This is an Open Access document downloaded from ORCA, Cardiff University's institutional
repository: http://orca.cf.ac.uk/124261/
This is the author’s version of a work that was submitted to / accepted for publication.
Citation for final published version:
Chen, Shiuan-An, Michaelides, Katerina, Grieve, Stuart W. D. and Singer, Michael Bliss 2019.
Aridity is expressed in river topography globally. Nature 573 , pp. 573-577. 10.1038/s41586-019-
1558-8 file
Publishers page: http://dx.doi.org/10.1038/s41586-019-1558-8 <http://dx.doi.org/10.1038/s41586-
019-1558-8>
Please note:
Changes made as a result of publishing processes such as copy-editing, formatting and page
numbers may not be reflected in this version. For the definitive version of this publication, please
refer to the published source. You are advised to consult the publisher’s version if you wish to cite
this paper.
This version is being made available in accordance with publisher policies. See
http://orca.cf.ac.uk/policies.html for usage policies. Copyright and moral rights for publications
made available in ORCA are retained by the copyright holders.
Aridity is expressed in river topography globally 1
2
Shiuan-An Chen
1
*, Katerina Michaelides
1,2
, Stuart W. D. Grieve
3
and Michael Bliss Singer
2,4,5
3
4
1
School of Geographical Sciences, University of Bristol, Bristol, BS8 1SS, UK.
2
Earth Research Institute, 5
University of California Santa Barbara, Santa Barbara, California 91306, USA.
3
University College 6
London, London, WC1E 6BT, UK.
4
School of Earth and Ocean Sciences, Cardiff University, Cardiff, CF10 7
3AT, UK.
5
Water Research Institute, Cardiff University, Cardiff, CF10 3AX, UK. 8
*e-mail: sc16970@bristol.ac.uk 9
10
It has long been suggested that climate shapes land surface topography, through interactions between 11
rainfall, runoff, and erosion in drainage basins
1-4
. The longitudinal profile of a river (elevation versus 12
distance downstream) is a key morphological attribute that reflects the history of drainage basin 13
evolution, so its form should be diagnostic of the regional expression of climate and its interaction with 14
the land surface
5-9
. However, both detecting climatic signatures in longitudinal profiles and 15
deciphering the climatic mechanisms of their development have been challenging due to the lack of 16
relevant data across the globe, and due to the variable effects of tectonics, lithology, land-surface 17
properties, and humans
10,11
. Here we present a global dataset of river longitudinal profiles (n = 18
333,502), and use it to explore differences in overall profile shape (concavity) across climate zones. 19
We show that river profiles are systematically straighter with increasing aridity. Through simple 20
numerical modeling, we demonstrate that these global patterns in longitudinal profile shape can be 21
explained by hydrological controls that reflect rainfall-runoff regimes in different climate zones. The 22
most important of these is the downstream rate-of-change in streamflow independent of drainage 23
basin area. Our results illustrate that river topography inherits a signature of aridity, suggesting that 24
climate is a first-order control on drainage basin evolution. 25
Conventional theory presents river longitudinal profiles (long profiles) as having a generally concave-up 26
shape, with knickpoints and other fluctuations expressing the interactions of several independent variables: 27
climate, tectonics, lithology, and human impacts
11-13
. This characteristic shape of long profiles has been 28
interpreted to arise due to downstream flow increase with drainage area, which erodes the riverbed, 29
transports sediment from upstream to downstream, and produces fining profiles in bed material grain 30
size
13,14
. However, there are long profiles with overall concavity much closer to zero (straighter) than the 31
typical concave-up profile shape
15-17
, yet there is limited understanding of the global distribution of long 32
profile concavities and their relation to climate. Stream power incision theory states that channel erosion is 33
intrinsically tied to an assumed relationship between river discharge (Q) and drainage area (Q~A
c
). Based 34
on this theory, an expression has been derived that links supply-limited river long profile concavity to the 35
exponent c
18
, illustrating that profiles will be concave up for c > 0, straight for c = 0, and convex for c < 0, 36
and a similar dependency of profile concavity on the Q-A relationship has been derived for 37
transport-limited fluvial systems
19
. Previous work has largely emphasized long profile concavity for cases 38
where c > 0, despite evidence that c in many river basins, especially in drylands, may vary flood to flood 39
between negative, zero, and positive values
8,17,20
. Of particular interest here is to ascertain whether the 40
climatic expression within river channel hydrology may be a first-order control on long profile shape, and 41
whether its climatic signature is preserved across the globe. 42
A river experiences a cascade: from climate to hydrology to erosion, which evolves its long profile. 43
Therefore, the climatic expression within streamflow should be a first-order control on long profile shape
6-8
. 44
Numerical analysis of profile shape responses to a distribution of flow events above the threshold for 45
bedrock incision has demonstrated part of this dependency
5,8,21
. However, there is limited global evidence 46
of how the hydrologic expression of climate affects long profiles, across a wide range of climate zones. 47
Climate determines the precipitation regime within a region. In turn, the precipitation regime controls the 48
rate and frequency of water supply to the land surface, a proportion of which generates runoff over drainage 49
basins, subject to losses by infiltration and evapotranspiration. Flow in rivers occurs when runoff reaches 50
the channel, with notable baseflow contributions from groundwater and subsurface drainage in humid 51
regions and potential for prolonged periods of no flow in arid channels. The flow of water within a river is 52
a key driver of landscape evolution, through the corresponding downstream force exerted on the stream bed, 53
the associated channel erosion, and the expression of local river incision at each elevation position along 54
the long profile. Therefore, we propose that the climate-streamflow relationship exerts a strong control on 55
long profiles. 56
Cimate is expressed differently in the downstream rate-of-change in streamflow between arid and humid 57
endmember rivers. In arid climates, streamflow tends to decrease downstream in all but extreme floods
22
58
for two main reasons: 1) Low annual rainfall, limited areal coverage of rainstorms, and short duration of 59
rainfall events generates partial area runoff
23
. This results in a small proportion of basin tributaries 60
contributing streamflow to the mainstem for limited periods of time. 2) Rivers are typically ephemeral (no 61
permanent flow)
24
, so channels lose water through dry, porous beds (transmission losses
22
) because water 62
tables are well below the channel
25
. Thus, the commonly assumed power law relationship between 63
streamflow and drainage area (with positive exponent c) breaks down
20
such that the long-term average 64
value of c may be negative, positive, or zero. In contrast, humid channels have perennial flow (all year 65
round), supported by baseflow from groundwater, and they accumulate flow from adjoining tributaries, 66
producing downstream increases in discharge
13
(positive c). We intuit that there is a spectrum of prevailing 67
downstream changes in streamflow across the globe based on the regional expression of climate within 68
discharge regimes (e.g., dryland hydrology, mountain front orography
5
), rather than simply on drainage 69
basin area. Given the obvious link between streamflow and riverbed erosion, we hypothesize that climatic 70
signatures are imprinted within river long profiles, superimposed upon other exogenous controls. In other 71
words, we expect a great deal of scatter typical of environmental data, but we hypothesize that climate will 72
reveal itself as a first-order control on long profile shape. 73
To test this hypothesis, we produced a new and unprecedented database of Global Longitudinal Profiles
74
(GLoPro) of rivers between 60°N and 56°S (Fig.1) extracted from NASA’s 30-m Shuttle Radar 75
Topography Mission Digital Elevation Model (SRTM-DEM)
26
. The profiles were extracted using 76
LSDTopoTools
27
, software with advanced capabilities in topographic analysis, employing a conservative 77
threshold for upstream drainage area and an algorithm of downstream flow accumulation, both of which 78
reduce the likelihood of Type 1 errors (Methods). For each profile we computed the Normalized Concavity 79
Index (NCI), a metric computed based solely on profile geometry (Methods; Extended Data Fig.1) that 80
allows for standardized comparisons of river profile shapes across the globe. The NCI is negative if the 81
profile is concave-up, zero if the profile is straight, and positive if the profile is convex-up. 82
We categorized each profile in GLoPro using the Köppen-Geiger (K-G) climate classification
28
and the 83
quantitative Aridity Index (AI = Precipitation/Potential Evapotranspiration)
29
, to investigate relationships 84
between climate and river long profile shape and to test whether the expression of aridity is detectable in 85
NCI. K-G is based on temperature and precipitation thresholds, emphasizing vegetation response to climate. 86
AI is a scale that represents the balance between precipitation and evaporative demand, and it declines with 87
aridity. Here we addressed the null hypothesis that there are no differences in NCI between climate 88
categories. We did not censor GLoPro for any other natural or anthropogenic factors, and it includes both 89
bedrock and alluvial rivers. We do not make any assumptions about whether the profiles are steady-state 90
(equilibrium) or transient, but we assumed that climate categories in K-G and AI have not changed 91
significantly over the timescales of long profile development (Methods). 92
The global distribution of NCI values does not suggest any strong geographic biases, although there are 93
clear concentrations of convex (Southern Siberia), concave (SE Asia), and nearly straight (Arabian 94
peninsula) rivers (Fig.1). NCI distributions of different climate classes (Fig.2a) overlap and display great 95
breadth, reflecting the large sample size and the many interacting independent variables (climate, tectonics, 96
lithology, and human factors) that affect drainage basin development. Nevertheless, statistically significant 97
differences between distributions are evident (Extended Data Fig.5). Comparing the four main K-G climate 98
zones, all NCI distributions are negatively skewed, revealing that river long profiles are generally 99
concave-up (Fig.2a). However, compared to the other three main climate zones (Tropical, Temperate, and 100