The Earth’s energy and water cycles govern the distribution and movement of energy and water in the atmosphere, oceans and land. Both energy and water are constantly being transported between different regions of the globe, and the two cycles are closely linked (‘coupled’) together. It is very important to determine the size and variability of these transports as well as the long-term average distributions of each quantity. This will enable us to detect trends due to climate change, as well as to provide insight into short term climatic variability such as the El Niño Southern Oscillation.
The Sun provides energy to the Earth via solar radiation; simultaneously, some energy is emitted back into space. It is therefore quite common to consider the energy imbalance at the top-of-atmosphere (TOA). There is currently a small downward imbalance which means the Earth is gaining heat. The incoming radiation is distributed between the atmosphere, land and oceans both vertically and horizontally. Water is transported around the globe by (e.g.) ocean currents, rivers, evaporation, and precipitation. The last two processes provide the crucial link between the energy and water cycles. When water changes state from vapour to liquid or vice versa it either releases or absorbs energy in quantities which are significant enough that they must be included in the energy cycle.
Our aim is to combine a variety of Earth observation (EO) data sets in order to determine the energy and water cycles with as much precision as possible. This work was pioneered by the NASA NEWS team (L’Ecuyer et al. 2015, Rodell et al. 2015) and as a first step we have reproduced their work. The Earth is divided into 16 regions (seven land and nine ocean) as shown in Figure 1. In each region we consider the energy and water content both in the atmosphere and at the surface, as well as the TOA radiation imbalance (no water is lost to space).
Figure 1: The regions used in this study.
Several complementary EO measurements of energy and water transport are combined in each of these regions. These measurements are expressed as fluxes (the rate of flow of energy through an area). The larger the flux in a particular region, the more energy is being added to (or removed from) that region. Both vertical and horizontal fluxes are exploited in this method. Figure 2 shows the initial net vertical energy flux at the Earth’s surface in each of the 16 regions. The initial net flux is obtained by simply adding the downward fluxes together and subtracting the upward fluxes from the total. The values shown in the figure are the average of about a decade’s worth of observations.
Figure 2: Initial values of net surface flux in each region. Positive values indicate heat gain from the atmosphere and negative values indicate heat loss to the atmosphere.
Now, we know that there are various physical balances that should be respected (such as conservation of energy). The measured fluxes don’t satisfy these balances so we have to modify each flux by adding or subtracting a particular amount to it. When doing this it is particularly important to consider the uncertainty on each flux, which quantifies how well we (think we) have measured it. Observations that are poorly measured have large uncertainties and will be allowed to move more than those that are well known. For example, the TOA radiation has been extremely precisely determined, so has a small uncertainty, but the horizontal movement of water in the atmosphere is poorly understood and has a large uncertainty. The latter can therefore be moved relatively further from its initial value when trying to ensure all the physical balances are respected.
The net surface fluxes obtained in this procedure are reproduced in Figure 3. Some features of the solution indicate that it may be possible to make improvements. For example, the North Atlantic (region 13 in Figure 1) is losing less heat than may be expected. A large amount of heat is transported northwards from the South Atlantic by a process known as the Atlantic Meridional Overturning Circulation (see e.g. Buckley and Marshall 2016). This sort of discrepancy has motivated us to look into ways to incorporate extra information into the solution and hopefully mitigate the problems. One method which seems promising is to account for large-scale correlations in satellite measurements of heat loss from the ocean to the atmosphere (technically known as latent and sensible heat fluxes). If the flux measurements over the oceans are positively correlated then they are more likely to all be measured too high or low at once. This could arise due to the data processing methods used to determine the fluxes from satellite data, for example.
Figure 3: Output values of net surface flux in each region. Positive values indicate heat gain from the atmosphere and negative values indicate heat loss to the atmosphere.
We have estimated the heat flux correlations over the oceans and if we include them in the fit we find that more heat is transported from the South to the North Atlantic, which consequently loses more energy to the atmosphere. This is a promising result and we are currently preparing a paper to describe it more fully. We also intend to explore additional possibilities such as using additional data sets, including reanalysis constraints, and enhancing the spatial resolution of the solution by dividing the globe into more regions. In the long term we have ambitions to include the carbon cycle alongside the energy and water cycles.