SupplementaryMaterial

Feedbacks between deforestation, climate, and hydrology in the Southwestern Amazon: implications for the provision of ecosystem services

Letícia S. Lima, Michael T. Coe, Britaldo S. Soares Filho, Santiago V. Cuadra, Lívia C. Dias, Marcos H. Costa, Leandro S. Lima, Hermann O. Rodrigues

Validation

Simulations using dynamic vegetation models –e.g. IBIS (Kucharik et al. 2000), and atmospheric general circulation models, AGCMs –e.g. CCM3 (Kiehl et al. 1998), have been applied in several studies of the Amazon basin and our results are broadly consistent with results from studies withdeforestation scenarios (Lean and Warrilow 1989; Shukla et al. 1990; Salati and Nobre 1991; Hahmann and Dickinson 1997; Costa and Foley 2000; Sampaio et al. 2007; Ramos da Silva et al. 2008; Coe et al. 2009).

Coupled CCM3-IBIS. CCM3-IBIS model was globally validated (Delire et al., 2002) and validated for the Amazon (Senna et al., 2009; Coe et al., 2009). It has been used in several studies to simulate the interactions among ecosystems and atmosphere in the Amazon basin to predict climate and hydrologic impacts of land use changes in this region (e.g., Costa and Foley, 2000; Coe et al., 2009). The Terrestrial Hydrology Model with Biogeochemistry – THMB, was validated for the Amazon basin by Coe et al. (2007, 2009).

The simulations of land cover changes in the coupled CCM3-IBIS and in IBIS stand-alone are performed replacing natural vegetation by tropical grasses (C4 species) in deforested areas, as in Costa et al. (2007). These grasses have lower leaf area indices (LAI) and shallower root systems, resulting in a lower transpiration rate per unit of leaf area when compared to Amazon tropical rain forest (dominated by C3 tree species).

Precipitation dataset. In order to evaluate the uncertainties in the precipitation dataset used in this study (CRU3.0, Mitchell and Jones, 2005), we compared the mean precipitation as simulated by the model against 4 different precipitation datasets: Cramer & Leemans (2001); Legates and Willmott (1990); Sheffield (2006) and CMAP (Xie and Arkin, 1997). The relative differences between the Control simulation(CTL), using CRU 3.0,and the average for all datasets were less than 4%; the greatest difference was found for the Madeira river basin (Table S1). Thegreatestdifference between CTL and the datasets (up to 11.1%) were found for the comparison with CMAP dataset (Xie and Arkin, 1997) in the Madeira basin.

Evapotranspiration. In CTL simulation, average evapotranspiration (ET) values were ~1147 mm/year (Madeira basin), ~1331 mm/year (Purus basin)and ~1356 mm/year (Juruá basin). Madeira basin is partially covered by savanna biome which could explain lower values of simulated ET for this basin when compared to the other basins. These values are in good agreement with previous studies(Costa and Foley 1999). Field campaigns in central Amazon found evaporation values of ~1368 mm/year (Shuttleworth et al. 1988). Estimated ET based on analysis of eddy covariance sites data (Fisher et al. 2009)found values of ~1370 mm/year for the entire Amazon.

Discharge.The simulated mean annual river discharge (CTL simulation) was compared against observations from three different stations in the mainstream of each basin.Although the simulated discharge of Juruá and Purus Rivers were in good agreement with the observed values, we found an almost constant underestimation (about 30%) for the Madeira River. In order to find a reason for this specific bias, we calculated the differences between the observed and simulated discharge for different river segments. We found that the major discharge deficit came from outside the Brazilian border (more than 25%, upstream station number 15320002). The same pattern was not found for Juruá and Purus river basins, whose areas are almost entirely inside Brazilian borders. In fact, almost 60% of Madeira basin area is located outside the Brazilian border. As the precipitation used as input data in our simulations are based on rain gauges measurements, we concluded that most of the Madeira discharge bias were related with the CRU3.0 precipitation data set,once this area have a very low rain gaugedensity. The same underestimation of precipitation were found on the other precipitation datasets (Table S1). Similar bias were also reported in previous studies (e.g., Costa and Foley, 1997; Coe et al., 2002). In order to correct this bias, we applied a constant correction of an additional 30% on the runoff data calculated by IBIS for the Madeira basin (similar to Coe et al., 2002; 2009). The discharge results and relative error (RE) considering this correction are shown in Table S2.

The simulated mean discharge is in good agreement with the observed data. The overall dischargeunderestimation is about 6% (Table S2) withthe Purus basinhaving the greatest difference from the observations. Despite the uncertainties associated with the precipitation datasets, other sources of factors may explain the model deviations: (i) During the period of the observed data deforestation occurred but in the CTLsimulation the Amazon land cover wasconstant and did not include any human interference. (ii) The IBIS and THMB models may have unknown sources of bias as a result of model representation of biophysics, parameter uncertainties and those caused by spatial resolution.

Analysis

Mean values of P and ET. In the simulations with climate feedback (LCC_CF) the evapotranspiration and precipitation changes for the period of simulation were evaluated in terms of mean values for each basin. The same was done for the evapotranspiration in the simulations without climate feedback (LCC_NoCF). The comparison between scenarios was done considering the average values for the entire period of simulation (1950-1999), although the first two years of simulation (1950 and 1951) were excluded from the analysis as it is the period required to enable the hydrologic model (THMB) to reach the equilibrium.

Changes in precipitation seasonality. In the LCC_CF simulation, we analyzed the changes in precipitation for each scenario for each season (Figure 3, paper).Firstly, we calculated the spatial average value for each basin and each month. Secondly, we calculated the monthly average values considering the entire period of simulation (excluding the first two years), presented in Table S3. Then we calculated the average for every group of three months (December-February; March-May; June-August; September-November). The resulted values of every scenario were then compared to the CTL simulation in terms of the relative difference (%).

Water deficit period. The water deficit period is analyzed in the LCC_CF simulation.It is defined as the period in which the values of precipitation minus evapotranspiration are less than zero (P-ET0). First, we calculated the monthly mean of each of these variables (P and ET) for the simulation period, averaged for the basin area. We discarded the first two years as they were used by THMB to reach the equilibrium. We calculated the difference (P-ET) for each monthly mean, and after that, the results for all scenarioswere plotted in the same graph in order to compare the simulations.

River discharge. Observed values of river discharge was obtained from the AgênciaNacional de Águas (ANA) website ( for the stations listed in Table S2. In order to compare simulated versus observed values we considered the initial year of observations in each station calculating the monthly average values, and then annual average values. For the results presented in the paper, we considered the values obtained in the closest stations to the mouth of the main rivers, which was the stations: “Gavião” in Juruá river (ANA station code 12840000); “Arumã-jusante” in Purus river (ANA station code 13962000); and “Manicoré” in Madeira river (ANA station code 15700000).

References

  1. Coe MT, Costa MH, Botta A, Birkett C (2002) Long-term simulations of discharge and floods in the Amazon basin. J Geophys Res 107, 17.
  2. Coe MT, Costa MH, Howard EA (2007) Simulating the surface waters of the Amazon River basin: impacts of new river geomorphic and flow parameterizations. Hydrol Process 22: 2542-2553.
  3. Coe MT, Costa MH, Soares-Filho BS (2009) The influence of historical and potential future deforestation on the stream flow of the Amazon river – land surface processes and atmospheric feedbacks. J Hydrol 369:165-174.
  4. Costa MH, Foley JA(1997) Water balance of the Amazon Basin: dependence on vegetation cover and canopy conductance. J Geophys Res 102, 17.
  5. Costa MH, Foley JA (1999) Trends in the hydrologic cycle of the Amazon basin. J Geophys Res 104:14189-14198.
  6. Costa MH, Foley JA (2000) Combined effects of deforestation and doubled atmospheric CO2concentrations on the climate of Amazonia. J Clim 13:18-34.
  7. Costa MH, Oliveira CHC, Andrade G, Bustamante TR, Silva FA, Coe MT (2002) A macroscale hydrological data et of river flow routing parameters for the Amazon basin. J Geophys Res 107, D20.
  8. Costa MH, Yanagi SNM, Souza PJOP, Ribeiro A, Rocha EJP (2007) Climate change in Amazonia caused by soybean cropland expansion, as compared to caused by pastureland expansion. Geophys Res Lett 34:L07706.
  9. Cramer WP, Leemans R(2001) Global 30-Year Mean Monthly Climatology, 1930-1960, version 2.1. Data set available at from Oak Ridge National Laboratory Distributed Active Archive Center, Oak Ridge, USA. Accessed August, 15, 2011.
  10. Delire C, Foley JA, Thompson S (2004) Long-term variability in a coupled atmosphere-biosphere model.J Clim 17:3947-3959.
  11. Fisher JB, Malhi Y, Bonal D, Rocha HR, Araújo AC, et al. (2009) The land-atmosphere water flux in the tropics. Glob Change Biol 15:2694-2714.
  12. Hahmann AN, Dickinson RE (1997) RCCM2-BATS model over Tropical South America: applications to tropical deforestation. J Clim 10:1944-1964.
  13. Kiehl JT, Hack JJ, Bonan GB, Boville BA, Williamson DL, Rasch PJ (1998) The National Center For Atmospheric Research Community Climate Model: CCM3. J Clim 11:1131-1149.
  14. Kucharik CJ, Foley JA, Delire C, Fisher VA, Coe MT, Lenters JD, Young-Molling C, Ramankutty N (2000) Testing the performance of a Dynamic Global Ecosystem Model: water balance, carbon balance, and vegetation structure. Glob Biogeochem Cycles 14:795-825.
  15. Lean J, Warrilow DA (1989) Simulation of the regional climatic impact of Amazon deforestation. Nature 342:411-413.
  16. Legates DR,Willmott CJ(1990) Mean seasonal and spatial variability in gauge-corrected, global precipitation. Int J Climatol 10:111-127.
  17. Mitchell TD, Jones PD (2005) An improved method of constructing a database of monthly climate observations and associated high-resolution grids. Int J Climatol 25:693-712.
  18. Ramos da Silva R, Werth D, Avissar R (2008) Regional impacts of future land-cover changes on the Amazon basin wet-season climate. J Clim 21:1153-1170.
  19. Salati E, Nobre CA (1991) Possible climatic impacts of tropical deforestation. Clim Change 19:177-196.
  20. Sampaio G, Nobre C, Costa MH, Sayamurty P, Soares-Filho BS, Cardoso M (2007) Regional climate change over eastern Amazonia caused by pasture and soybean cropland expansion. Geophys Res Lett 34:L17709.
  21. Senna MCA, Costa MH, Pinto LIC, Imbuzeiro HMA, Diniz LMF, Pires GF (2009) Challenges to Reproduce Vegetation Structure and Dynamics in Amazonia Using a Coupled Climate–Biosphere Model. Earth Interact 13:1-28.
  22. Sheffield J, Goteti G, Wood EF (2006)Development of a 50-Year high-resolution global dataset of meteorological forcingsfor land surface modeling. J Clim 19:3088-3111.
  23. Shukla J, Nobre C, Sellers P (1990) Amazon Deforestation and Climate Change. Science 247:1322-1325.
  24. Shuttleworth WJ (1988) Evaporation from Amazonian Rainforest. Proc R SocLond B 233(1272):321-346.
  25. Xie P, Arkin PA(1997) Global precipitation: A 17-year monthly analysis based on gauge observations, satellite estimates, and numerical model outputs. Bull Am MeteorolSoc78:2539-2558.

Fig. S1: Deforestation scenarios for the Amazon used in the simulations (original spatial resolution). (a) End of Deforestation by 2020 (ED202020) (Nepstad et al., 2009); (b) Business as Usual by 2030 (BAU2030) (Soares-Filho et al., 2006), and (c) Business as Usual by 2050 (BAU2050) (Soares-Filho et al., 2006). The original dataset was resampled to model’s spatial resolution (CCM3: 2.81°; IBIS: 0.0833°).

Juruá / Purus
Dataset / mm/day / RE (%) / Dataset / mm/day / RE (%)
Sheffield / 6.2 / -1.6 / Sheffield / 6.0 / -0.7
Cramer / 6.0 / 2.3 / Cramer / 6.0 / -0.8
CMAP / 5.6 / 9.8 / CMAP / 5.4 / 10.0
Legates / 6.6 / -7.6 / Legates / 6.1 / -2.7
Average / 6.1 / 0.0 / Average / 5.9 / 0.9
Control (CRU) / 6.1 / 0.0 / Control (CRU) / 5.9 / 0.0
RelativeError (%)
Madeira / RE(%) = (Control/Dataset-1)*100
Dataset / mm/day / RE (%)
Sheffield / 4.4 / 7.3 / Dataset / Time span / SpatialResolution(°)
Cramer / 4.6 / 3.5 / Sheffield / 1970-1999 / 1.0
CMAP / 4.3 / 11.1 / Cramer / 1930-1960 / 0.5
Legates / 4.8 / -0.9 / CMAP / 1979-2000 / 2.5
Average / 4.6 / 3.6 / Legates / 1920-1980 / 2.5
Control (CRU) / 4.7 / 0.0 / Control (CRU) / 1950-1999 / 0.5

Table S1 – comparison of mean values (first column) and relative errors (second column) between different precipitation datasets and the CTL simulation using CRU 3.0 (Mitchell and Jones, 2005). The gray box depicts the time span and spatial resolution of each dataset.

Stationcode
Agência Nacional de Águas / Station number
(Fig. 1, paper) / Station name / Observed mean river
discharge (m3/s) / Simulated mean river
discharge (m3/s) / Relative Error (RE)
(%)
Juruá River
12550000 / 6 / Eirunepé - montante / 1833.9 / 1799.2 / -2
12680000 / 5 / Envira / 1296.8 / 1310.4 / 1
12840000 / 9 / Gavião / 4780.3 / 4368.7 / -9
Purus River
13650000 / 3 / Floriano Peixoto / 590.4 / 576.2 / -2
13600002 / 1 / Rio Branco / 344.2 / 370.7 / 8
13880000 / 7 / Canutama / 6537.9 / 5574.7 / -15
13962000 / 10 / Arumã-jusante / 10469 / 9244.8 / -12
Madeira River
15320002 / 2 / Abunã / 18099.2 / 16233.1 / -10
15630000 / 4 / Humaitá / 21829.3 / 20367.7 / -7
15700000 / 8 / Manicoré / 24726 / 22395 / -9
Averagerelativeerror (%) / -6

Table S2 – Observed versus simulated mean river discharge (CTL simulation) and relative error (RE) for ten stations in the study area.

Juruá
Precip.
(mm/day) / CTL / LCC_CF / Purus
Precip.
(mm/day) / CTL / LCC_CF
Control / ED2020 / BAU2030 / BAU2050 / Control / ED2020 / BAU2030 / BAU2050
Jan / 8.9 / 8.7 / 8.5 / 8.0 / Jan / 9.3 / 9.2 / 9.4 / 8.9
Feb / 9.4 / 8.7 / 8.6 / 7.9 / Feb / 9.9 / 9.4 / 9.6 / 8.9
Mar / 9.6 / 9.2 / 9.0 / 8.2 / Mar / 9.7 / 9.6 / 9.6 / 9.0
Apr / 7.5 / 7.4 / 7.2 / 6.9 / Apr / 7.3 / 7.0 / 7.0 / 6.5
May / 4.6 / 4.4 / 4.4 / 4.4 / May / 4.3 / 4.0 / 4.1 / 4.1
Jun / 2.5 / 2.4 / 2.4 / 2.4 / Jun / 1.9 / 1.8 / 1.7 / 1.7
Jul / 1.9 / 1.9 / 1.8 / 1.8 / Jul / 1.2 / 0.9 / 0.8 / 0.7
Aug / 2.5 / 2.2 / 2.1 / 1.9 / Aug / 1.8 / 1.2 / 1.0 / 0.8
Sep / 4.3 / 3.4 / 2.3 / 1.6 / Sep / 4.0 / 3.2 / 2.3 / 1.8
Oct / 6.3 / 5.6 / 4.8 / 4.0 / Oct / 5.7 / 4.9 / 3.9 / 3.2
Nov / 7.3 / 7.8 / 7.8 / 7.3 / Nov / 7.7 / 8.2 / 8.2 / 7.4
Dec / 8.5 / 8.8 / 8.6 / 8.3 / Dec / 8.5 / 8.7 / 8.7 / 8.1
Madeira
Precip.
(mm/day) / CTL / LCC_CF
Control / ED2020 / BAU2030 / BAU2050
Jan / 8.2 / 8.1 / 8.1 / 8.4
Feb / 8.5 / 8.0 / 8.3 / 8.2
Mar / 7.1 / 6.8 / 6.8 / 6.7
Apr / 4.3 / 4.1 / 4.1 / 4.0
May / 2.6 / 2.5 / 2.5 / 2.5
Jun / 1.5 / 1.4 / 1.4 / 1.3
Jul / 1.0 / 0.9 / 0.9 / 0.9
Aug / 1.3 / 1.1 / 0.9 / 0.9
Sep / 2.5 / 2.0 / 1.7 / 1.6
Oct / 4.2 / 3.5 / 3.2 / 3.1
Nov / 5.7 / 5.3 / 4.7 / 4.9
Dec / 7.4 / 7.1 / 7.0 / 7.1

Table S3: Average values of precipitation for each basin in each scenario of the LCC_CF simulations, averaged for the entire period of simulation (1950-1999).