Individuals and political organizations alike often define “climate” as climate scientist John Kennedy did on Twitter: “Practically speaking: weather’s how you choose an outfit, climate’s how you choose your wardrobe.” Meanwhile, the scientific literature rarely defines climate more specifically than as the “statistics of weather.” Such vague definitions may impede policy making on climate change because politics is best at tackling well-defined problems. Without a clear definition of what climate is, local and global perspectives on climate change contrast with each other, presenting a problem in multilateral negotiations [Miller, 2004] and reducing the relevance of discussions on climate and its changes to local communities [Hulme, 2016]. The lack of a robust definition may even prevent agreement in discussions on climate change policies.
The concepts of climate and weather developed over centuries in various cultural and geographic settings [Hulme, 2016]. The resulting plurality of views implies that communication about climate involves cultural and linguistic translations [Rudiak-Gould, 2012]. We can observe such translations (or their failure) every day in the media’s reception of climate science literature and in how climate scientists adopt results from colleagues with different backgrounds.
A Classification of Convenience
Scientists’ definitions of climate have evolved with their understanding of the world [Heymann and Achermann, 2018]. Possibly the most common current definition is that from the Intergovernmental Panel on Climate Change’s (IPCC) Fifth Assessment Report, which distinguishes between “climate in a narrow sense…as the average weather…over a period of time ranging from months to…millions of years” and “climate in a wider sense” as “the state…of the climate system.” Both senses are valid descriptions of climate. They are instances of a common template, that is, a common concept of climate. Other definitions differ from the IPCC’s in the parts of the climate system and the methods that they consider [see Conradie, 2015]. Werndl  asserts, “How to define climate and climate change is nontrivial and contentious.”
The idea of climate is a classification of convenience. It is a tool that helps us deal with our ever-changing surroundings [Hulme, 2016] in colloquial, scientific, philosophical, and political contexts. Then, climate is an inherently subjective concept [Lucarini, 2002]. Any societal group, any person, any business, and any academic may have a distinct view on climate. This view depends on the specific application, the regional and temporal focus, and the person’s or group’s unique experiences. Any societal actor’s view is one of a myriad of possible instances of climate. The scientific literature on climate largely imposes a North American and European perspective (compare Rudiak-Gould ). But no single instance can define climate; rather, we need a template definition for the concept.
If we accept that the view of climate as the “statistics of weather” is unclear, we may clarify this view using a more precise definition of our terms. What do these statistics of weather usually represent? What constitutes the “weather” whose statistics we want to consider? That is, we have to spell out what is climate and what is weather. Indeed, the transition between weather and climate is ambiguous, with gaps between these concepts in terms of their coverage and their temporal delineation.
Earth’s climate system includes its atmosphere, hydrosphere, cryosphere, biosphere, and lithosphere. Besides statistics, descriptions of the climate system’s behavior may employ thermodynamics and fluid dynamics, along with chemistry, electromagnetism, rheology, and plasma physics.
The idea of climate allows us to compare parts of the system in different locations, at different times, and with different sources of data. We can consider spatial scales ranging from local to global at the surface, in the upper atmosphere, or in the ocean. And we may classify climate over time periods of millions of years, three well-observed decades, or even shorter periods to highlight short-term variations. Each data point in studying climate is an uncertain, imperfect estimate. Only by considering uncertainty can we make practical use of models of climate, which are uncertain by construction.
We speak of climate change when, according to a certain criterion, we detect differences between statistics of weather for different instances of interest, such as different time periods. Conversely, “climate variability” refers to variations within the reference period for one instance. Even a gradual change of the period of interest leads to another instance of climate. Timescales of climate depend on the physical properties of the system and differ between considered systems. That is, the climate system of another celestial body has different properties than Earth’s climate system. The shortest climate timescale that signifies the transition between weather and climate has to allow defining statistics.
If these are the simplified dimensions and components of what we commonly consider climate, what defines the weather whose statistics produce the climate? And if the smallest climate timescale on Earth is monthly, what are the timescales of this weather?
States of an Atmosphere
Weather is more tangible than climate; we experience it on a day-to-day basis, and we can point to it [Hulme, 2016]. Our perceptions of weather may extend to phenomena beyond the atmosphere. If I am on a boat, wave height is a relevant part of my weather observations. But is desert dust collecting on your car part of weather? Is the transport of industrial emissions weather? Do you perceive the flooding of a field as part of weather?
The American Meteorological Society defines weather as the “state of the atmosphere” at a specific time and emphasizes the minutes-to-days scale of its variations and how these variations influence life or a celestial body. The state of a system describes the system in all its parts of interest. Thus, weather completely describes the atmosphere in the sense that the concept of weather encompasses everything needed for a description.
Weather service forecast models distinguish between external (prescribed) properties and internal (predicted) properties. External properties are slowly varying aspects like land use and the surface temperature of the ice-free ocean that provide boundary conditions for the model run. Internal properties are generally fast-changing variables, such as the properties of snow on the surface or the land surface temperature, that directly influence or are influenced by local weather; these internal model variables may be external to the atmosphere.
External properties are necessary to describe the state of the atmosphere because they are part of weather. If climate is the “statistics of weather,” one cannot restrict weather to the atmosphere but must allow weather to extend beyond the atmosphere. For example, the biosphere, cryosphere, and hydrosphere all influence the atmosphere through fluxes of energy and moisture.
If weather on Earth takes place over minutes to days and the shortest climate timescale is monthly, we need a conceptual definition of the transition between weather and climate. The IPCC separates climate variability from the scales of “individual weather events.” This view originates in the classical scale diagrams of meteorology, which plot typical spatial scales against typical temporal scales [e.g., Orlanski, 1975].
Using this view, we can take the transition as the scale at which the predictability of weather reaches its limit [e.g., Lovejoy, 2013; E. N. Lorenz, unpublished manuscript]. In this view, the length of time over which the initial conditions dominate limits the timescales of weather. Beyond this time limit, the sum of our experiences and the resulting expectation of the typical weather describe the system more reliably than a deterministic forecast starting from our experience of the current state of the system.
Above I tried to clarify the temporal transition between weather and climate and to highlight which parts of a planetary system weather and climate describe. This does not fully remove the ambiguity of the classical “climate is the statistics of weather” definition. Contrasts between local, regional, and global instances of climate, its change, and its impacts [Clark, 1985] likely will become even more important with the emergence of plans for climate engineering applications. These applications require even more that discussions are specific about the climate instances of interest. Acknowledging the inherent subjectivity in current definitions of climate may pave the way to a less contentious definition.
Then again, maybe it is not possible to conclusively delineate weather and climate—at least their temporal separations and terminological dependence. The differing origins of the terms may prevent such a distinction [e.g., Edwards, 2010]. If this is the case, then there may always be the traditional term “weather” and an evolving understanding of what climate is. The colloquial view would contrast subjective expectations and experiences, whereas the Earth sciences would use weather as the concept of meteorology and climate as a description of the interplay between the various Earth system components.
However one applies the term “climate,” it was, is, and will be an evolving concept [Heymann and Achermann, 2018]. It grows as our understanding of the Earth system and of the factors influencing it and its processes—including weather—grows larger.
Comments by anonymous colleagues helped to improve the manuscript. I want to thank the referee of this article and the four anonymous referees of an earlier version as well as the editors for their valuable comments and their support of this article.