are probably among the least troubling in comparison to other continents. Our methodological framework could potentially be implemented at a global scale, and 26/35possibly through Google Earth Engine (GEE)(van den Hoogen et al, 2021) or through the European Space Agency's OpenEO platform (https://openeo.cloud/) to produce high resolution(10-30 m) …”

mentioning

confidence: 99%

Forest tree species distribution for Europe 2000–2020: mapping potential and realized distributions using spatiotemporal Machine Learning

Bonannella¹,

Hengl²,

Heisig

et al. 2022

Preprint

View full text Add to dashboard Cite

This paper describes a data-driven framework based on spatiotemporal machine learning to produce distribution maps for 16 tree species (Abies alba Mill., Castanea sativa Mill., Corylus avellana L., Fagus sylvatica L., Olea europaea L., Picea abies L. H. Karst., Pinus halepensis Mill., Pinus nigra J. F. Arnold, Pinus pinea L., Pinus sylvestris L., Prunus avium L., Quercus cerris L., Quercus ilex L., Quercus robur L., Quercus suber L. and Salix caprea L.) at high spatial resolution (30 m). Tree occurrence data for a total of 3 million of points was used to train different algorithms: random forest, gradient-boosted trees, generalized linear models, k-nearest neighbors, CART and an artificial neural network. A stack of 305 coarse and high resolution covariates representing spectral reflectance, different biophysical conditions and biotic competition was used as predictors for realized distributions, while potential distribution was modelled with environmental predictors only. Logloss and computing time were used to select the three best algorithms to tune and train an ensemble model based on stacking with a logistic regressor as a meta-learner. An ensemble model was trained for each species: probability and model uncertainty maps of realized distribution were produced for each species using a time window of 4 years for a total of 6 distribution maps per species, while for potential distributions only one map per species was produced. Results of spatial cross validation show that the ensemble model consistently outperformed or performed as good as the best individual model in both potential and realized distribution tasks, with potential distribution models achieving higher predictive performances (TSS = 0.898, R2 logloss = 0.857) than realized distribution ones on average (TSS = 0.874, R 2 logloss = 0.839). Ensemble models for Q. suber achieved the best performances in both potential (TSS = 0.968, R2 logloss = 0.952) and realized (TSS = 0.959, R2 logloss = 0.949) distribution, while P. sylvestris (TSS = 0.731, 0.785, R2 logloss = 0.585, 0.670, respectively, for potential and realized distribution) and P. nigra (TSS = 0.658, 0.686, R2 logloss = 0.623, 0.664) achieved the worst. Importance of predictor variables differed across species and models, with the green band for summer and the Normalized Difference Vegetation Index (NDVI) for fall for realized distribution and the diffuse irradiation and precipitation of the driest quarter (BIO17) being the most frequent and important for potential distribution. On average, fine-resolution models outperformed coarse resolution models (250 m) for realized distribution (TSS = +6.5%, R2 logloss = +7.5%). The framework shows how combining continuous and consistent Earth Observation time series data with state of the art machine learning can be used to derive dynamic distribution maps. The produced predictions can be used to quantify temporal trends of potential forest degradation and species composition change.

show abstract

A geospatial mapping pipeline for ecologists

Cited by 19 publications

References 17 publications

ForestTemp – Sub‐canopy microclimate temperatures of European forests

ForestTemp – Sub‐canopy microclimate temperatures of European forests

Global maps of soil temperature

Forest tree species distribution for Europe 2000–2020: mapping potential and realized distributions using spatiotemporal Machine Learning

Contact Info

Product

Resources

About