Weather station networks, reinvented

We need a different kind of weather station networks to answer most ecological questions. We argue that what we need is countrywide ‘microclimate networks’, that measure weather conditions there where they matter for ecology and nature around us.

Where weather stations are insufficient

To measure global weather and climate, the world is increasingly covered by a network of weather or meteorological stations. On land, these stations are designed in such a way that it ensures all stations, whether in Belgium or Brazil, are recording climate in the same standardized manner: in open landscapes, above short grass, and well away from trees, buildings or mountains. While macro-meteorologists, those who design these networks, did their best to remove these local sources of “noise” in the data, these sources of “noise” are meaningful to many organisms. For example, although trees may cause chaos for meteorologists, because forest cover interrupts and changes local temperatures, these differences in forest cover between for example a beech forest in Flanders and a dense tropical forest in the Amazon rainforest of South America have huge meaning and importance for the animals and plants that live there. As a result, our current weather station network is ill-equipped to provide meaningful data to scientists wanting to know how climate impacts biodiversity. 

To do: build countrywide microclimate networks

A call to action has been made in a recently published study, which calls for a globally coordinated effort to create a new kind of weather station network – one that can tell us what a small lizard is ‘feeling’ in the remotest of habitats. We strongly believe that the world needs microclimate networks in parallel to the existing countrywide weather station network as established by national meteorological institutes, such as the RMI in Belgium. We also believe that we need to be quick about it. In their recent publication, the ecologists stress their case. Actually, we need two things: mini weather stations that measure conditions close to and in the soil, and a network of these sensors that measures in the most relevant environments. From cities to the countryside, from forests to mountain slopes, all those locations traditionally avoided by weather stations should be preferentially sampled by these new microclimate networks. By implementing this kind of network, we will be better equipped to understand climate change impacts on species, ecosystems and our agricultural systems.

A ‘mini weather station’ in action, measuring temperature and soil moisture in a Flemish broadleaf forest.

Selection algorithm

To achieve these countrywide microclimate networks, we provide a handy algorithm to decide where exactly these new networks should come. It selects measurement locations based on the expected microclimate variability of a region, by selecting locations in as broad a range of landscape types as possible. For this, we use a multivariate analysis of the environmental space, based on characteristics from which we know that they matter for microclimate: topography, vegetation cover, urbanity, macroclimate. The algorithm quantifies the distribution of these habitats across the country or region and proposes a set of locations that maximally covers this variability with as few measurement locations as possible.

Next up? Implementation! We hope that governments, scientists  or anybody else pay heat to our plea for microclimate networks and use our proposed selection tool to implement them in their own region. As such, we reach out to anybody else to start thinking on this parallel network of microclimate sensors to complement the global weather station network.

Our selection algorithm exemplified for France, with 453 suggested measurement locations on the left, and distribution of these locations (red dots) in the environmental space of France on the right, exemplified for elevation, the Topographic Roughness Index (the complexity of the terrain), and FAPAR (the amount of light absorbed by plants, a proxy of how green the area is).


With the global SoilTemp-network, we made a great start in that regard, bringing together microclimate data from over 13.000 sensors from over 60 countries. The algorithm proposed here can however help us to switch gears even more, and design these networks specifically with microclimate measuring and global and regional coverage of microhabitats in mind.

We also already have a first example to showcase their approach: the new citizen science project ‘Nosy parkers in the garden’ ( has used it to select gardens for their project on heat and drought impacts on lawns. In this large citizen science project, 4400 citizens were asked to install a mini weather station (‘the garden dagger’) in their garden to measure the impact of heat and drought on urban environments.

A microclimate sensor (little white dot in the grass) taking the heat of a forest clearing

Link to the publication in Global Ecology & Biogeography:

This entry was posted in Science and tagged , , , , , , , . Bookmark the permalink.

1 Response to Weather station networks, reinvented

  1. Pingback: And we are live! | The 3D lab

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s