Local biomass burning is a dominant cause of the observed precipitation reduction in southern Africa

Abstract

Observations indicate a precipitation decline over large parts of southern Africa since the 1950s. Concurrently, atmospheric concentrations of greenhouse gases and aerosols have increased due to anthropogenic activities. Here we show that local black carbon and organic carbon aerosol emissions from biomass burning activities are a main cause of the observed decline in southern African dry season precipitation over the last century. Near the main biomass burning regions, global and regional modelling indicates precipitation decreases of 20-30%, with large spatial variability. Increasing global CO₂ concentrations further contribute to precipitation reductions, somewhat less in magnitude but covering a larger area. Whereas precipitation changes from increased CO₂ are driven by large-scale circulation changes, the increase in biomass burning aerosols causes local drying of the atmosphere. This study illustrates that reducing local biomass burning aerosol emissions may be a useful way to mitigate reduced rainfall in the region.

ARTICLE

Received 10 Aug 2015 | Accepted 4 Mar 2016 | Published 12 Apr 2016

Observations indicate a precipitation decline over large parts of southern Africa since the 1950s. Concurrently, atmospheric concentrations of greenhouse gases and aerosols have increased due to anthropogenic activities. Here we show that local black carbon and organic carbon aerosol emissions from biomass burning activities are a main cause of the observed decline in southern African dry season precipitation over the last century. Near the main biomass burning regions, global and regional modelling indicates precipitation decreases of 2030%, with large spatial variability. Increasing global CO2 concentrations further contribute to precipitation reductions, somewhat less in magnitude but covering a larger area. Whereas precipitation changes from increased CO2 are driven by large-scale circulation changes, the increase in biomass burning aerosols causes local drying of the atmosphere. This study illustrates that reducing local biomass burning aerosol emissions may be a useful way to mitigate reduced rainfall in the region.

DOI: 10.1038/ncomms11236 OPEN

Local biomass burning is a dominant cause of the observed precipitation reduction in southern Africa

ivind Hodnebrog1, Gunnar Myhre1, Piers M. Forster2, Jana Sillmann1 & Bjrn H. Samset1

1 Center for International Climate and Environmental Research-Oslo (CICERO), P.O. Box 1129 Blindern, N-0318 Oslo, Norway. 2 School of Earth and Environment, University of Leeds, Leeds LS2 9JT, UK. Correspondence and requests for materials should be addressed to .H.(email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 7:11236 | DOI: 10.1038/ncomms11236 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11236

Black carbon (BC) aerosols, in the atmosphere, absorb solar radiation and may affect the hydrological cycle through changes in the atmospheric and surface energy budgets15.

In contrast to most greenhouse gases, BC is removed from the atmosphere only after a few days6,7. Absorbing aerosols such as biomass burning BC modify cloud distribution, mainly by suppressing convective activity, albeit with some debate existing in the literature812. On a global scale, the altitude of added absorbing aerosol concentration is shown to inuence the magnitude and sign of global average precipitation changes, with a net decrease in precipitation expected unless BC is added close to the surface3,1317. Organic carbon (OC) aerosols are emitted together with BC and have approximately the same lifetime, but OC scatters sunlight to a much larger extent than BC, and therefore cools the atmosphere-surface system6. Biomass burning is a major source of BC and OC aerosols and normally takes place during the dry season, which is therefore the focus of this study.

Observational data reveal changes in precipitation patterns since the middle of the past century, with some regions in the world getting increased precipitation while other regions are experiencing less precipitation18,19. In general, increases in anthropogenic emissions have led to more precipitation on a global scale due to surface temperature warming20. However, the detailed effects of changes in greenhouse gases and aerosols on the hydrological cycle are complex2,21. Precipitation responses depend on the climate forcing mechanism, and also differ with timescale15,16,22. Responses in precipitation often show strong regional differences. This is mainly due to large-scale changes in circulation patterns and land-ocean contrasts, but could also arise from strictly local interactions, which are often caused by constituents that have short residence times in the atmosphere, such as BC.

Southern Africa is a region where observations indicate reduced precipitation and increased dryness since the 1950s (refs 18,19). The region is highly vulnerable to climate change (see ref. 23 and references therein), and has strong connections between climate and water, energy and food security24. Global climate models predict a further decrease in precipitation and a strong increase in the number of consecutive dry days near the end of the century for a business-as-usual scenario, due to widening of the Hadley Circulation25. While previous work on observed precipitation has focused mostly on annual trends, the focus in this study is on the dry season during which emissions from biomass burning are particularly strong. Although long-term observational data are sparse in the region, previous studies have shown evidence of declining trends in annual precipitation for several stations. In the latter half of the 20th century, signicant declining precipitation trends have been shown in Zambia and southwest Tanzania, in addition to a more widespread increase in the annual maximum of consecutive dry days26. Annual precipitation has also decreased in tropical rainforest regions, such as in Congo, the Democratic Republic of the Congo and Gabon27,28.

The present study focuses on explaining causes of observed precipitation decline during the dry season in southern Africa. We use a global precipitation observation data set (Global Precipitation Climatology Centre (GPCC)29,30) and a global (CESM1.0.4/CAM4 (refs 31,32)) and regional climate model (Weather Research and Forecasting (WRF)33). Our results show that the combination of BC/OC from local anthropogenic biomass burning activities, and increases in the global CO2 concentration, can explain the precipitation decline in the region. These ndings are further supported by analysis of the regional energy budget, which shows a local drying of the atmosphere from added biomass burning aerosols.

ResultsSpatial distributions of precipitation changes. Some of the largest BC and OC emissions in the world take place in southern Africa due to extensive biomass burning during the dry season34,35, approximately from June to September each year. Due to human activities, BC biomass burning emissions in the region almost doubled during the 20th century, with the majority of the increase in the latter half35 (Supplementary Fig. 1), leading to a 60% increase in the BC burden36 (Table 1). The increase in OC burden over the period is about 70% (Table 1).

In the present study, we show the modelled precipitation impact of increases in local BC and OC emissions due to biomass burning over the past one-and-a-half centuries, and compare them with the possible inuence of increases in atmospheric CO2 concentrations over the same period. We quantify and interpret the results from global and regional model results in terms of the regional energy budget37 to uncover the mechanisms behind modelled changes in precipitation.

Throughout the manuscript, we show differences between model experiments (see Table 1) and denote BB as the impact of changes in biomass burning emissions of BC and OC from 1850 to 2000 (experiments BASE BB1850), CO2 as the impact of

changes in CO2 volume mixing ratios over the same time period (BASE CO21850) and CO2BB as the combined impact of

BC/OC biomass burning and CO2 concentration changes (BASE CO2BB1850). Due to large variability in the

precipitation changes, ve-member ensemble runs were performed for each scenario.

Observations over southern Africa in the past century indicate a precipitation decline during the dry season from June to September (Fig. 1a,b), with a stronger decline in the latter part of the period, coinciding with the period of strongly increasing biomass burning emissions (Supplementary Fig. 1b).

A strong impact of anthropogenic biomass burning particles on precipitation during the dry season can be seen in the regional and the global climate model (Fig. 1c,d; Supplementary Fig. 2c,d), with precipitation decreases in the order of 1040% in large parts of the region. In both models the strongest decline in absolute precipitation occurs in the northwest, which also is the region with the largest increase in atmospheric biomass burning aerosol

Table 1 | Overview of model experiments.

Experiment* Short description Global CO2 vmr (p.p.m.) BC burdenw (mg m 2) OC burdenw (mg m 2) BASE Year 2000 conditions 367 1.85 11.8BB1850 Biomass burning emissions of BC/OC for 1850 367 1.15 6.79 CO21850 CO2 concentrations for 1850 285 1.85 11.8

CO2BB1850 Both CO2 and BB changed to 1850 285 1.15 6.79

*BB BASE BB1850; CO BASE CO21850; CO2BB BASE CO2BB1850.

wBurdens are averaged over southern Africa (subregion shown by dashed rectangle in Fig. 1) and over the June to September season.

2 NATURE COMMUNICATIONS | 7:11236 | DOI: 10.1038/ncomms11236 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11236 ARTICLE

Absolute precipitation changes (mm month1)

a b

Trend 1901:2010

BB BB

CO2 CO2BB

1981:2010 1901:1930 CO2BB

1981:2010 1901:1930

18 16 14 12

mm month1

10

8 6 4

1900 1920 1940 1960 1980 2000

WRF

CESM

Trend:0.038 mm month decade Pvalue: 0.42

Trend:0.11 mm month decade Pvalue: 0.23

GPCC

GPCC

Tanzania Mozambique

Zimbabwe

Gabon

Congo

DR Congo

Angola

Zambia Namibia

Botswana

South Africa

c d

CESM

CESM

e

f

mm month1

6 4 2 0 2 4 6

Relative precipitation changes (%)

g h

GPCC CESM

%

40 20 0 20 40

Figure 1 | Distributions of observed and modelled precipitation changes. Trend in observed precipitation (mm month 1) over the June to September season (a), absolute precipitation changes (mm month 1) for the same season from observations and various model experiments (bf), and relative precipitation changes (%) from observations and the CO2BB model experiment (gh). Values in (a) are averaged over the subregion shown by the dashed rectangle in (b). Panels a,b and g are from observations (GPCC) whereas the ve other panels are from models results, one with the regional WRF model (c) and four from the global CESM model (df,h). Model results are shown as differences from the BASE simulation and show differences between 2000 and 1850 conditions in each case; see text and Table 1 for notation of the experiments. Observed precipitation trends in (a) are shown for the whole GPCC time period from 1901 to 2010 (red) and from 1951 to 2010 (blue). Observations in (b) are shown as the difference 1981:2010 1901:1930 because 30-year

averages are needed to reduce noise (that is, the inuence of natural variability). Note that in most parts of the region, the number of stations per grid cell is low in the GPCC data. In plots (bf), symbols denote grid boxes where changes are signicant (P value o0.05) according to a two-tailed Students t-test (grey symbols), and when also accounting for multiple statistical testing according to ref. 54 (black symbols).

NATURE COMMUNICATIONS | 7:11236 | DOI: 10.1038/ncomms11236 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11236

burden (Supplementary Fig. 1c). In this region, the global (Community Earth System Model (CESM)) model BB and CO2BB simulations show statistically signicant reductions in precipitation. The BB simulation in the regional (WRF) model supports this nding, showing precipitation reduction in the same area, although not statistically signicant due to a smaller sample of years. An overall decrease in precipitation can be seen in southern Africa, but with some smaller regions experiencing increased precipitation due to biomass burning. Such regions of increased precipitation downwind of regions where BB leads to precipitation decreases can be expected and have been seen before11. The regional model, with higher spatial resolution than the global model, shows a more inhomogeneous distribution with more local scale features.

The modelled impact of industrial-era changes in CO2 concentrations on precipitation is an increase in the northeast of southern Africa and a decrease of up to 30% in most of the remaining part (Fig. 1e; Supplementary Fig. 2e). Thus, our results indicate that the known changes in both BB and CO2 concentrations may have led to a precipitation decrease since 1850, but with slightly different patterns. This decrease is enhanced in the simulation with combined changes in BB and CO2 concentrations (Fig. 1f,h). The precipitation pattern in the subregion in CO2BB resembles that of the sum of BB and CO2 (not shown), indicating that the two effects are to a large extent additive. Distributions of changes in the number of dry days (that is, days with precipitation o1 mm) and maximum number of consecutive dry days show similar tendencies of a drying due to BB and increased CO2 concentrations (Supplementary Figs 3 and 4).

Modelled precipitation changes compare relatively well with what has been observed over the previous century (Fig. 1; Supplementary Fig. 2). Notably, observed precipitation decreases close to regions where biomass burning emissions have increased most are well reproduced by both models BB experiment, particularly in Angola and south of the Democratic Republic of the Congo. Correlation coefcients for the subregion conrm that observed spatial patterns of absolute precipitation change correlate well with modelled precipitation changes induced by increased biomass burning (correlation coefcients 40.5 at 2.5

resolution, and signicant at Po0.05 level), while there is no correlation with modelled precipitation change due to increased CO2 (correlation coefcient B0.0) (Supplementary Table 1).

Uncertainties usually increase when going from larger to smaller scales in studies of climate change impacts, and a small displacement in simulated patterns may therefore lead to poor correlation. We note here that correlation coefcients calculated after averaging to coarser resolution are much higher (for example, 0.70.8 for BB at 5 resolution). It is also encouraging that the models are able to reproduce the spatial patterns of present-day precipitation climatology, with signicant (P valueo0.05) correlation coefcients of 0.97 and 0.78 for WRF and CESM, respectively (Supplementary Fig. 5; Supplementary Table 1).

The domination of BB aerosols in driving precipitation changes is unique for the dry season in the southern Africa region. Anthropogenic biomass burning aerosols are not found to dominate precipitation changes in other regions or seasons. In central Africa, biomass burning activities mainly take place from December to February, but the increase in biomass burning aerosol emissions since pre-industrial times has not led to a clear impact on precipitation (Supplementary Fig. 6). The third region of the world, in addition to southern and central Africa, with strong increases in biomass burning activities is the central part of South America. Here observations indicate a precipitation decline during its dry season from June to September, and global model results show that this is caused by changes in CO2 rather than changes in BB (Supplementary Fig. 7).

Changes in mean precipitation and number of dry days. In the following we focus on the southern African subregion indicated by the dashed rectangle in Fig. 1. This region has had a strong increase in biomass burning emissions, and is a region with minimal precipitation during the June to September period, implying that a climatic change in precipitation could have important societal effects.

The reduction in precipitation in the subregion due to BB and anthropogenic CO2 emissions is evident from Fig. 2, with a decrease in monthly mean precipitation and an increase in the number of dry days. This precipitation reduction is also observed, and although the absolute precipitation change of around 1 mm month 1 seems small, the relative change is as much as 510% (Supplementary Fig. 8).

It is also an average over a relatively large area, locally observed changes are much larger (Fig. 1b,g). With the exception of one ensemble member for the BB case and one for the CO2BB case, the decrease in mean precipitation and increase in the number of dry days are consistent for all cases, between all ensemble members, and across the two models (Fig. 2; Supplementary Fig. 8). In the BB and CO2BB cases, the results are statistically signicant (P valueo0.05) for all ensemble members aggregated (Fig. 2). The WRF model result is based on fewer years and therefore not statistically signicant at the same level, but is still in general agreement with the mean of the CESM ensembles.

1

0

1

mm month1

Number of dry days

2

3 CO2BB

1 CO2BB

CO2

a

1981:2010

1951:1980

BB

BB

1981:2010

1901:1930

b

2

1

0

CO2

BB

BB

CESM WRF GPCC

Figure 2 | Southern Africa average change in precipitation. Change in mean precipitation (mm month 1) (a) and number of dry days (b) over the

June to September season within the subregion (region indicated by dashed rectangle in Fig. 1) for various model experiments. Model results are shown as differences from the BASE simulation, where CESM denotes the global climate model and WRF the regional model. GPCC observations are both shown as the difference 1981:2010 1901:1930 and 1981:2010 1951:1980.

Error bars show the 95% condence interval derived from a two-tailed Students t-test.

4 NATURE COMMUNICATIONS | 7:11236 | DOI: 10.1038/ncomms11236 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11236 ARTICLE

Precipitation and the regional atmospheric energy budget. Analysis of the energy budget, including changes in horizontal energy transport, have proved useful in understanding global and regional precipitation changes21,37. The local atmospheric energy budget representing the difference between two climate situations can be approximated by

LdP dQ dHdry dLH dH

moist; 1 where L is the latent heat of condensation, dP is the change in surface precipitation ux, dQ is the change in diabatic cooling of the atmosphere, dHdry and dHmoist denote the change in

net inward horizontal dry and moist static energy transport, respectively, and dLH is the net upward latent heat ux from the surface (see Methods section and refs 21,37 for details). The diabatic cooling term (dQ) can be written dQ (dSW dLW dSH), where dSW and dLW is the net

atmospheric shortwave and longwave radiative heating, respectively, and dSH is the upward sensible heat ux from the surface. Equation 1 shows that precipitation changes are constrained by the atmospheric energy balance21.

Regional energy budgets for each model experiment are shown in Fig. 3 and Supplementary Fig. 9. Net atmospheric absorption of solar radiation from added BC in southern Africa is strong and exceeds 2 W m 2 (Fig. 3a,b). The level of agreement between the two models and between the various ensemble members is high relative to the other terms that experience much larger natural year-to-year variability.

The atmospheric radiative heating due to solar absorption by BC, or in other words, the reduced diabatic cooling term (dQ)

caused by BC, leads to a decrease in surface precipitation ux (assuming L is constant). However, this is partly balanced by reduced sensible heat ux from the surface to the atmosphere caused by reduced solar radiation reaching the surface. Scattering by OC particles also contributes to the reduced sunlight at the surface in the BB case. Absorbing aerosols have previously been shown to cause circulation changes, due to the induced temperature gradients, and this depends principally on the vertical location and magnitude of the aerosols38. Here, in the global model, increased net horizontal transport of dry static energy outward from the atmospheric column, presumably resulting from circulation changes, draws in the same direction as the sensible heat term (Fig. 3b). In the regional model, the horizontal transport terms are negligible (Fig. 3a), and this difference seems to be the main reason for the larger precipitation reduction in the regional model, compared with the global model, seen in the BB case in Fig. 2a.

The local energy budget for increased CO2 concentrations is fundamentally different from that for the BB case (Fig. 3c). Increased CO2 leads to increasing global sea surface temperatures, thus feedback effects dominate the atmospheric energy budget. The main change in the local atmospheric energy budget is from the LW radiative cooling term, caused by increased surface temperature, partly offset by atmospheric absorption39, and contributes to increased precipitation. However, this precipitation increase is counterbalanced by increases in atmospheric SW heating caused mainly by increased solar absorption by water vapour37, sensible heat ux from the surface, and net inward horizontal energy ux due to large scale circulation changes. Increase in CO2 leads to a precipitation decline; despite a small increase in latent heat ux from the surface to the atmosphere, a net loss of moist static energy in the horizontal direction leads to an overall decrease. Although the effect of increased CO2 concentrations is a precipitation increase on a global scale, this local decrease in precipitation is not unexpected as the spatial pattern of precipitation changes due to anthropogenic climate change is to a large extent affected by circulation changes37.

In fact, analysis of modelled vertical velocity shows a widening of the Hadley cell in the increased CO2 case, consistent with what has been observed40. This widening leads to more subsidence, and presumably less precipitation, south of B18S in the subregion, and slightly less subsidence north of this latitude (not shown), largely consistent with modelled precipitation changes (Fig. 1e; Supplementary Fig. 2e).

When both adding biomass burning aerosols and increasing CO2, the decrease in precipitation is driven by shortwave atmospheric absorption and horizontal transport of both dry and moist static energy (Fig. 3d). Overall the regional energy budget explains the precipitation changes illustrated in Figs 1 and 2, with dissimilar impact on the energy budget from CO2 and BB. Comparison of an additional transient simulation that includes changes in BC/OC from biomass burning and increased CO2 concentrations, and a transient simulation including all anthropogenic climate forcers, indicate that increases in CO2 and biomass burning BC/OC are the main drivers of changes to the atmospheric energy budget since pre-industrial time (Supplementary Fig. 10).

DiscussionIt should be noted that long-term measurements of precipitation in most of southern Africa are sparse23, and different observed precipitation products show relatively large discrepancies41. There may also be other drivers additional to BB and CO2,

such as non-CO2 greenhouse gases and other aerosols, which may have contributed to the observed precipitation changes. However, reective aerosols such as sulphate are not likely to contribute substantially because emissions of its main precursor, SO2, are

relatively small in and around the surrounding region35. Furthermore, remote inuences from the mid-latitudes cause only small effects in southern Africa42.

Previous studies have also found a clear impact of biomass burning aerosols on the hydrological cycle over southern and central Africa5,11,43,44. Although different experimental setups and focuses make it difcult to compare, there are some contradictory results in previous studies; some studies indicate more precipitation5,43 from biomass burning aerosols and others indicate less precipitation11,44, as is found here. Impacts of BC on clouds and precipitation are known to depend strongly on the altitude of added BC3,14, and this could be one reason for the differing results. The BC and OC aerosol concentrations in this study are from a model that includes a plume rise assumption for biomass burning emissions (OsloCTM2), and this leads to a portion of the aerosols being emitted above the boundary layer. Consequently, the main atmospheric heating from biomass burning BC aerosols occurs at an altitude high enough to inhibit vertical instability under certain conditions, and biomass burning BC may therefore suppress convective activity in our simulations. While we acknowledge that uncertainties in BC vertical distribution are large, it is well established that biomass burning emissions are rapidly lifted to high altitudes45. OsloCTM2 vertical proles of extinction coefcients compare very well with the CALIOP satellite instrument over the southern Africa region (see Fig. 4 in ref. 46).

Although the ndings in this study are supported by a regional and a global model, observations, and analysis of the regional energy budget, uncertainties exist when studying changes in precipitation on a regional scale. A multi-model study would be required to provide further robustness to the ndings.

The global climate model calculations in the present study were performed using a slab ocean setup, to reduce the required computational resources. We do not expect that our results would differ largely if a full ocean rather than a slab ocean model had

NATURE COMMUNICATIONS | 7:11236 | DOI: 10.1038/ncomms11236 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11236

a b

Experiment: BB. Model: WRF. Experiment: BB. Model: CESM.

TOA

TOA

ATM

SFC

TOA

ATM

SFC

W m2

W m2

W m2

4 2 0 2 4

2 4

2 0 2 4

W m2

W m2

W m2

4 2 0 2 4

2 4

2 0 2 4

ATM

SFC

P decreases

P increases

2 0

2 0

P decreases

P increases

SW LW SH Hdry LH Hmoist SW LW SH Hdry LH Hmoist

c

Experiment: CO2. Model: CESM. Experiment: CO2BB. Model: CESM.

d

W m2

W m2

W m2

4 2 0 2 4

2 4

2 0 2 4

TOA

ATM

SFC

W m2

W m2

W m2

4 2 0 2 4

2 4

2 0 2 4

2 0

P decreases

P increases

2 0

P decreases

P increases

SW LW SH Hdry LH Hmoist SW LW SH Hdry LH Hmoist

Figure 3 | Regional atmospheric energy budget for southern Africa. Atmospheric energy budget over the June to September season and averaged over the subregion (shown by the dashed rectangle in Fig. 1) for different models and experiments. Results are shown as differences from the BASE simulation, where CESM denotes the global climate model and WRF the regional model. At top-of-atmosphere (TOA) and surface (SFC), net downward energy uxes have positive values. The energy budget of the atmosphere (ATM) is dened as the difference between the energy balance at TOA and SFC, except for the horizontal transport terms (Hdry and Hmoist), which are positive when net energy transport is into the atmospheric column from the outside. Direction of brown and blue arrows on the right-hand-side of each plot denotes contribution towards decreased or increased surface precipitation according to equation 1 ( LdP dSW dLW dSH dHdry dLH dH

moist), and the total change in precipitation in each experiment is indicated by a black line embedded on these arrows. Error bars show the 95% condence interval derived from a two-tailed Students t-test.

been used. In a study using an earlier version of the CESM model47, a fully coupled ocean model simulation led to a slightly higher climate sensitivity. Hence, a slightly stronger response in the hydrological cycle could possibly have been expected in the increased CO2 case if a full ocean model would have been used.

The impact of BC on the hydrological cycle is however dominated by rapid adjustments15 and is therefore not likely to be strongly affected by the choice of slab ocean versus full ocean.

Our results show that local BC and OC aerosol emissions from anthropogenic biomass burning activities and increasing global CO2 concentrations are the main causes of the observed decline

in southern African dry season precipitation over the last century. This implies that, in addition to improving air quality, reducing biomass burning emissions of BC and OC may be a useful way to mitigate reduced rainfall in the southern Africa region. This conclusion is strengthened by the suggestion that these aerosols may be part of a positive feedback loop where increased drought could increase the risk of igniting more res11. Precipitation in the region is expected to decrease further in the future if greenhouse gas emissions continue unabated25. Hence, initiation of stringent air quality legislation, involving BC mitigation, may help to counteract this precipitation decline in the short term.

6 NATURE COMMUNICATIONS | 7:11236 | DOI: 10.1038/ncomms11236 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11236 ARTICLE

Methods

Observation data. We have used long-term observation data from the GPCC29,30, which provides monthly values of precipitation from 1901 to 2010 from 67,200 stations world-wide, gridded to a horizontal resolution of 0.5 0.5. This

observation dataset has been chosen because it gives a better coverage than other long-term gridded precipitation observation datasets, such as CRU and GHCN see TFE.1, Fig. 2 in ref. 48. In regions where GPCC coverage overlaps with CRU and GHCN, they give similar estimates of annual trends. The trend analysis in Fig. 1a has been calculated using the MannKendall method49, and P values in the observed trends are calculated using a two-tailed z-test.

Global climate model. Global climate model simulations have been carried out using the National Center for Atmospheric Research (NCAR) CESM1.0.4 (ref. 31). The atmospheric component, Community Atmosphere Model (CAM4) (ref. 32), was in our simulations run with 1.9 2.5 horizontal resolution and 26 vertical

layers, and coupled with the Community Land Model (CLM4) (ref. 50). Five different ensemble members have been run for 60 years each, whereof the rst 10 years were considered as spin-up and are not part of the analysis. The ensemble members were started from different initial conditions but the same climatological state, and we have used a conguration where the atmospheric component is coupled with a slab ocean model51. Results are shown as averages of all ensemble members. Atmospheric distributions of aerosol concentrations have been prescribed for black and organic carbon, and sulphate, using monthly mean 3-D concentrations from the chemistry-transport model OsloCTM2 from a previous study36. This model has previously shown good agreement with observed pattern and magnitude of aerosol optical depth in southern Africa during the biomass burning season34, and with vertical aerosol extinction coefcient proles over the same region46. In the CESM model experiments, both the CO2 and BC/OC concentrations have been changed globally. To test the effect of global versus local perturbation of biomass burning BC/OC, the rst two ensemble members of the BB1850 simulation were re-run, but with biomass burning BC/OC concentrations changed to 1850 levels only during June to September and only in and around southern Africa. The results in terms of southern Africa dry season precipitation change were within 25% of the results obtained when biomass burning BC/OC were changed globally and annually.

Regional climate model. Additional climate model simulations have been conducted using the regional WRF model33 version 3.5. The model has been run in 50 50 km horizontal resolution over Africa, and with 50 vertical layers from the

surface and up to 10 hPa. The extent of the horizontal domain is the same as dened in CORDEX-Africa52. WRF has been run for the 30 year period, 19842013, in 1 year time slices, each re-initialized in December of the previous year to allow one month of spin-up. Meteorological initial and boundary conditions, as well as sea-surface temperatures, have been taken from the 6-hourly 1 1 resolution nal reanalysis data of the Global Forecast System (GFS) model

of the National Centers for Environmental Prediction (NCEP) from year 2000 onward, and NCEP/NCAR 2.5 2.5 resolution reanalysis for the preceding

years53. The same radiation scheme (CAM) as in the CESM has been used in WRF. In the standard version of WRF, however, the CAM scheme only allows one xed column value of each aerosol species to represent the entire domain for the whole year. Here, the WRF code was modied to enable prescribed monthly mean 3-D concentration elds of BC, organic carbon and sulphate from OsloCTM2, for consistency with what is used in the CESM.

Energy budget. In the calculation of dHdry, we have used the same approach as in ref. 37 and calculated it as the residual of equation 1 (dHdry LdP dQ), assuming

L 29 W m 2 mm 1 day, rather than a direct calculation of the total dry static

energy ux. This approximation has been shown to introduce only small errors37. In Fig. 3, we have represented differences in horizontal transport of energy in and out of the atmospheric region both as dry static energy ux (dHdry) and as moist

static energy ux (dHmoist). The latter term is again calculated as a residual,

dHmoist dLH-LdP, and both LH and P are available from the model output.

References

1. Ramanathan, V., Crutzen, P. J., Kiehl, J. T. & Rosenfeld, D. Atmosphere -Aerosols, climate, and the hydrological cycle. Science 294, 21192124 (2001).

2. Allen, M. R. & Ingram, W. J. Constraints on future changes in climate and the hydrologic cycle. Nature 419, 224232 (2002).

3. Ban-Weiss, G. A., Cao, L., Bala, G. & Caldeira, K. Dependence of climate forcing and response on the altitude of black carbon aerosols. Clim. Dyn. 38, 897911 (2012).

4. Kawase, H. et al. Physical mechanism of long-term drying trend over tropical North Africa. Geophys. Res. Lett. 37, L09706 (2010).

5. Randles, C. A. & Ramaswamy, V. Direct and semi-direct impacts of absorbing biomass burning aerosol on the climate of southern Africa: a Geophysical Fluid Dynamics Laboratory GCM sensitivity study. Atmos. Chem. Phys. 10, 98199831 (2010).

6. Boucher, O. et al. in Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (eds Stocker, T. F. et al.) Ch. 7 571658 (Cambridge University Press, 2013).

7. Samset, B. H. et al. Modelled black carbon radiative forcing and atmospheric lifetime in AeroCom Phase II constrained by aircraft observations. Atmos. Chem. Phys. 14, 1246512477 (2014).

8. Koren, I., Kaufman, Y. J., Remer, L. A. & Martins, J. V. Measurement of the effect of Amazon smoke on inhibition of cloud formation. Science 303, 13421345 (2004).

9. Feingold, G., Jiang, H. L. & Harrington, J. Y. On smoke suppression of clouds in Amazonia. Geophys. Res. Lett. 32, L02804 (2005).

10. Rosenfeld, D. et al. Flood or drought: How do aerosols affect precipitation? Science 321, 13091313 (2008).

11. Tosca, M. G., Diner, D. J., Garay, M. J. & Kalashnikova, O. V. Human-caused res limit convection in tropical Africa: rst temporal observations and attribution. Geophys. Res. Lett. 42, 64926501 (2015).

12. Koch, D. & Del Genio, A. D. Black carbon semi-direct effects on cloud cover: review and synthesis. Atmos. Chem. Phys. 10, 76857696 (2010).

13. Ming, Y., Ramaswamy, V. & Persad, G. Two opposing effects ofabsorbing aerosols on global-mean precipitation. Geophys. Res. Lett. 37, L13701 (2010).

14. Samset, B. H. & Myhre, G. Climate response to externally mixed black carbon as a function of altitude. J. Geophys. Res.-Atmos. 120, 29132927 (2015).

15. Kvalevg, M. M., Samset, B. H. & Myhre, G. Hydrological sensitivity to greenhouse gases and aerosols in a global climate model. Geophys. Res. Lett. 40, 14321438 (2013).

16. Andrews, T., Forster, P. M., Boucher, O., Bellouin, N. & Jones, A. Precipitation, radiative forcing and global temperature change. Geophys. Res. Lett. 37, L14701 (2010).

17. Hodnebrog, ., Myhre, G. & Samset, B. H. How shorter black carbon lifetime alters its climate effect. Nat. Commun. 5, 5065 (2014).

18. Hartmann, D. L. et al. in Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (eds Stocker, T. F. et al.) Ch. 2 159254 (Cambridge University Press, 2013).

19. Greve, P. et al. Global assessment of trends in wetting and drying over land. Nat. Geosci. 7, 716721 (2014).

20. Knutti, R. & Sedlacek, J. Robustness and uncertainties in the new CMIP5 climate model projections. Nat. Clim. Change 3, 369373 (2013).21. OGorman, P. A., Allan, R. P., Byrne, M. P. & Previdi, M. Energetic constraints on precipitation under climate change. Surv. Geophys. 33, 585608 (2012).

22. Bala, G., Caldeira, K. & Nemani, R. Fast versus slow response in climate change: implications for the global hydrological cycle. Clim. Dyn. 35, 423434 (2010).

23. Niang, I. et al. in Climate Change 2014: Impacts, Adaptation, and Vulnerability. Contribution of Working Group II to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (eds Barros, V. R. et al.) Ch. 22 11991265 (Cambridge University Press, 2014).

24. Conway, D. et al. Climate and southern Africas water-energy-food nexus. Nat. Clim. Change 5, 837846 (2015).

25. Collins, M. et al. in Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (eds Stocker, T. F. et al.) Ch. 12 10291136 (Cambridge University Press, 2013).

26. New, M. et al. Evidence of trends in daily climate extremes over southern and west Africa. J. Geophys. Res.-Atmos. 111, D14102 (2006).

27. Aguilar, E. et al. Changes in temperature and precipitation extremes in western central Africa, Guinea Conakry and Zimbabwe, 1955-2006. J. Geophys. Res.-Atmos. 114, D02115 (2009).

28. Malhi, Y. & Wright, J. Spatial patterns and recent trends in the climate of tropical rainforest regions. Philos. Trans. R. Soc. Lond. Ser. B-Biol. Sci 359, 311329 (2004).

29. Schneider, U. et al.GPCC Full Data Reanalysis Version 6.0 at 0.5: Monthly Land-Surface Precipitation from Rain-Gauges built on GTS-based and Historic Data (2011).

30. Becker, A. et al. A description of the global land-surface precipitation data products of the global precipitation climatology centre with sample applications including centennial (trend) analysis from 1901present. Earth Syst. Sci. Data 5, 7199 (2013).

31. Gent, P. R. et al. The community climate system model version 4. J. Clim. 24,

49734991 (2011).

32. Neale, R. B. et al.Description of the NCAR Community Atmosphere Model (CAM 4.0). (NCAR Technical Report, NCAR/TN-485 STR (2010).

33. Skamarock, W. C. & Klemp, J. B. A time-split nonhydrostatic atmospheric model for weather research and forecasting applications. J. Comput. Phys. 227, 34653485 (2008).

NATURE COMMUNICATIONS | 7:11236 | DOI: 10.1038/ncomms11236 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11236

34. Myhre, G. et al. Modeling the solar radiative impact of aerosols from biomass burning during the Southern African Regional Science Initiative (SAFARI-2000) experiment. J. Geophys. Res.-Atmos. 108, 8501 (2003).

35. Lamarque, J. F. et al. Historical (18502000) gridded anthropogenic and biomass burning emissions of reactive gases and aerosols: methodology and application. Atmos. Chem. Phys. 10, 70177039 (2010).

36. Skeie, R. B. et al. Anthropogenic radiative forcing time series from pre-industrial times until 2010. Atmos. Chem. Phys. 11, 1182711857 (2011).

37. Muller, C. J. & OGorman, P. A. An energetic perspective on the regional response of precipitation to climate change. Nat. Clim. Change 1, 266271 (2011).

38. Lee, S. S. et al. Effect of gradients in biomass burning aerosol on shallow cumulus convective circulations. J. Geophys. Res.-Atmos. 119, 99489964 (2014).

39. Pendergrass, A. G. & Hartmann, D. L. The atmospheric energy constraint on global-mean precipitation change. J. Clim. 27, 757768 (2014).

40. Seidel, D. J., Fu, Q., Randel, W. J. & Reichler, T. J. Widening of the tropical belt in a changing climate. Nat. Geosci. 1, 2124 (2008).

41. Nikulin, G. et al. Precipitation climatology in an ensemble of CORDEX-Africa regional climate simulations. J. Clim. 25, 60576078 (2012).

42. Shindell, D. T., Voulgarakis, A., Faluvegi, G. & Milly, G. Precipitation response to regional radiative forcing. Atmos. Chem. Phys. 12, 69696982 (2012).

43. Tummon, F., Solmon, F., Liousse, C. & Tadross, M. Simulation of the direct and semidirect aerosol effects on the southern Africa regional climate during the biomass burning season. J. Geophys. Res.-Atmos. 115, D19206 (2010).

44. Tosca, M. G., Randerson, J. T. & Zender, C. S. Global impact of smoke aerosols from landscape res on climate and the Hadley circulation. Atmos. Chem. Phys. 13, 52275241 (2013).

45. Freitas, S. R. et al. Including the sub-grid scale plume rise of vegetation res in low resolution atmospheric transport models. Atmos. Chem. Phys. 7, 33853398 (2007).

46. Kof, B. et al. Application of the CALIOP layer product to evaluate the vertical distribution of aerosols estimated by global models: AeroCom phase I results.J. Geophys. Res.-Atmos. 117, D10201 (2012).47. Danabasoglu, G. & Gent, P. R. Equilibrium climate sensitivity: is it accurate to use a slab ocean model? J. Clim. 22, 24942499 (2009).

48. Stocker, T. F. et al. in Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (eds Stocker, T. F. et al.) Ch. TS 33115 (Cambridge University Press, 2013).

49. Sen, P. K. Estimates of the regression coefcient based on Kendalls Tau. J. Am. Stat. Assoc. 63, 13791389 (1968).

50. Lawrence, D. M. et al. Parameterization improvements and functional and structural advances in version 4 of the Community land model. J. Adv. Model. Earth Sy. 3, M03001 (2011).

51. Kiehl, J. T., Shields, C. A., Hack, J. J. & Collins, W. D. The climate sensitivity of the community climate system model version 3 (CCSM3). J. Clim. 19, 25842596 (2006).

52. Kim, J. et al. Evaluation of the CORDEX-Africa multi-RCM hindcast: systematic model errors. Clim. Dyn. 42, 11891202 (2014).

53. Kalnay, E. et al. The NCEP/NCAR 40-year reanalysis project. Bull. Amer. Meteorol. Soc. 77, 437471 (1996)077o0437:tnyrp42.0.co;2 (1996).

54. Ventura, V., Paciorek, C. J. & Risbey, J. S. Controlling the proportion of falsely rejected hypotheses when conducting multiple tests with climatological data. J. Clim. 17, 43434356 (2004).

Acknowledgements

GPCC precipitation data are provided by the NOAA/OAR/ESRL PSD, Boulder,

Colorado, USA, from their web site at http://www.esrl.noaa.gov/psd/

Web End =http://www.esrl.noaa.gov/psd/ . Funding from the

Research Council of Norway through the NAPEX (no. 229778) and HYPRE (no. 243942)

projects is acknowledged. P.M.F. acknowledges IAGP grant EP/I014721/ and a Royal

Society Wolfson Merit Award.

Author contributions

.H. wrote the paper, but with contributions from all the authors. G.M. developed the

idea. G.M. and .H. designed the GCM and RCM model experiments and .H. carried

out the experiments. All authors participated in analysis of the model results.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/

http://www.nature.com/naturecommunications

Web End =naturecommunications Competing nancial interests: The authors declare no competing nancial interests. Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions

Web End =http://npg.nature.com/

http://npg.nature.com/reprintsandpermissions

Web End =reprintsandpermissions /

How to cite this article: Hodnebrog, . et al. Local biomass burning is a dominant cause

of the observed precipitation reduction in southern Africa. Nat. Commun. 7:11236

doi: 10.1038/ncomms11236 (2016).

This work is licensed under a Creative Commons Attribution 4.0

International License. The images or other third party material in this

article are included in the articles Creative Commons license, unless indicated otherwise

in the credit line; if the material is not included under the Creative Commons license,

users will need to obtain permission from the license holder to reproduce the material.

To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

8 NATURE COMMUNICATIONS | 7:11236 | DOI: 10.1038/ncomms11236 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications