Climatic Change (2015) 129:117–129 DOI 10.1007/s10584-015-1338-x

Impacts of future climatic and land cover changes on the hydrological regime of the Madeira River basin J. L. Siqueira Júnior & J. Tomasella & D. A. Rodriguez

Received: 13 December 2013 / Accepted: 18 January 2015 / Published online: 3 February 2015 # Springer Science+Business Media Dordrecht 2015

Abstract Brazilian strategic interest in the Madeira River basin, one of the most important of the southern Amazon tributaries, includes the development of hydropower to satisfy the country’s growing energy needs and new waterways to boost regional trade and economic development. Because of evidences that climate change impacts the hydrological regime of rivers, the aim of this study was to assess how global climate change and regional land cover change caused by deforestation could affect the river’s hydrological regime. To achieve this goal, we calibrated a large-scale hydrological model for the period from 1970–1990 and analyzed the ability of the model to simulate the present hydrological regime when climate model simulations were used as input. Climate change projections produced by climate models were used in the hydrological model to generate scenarios with and without regional land-use and land-cover changes induced by forest conversion to pasture for the period from 2011– 2099. Although results show variability among models, consensus scenarios indicated a decrease in the low-flow regime. When the simulations included forest conversion to pasture, climate change impacts on low flows were reduced in the upper basin, while, in the lower basin, discharges were affected along the whole year due to the more vigorous land-use conversion in the Brazilian region of the basin.

1 Introduction The Amazon basin, generally considered to be one of the world’s critical biodiversity hotspots, is a mosaic of ecosystems, vegetation types and various forest formations in addition to including several types of savanna and natural pasture formations. The basin includes a large variety of landscapes, soils, geological domains and different climate regimes that determine contrasting hydrological regimes (Molinier et al. 1996). Recently, the region has drawn the attention of the scientific community because of the occurrence of extreme events, all of them with approximately 100 years of recurrence, including severe hydrological droughts from 2005

Electronic supplementary material The online version of this article (doi:10.1007/s10584-015-1338-x) contains supplementary material, which is available to authorized users.

J. L. S. Júnior (*) : J. Tomasella : D. A. Rodriguez Centro de Ciência do Sistema Terrestre - INPE, Cachoeira Paulista, Brazil e-mail: [email protected]

118

Climatic Change (2015) 129:117–129

to 2010 (Marengo et al. 2008; Tomasella et al. 2011) and major floods from 2009 to 2012 (Marengo et al. 2013). The Madeira River is already one of the Amazon tributaries experiencing intense deforestation (Trancoso et al. 2009). The size and the rapid rate at which the forest is being converted to either agricultural or pasture have fueled scientific interest about the potential impacts of those changes on the hydrological regime of the Amazon River (Coe et al. 2009). In addition, climate change (CC) scenarios predicted using regional climate models indicate that the central and eastern Amazon regions may experience rainfall deficiencies in the future and an increased frequency of extreme precipitation events over most of western Amazonia (Marengo et al. 2011). In regard to identifying the determinant factors of land-use and land-cover changes (LUCC) in Amazonia, the essential role of the combination of economic factors and national policies is undeniable (Geist et al. 2006). In this context, the Brazilian federal government has set out plans to push economic development through an initiative known as the Growth Acceleration Program - PAC (PAC 2013). The PAC has made investments of public and private funds totaling $52 billion in the Amazon region between 2012 and 2014 and is planning an additional investment of $36 billion through 2020. Of the total amount invested in the Brazilian Amazon, 37 % will be allocated in the construction of hydropower dams, which corresponds to 45 % of the planned energy expansion during the period. The new dams will increase the Brazilian Amazon’s share of power generation to 23 % of the national output, up from 10 % today. Therefore, the decisive role of the Amazon region in balancing Brazil’s future energy requirements is crucial. The objective of this paper is to analyze how land use, land cover and global climate may impact the hydrological regime of the Madeira River. We calibrated a large-scale hydrological model to the Madeira Basin using comprehensive quality controlled meteorological and hydrological data from Bolivia, Brazil and Peru, rather than reanalysis (Ribeiro Neto et al. 2008), or satellite data (Collischonn et al. 2008), or models comparisons (Guimberteau et al. 2012) as in previous studies in Amazonia. We verified the performance of the hydrological model in terms of river discharges in 19 sub-basins using both observed data and climate model simulations as input for the period 1970–1990. This analysis, generally not considered in previous studies, is critical to reduce the uncertainties of the projected scenarios. We assessed the potential impacts on the hydrological regime of several global climate projections with and without the inclusion of regional deforestation trajectories in two key locations in Bolivia and Brazil. While in most previous studies in Amazonia (for instance Guimberteau et al. 2012) the impacts on hydrological regime have been assessed in terms of monthly and sometimes annual discharges, we analyzed impacts in terms of high and low discharges extracted from the flow duration curve. Since the impacts of global climate change and regional land cover changes are more significant in the extremes rather than on the mean values, we concluded that the most previous studies could be underestimating the potential impacts.

2 Study area With drainage area of approximately 1,420,000 km2 across Bolivia (51 %), Brazil (42 %) and Peru (7 %), the Madeira River is one of the most important Amazonian tributary. The Madeira river mean discharge at its mouth to the Amazon is approximately 31,704 m3s−1, which corresponds to 15 % of the Amazon mean discharge. Andean tributaries of the Madeira River drain semi-arid areas of high altitude and areas of tropical humid forest of the piedmont. The

Climatic Change (2015) 129:117–129

119

basin mean average rainfall is approximately 1834 mm year−1 (Molinier et al. 1996), with a strong spatial variability ranging from 255 mm year−1 at the station of Caracato (2650 masl) in the Bolivian Andes to more than 3000 mm year−1 in stations located at less than 1500 masl (Espinoza Villar 2009).

3 Data and methodology This section provides a brief summary of the type of data, data processing and models used in this paper. Supplementary Material, sections S.1-6, presents a detailed description of the sources of data, the hydrological model formulation, climate models used and the methodology. We assembled meteorological, hydrological and soil data from different sources of Bolivian, Brazilian and Peruvian sources (Fig. 1). Topographic and historical LUCC trends were derived from satellite data. For future projections, we combined regional LUCC trajectories with global CC from a suite of IPCC models. The distributed hydrological model developed at the Instituto Nacional de Pesquisas Espaciais, hereafter referred as MHD-INPE (Tomasella and Rodriguez 2014), was calibrated in 19 sub-basins of the Madeira River using observed discharges and interpolated meteorological data as input) for the period from 1970 to 1990 by applying the Shuffled Complex Evolution Algorithm (Duan et al. 1994) to the simple average of two objective functions (performance criteria). In addition, the performance of the model was evaluated using Flow Duration Curve—FDC signatures (Ley et al. 2011). Observed meteorological data was also used to correct the bias of climate models simulations, both for present (1970–1990) and future (2011–2099) climate. To assess the ability of the hydrological model to simulate discharge when using climate model results as input, we compared observed flow duration curves with simulations of the hydrological model for the present climate. Bias-corrected climate projections were then used as the input of the hydrological model to assess the impact of CC on river discharges on the monthly means and various FDC signatures, including and excluding regional LUCC trajectories. We used the non-parametric Wilcoxon test for medians to assess statistically significant changes in FDC signatures. Finally, we analyzed the potential impact of CC, both including and not including LUCC trajectories in two representative discharge stations in the upper and lower Madeira basin; and in terms of hydropower generation in Santo Antonio run-of-river plant, currently under construction in the lower Madeira basin.

4 Results and discussion 4.1 Climate model projections After bias correction the climate projections (details in the Supplementary Material S.7) indicate a reduction of the mean annual rainfall of −13.8, −8.1, −13.5 and −12.8 % for the models M1, M2, M3 and M4, respectively, at the end of the century. The global models MR5, CSR and HD2 showed a decrease in rainfall of −0.01, −1.3 and −4.0 %, respectively. These results are consistent with those of Marengo et al. (2011). Finally, the projections from IPS show a different sign and variability, suggesting an increase of 2.8 % on average by the end of the century. In terms of temperature, model projections suggest an increase of the average temperature for the period from 2071–2099 ranging from 3.8 °C in M2 to 6.2 °C in M4.

120

Climatic Change (2015) 129:117–129

Fig. 1 On the top shows the location of the Amazon and Madeira River basins; at the bottom shows the elevation and drainage network map of the Madeira River basin showing rainfall, discharge stations and the hydropower plants

Climatic Change (2015) 129:117–129

121

4.2 Discharge simulations for the period from 1970–1990 The results of the calibration procedure of the hydrological model for 19 sub-basins using observed data as input showed that the values of NS and NSlog were above 0.7 in all catchments, except for the Miraflores and Ariquemes sub-basins. The worst model performance is observed in Miraflores with NS equal to 0.53, which is acceptable for the purpose of this study (Moriasi et al. 2007). Discharge simulated for historical period (1970–1990), using atmospheric data as input, are consistent with results obtained from the calibration stage. Percentage biases of the runoff were less than 20 % and FDC signatures were consistent between simulations using observed meteorological data and climate model outputs. Supplementary Material S.8 includes details about data processing, the hydrological model implementation and calibrations, as well as the performance of the model for each sub-basins using meteorological observations and atmospheric simulations as input. 4.3 Global climate and land-use and land-cover changes impacts For the sake of simplicity, we show the results for two representative gauging stations: Puerto Siles, in the upper basin, and Fazenda Vista Alegre, in the lower basin (Fig. 1). For clarity, baseline information shown in the figures of this section corresponds to the average of monthly discharge of all climate models for the period 1970–1990, although hydrological impacts were calculated considering the difference between the results for future projections of each model and the simulation of the same model for the period 1970–1990. The statistical significance of the differences between the projections and baselines are shown in Tables S.8 and S.9 in section S9 of Supplementary Material. We considered that the potential impacts on the hydrological cycle due to LUCC and global climate change are more reliable when the discharges projected shown consensus in signal and along different time slices. For the scenarios where regional LUCC are not included, most of the models indicated a reduction in mean monthly discharges at Puerto Siles, as shown in Fig. 2 (a, c and e). However, CSR and HD2 projections indicated discharges higher than the baseline, mainly during the wet season, for the periods 2011–2040 and 2071–2099 respectively. With the inclusion of LUCC (Fig. 2b, d and f), the impacts of CC are partially counterbalanced and the dry season discharges increased for several of the models, except for the CSR projection for the period 2011–2040, which showed a reduction in low discharges. Moreover, the effects of LUCC in the wet season discharges were relatively less important than in the lower discharges. Regarding the FDC signatures, M1, M2, M4 and MR5 projections showed a reduction the slope of the FDC for medium range discharges, QSM. Meanwhile, M3, CSR, IPS and HD2 projections indicate a gradual increase, up to a 26 % increase in the case of the IPS simulation (Fig. 3a). However, it is important to note that only the M2 projection for 2041–2070 and HD2 projection for 2071–2099 were statistically significant. When LUCC trajectories were included, QSM decreased for at least one period in the M3, MR5, CSR, IPS and HD2 simulations (Fig. 3b). The mid-segment of the FDC is associated with flows from moderate size precipitation and to the intermediate-term primary and secondary base flow relaxation response of the basin (Yilmaz et al. 2008; Ley et al. 2011). QSM signature also represents the medium-range variability of runoff coefficients (Ley et al. 2011). Therefore, steeper slopes of the FDC indicate an increased flashiness of the discharge responses to precipitation (Ley et al. 2011), which are a response to increasing intra-seasonal variability of rainfall affecting the rainfallrunoff responses as observed, for instance in IPS projection (Fig. 3a). This effect was reduced

122

Climatic Change (2015) 129:117–129

Fig. 2 Comparison between averaged monthly discharge of all climate models for future projections (the periods from 2011–2040, 2041–2070 and 2071–2099) and the mean monthly discharge of all models for the period from 1970–1990 (referred to as the baseline) in the Puerto Siles station. Left side does not include projected deforestation trajectories; right side includes projected deforestation trajectories

when LUCC was included in the simulations (Fig. 3d) because the additional excess of soil water due to the reduction of evaporation increased aquifer recharge, and consequently the amount of base flow affecting directly the QSM signature. The differences between wet and dry season discharges, represented by SEASON, suggested an increased seasonality in the case of the M3, M4, MR5 and IPS simulations (Fig. 3c), though only the changes of M4 simulation was statistically significant. Again, when LUCC was included, almost all of the projections indicated a reduction in the simulated seasonality (Fig. 3d), with statistically significant changes in M1, M2 and HD2 projections but not in all time-slices. Because forest conversion to pasture reduces evaporation and the seasonal variability of soil moisture, this conversion affects the differences between wet and dry season discharges. This result is consistent with the trend detected in the case of the QSM signature, although QSM reflects the intra-seasonal variability, rather than differences between wet and dry season variability.

Climatic Change (2015) 129:117–129

123

Fig. 3 Variability between modeled and simulated signatures of the flow duration curve—FDC in the Puerto Siles station: slope of the FDC at the medium range—QSM (a–b); differences between wet and dry season discharges—SEASON (c–d); low flow segment of the FDC – MWL (e–f); and high flow segment of the FDC – MWH (g–h); Simulated discharges are the result of hydrological model simulations using data as input for all climate models using data for the period from 1970–1990. Left side panels do not include projected deforestation trajectories; right side include projected deforestation trajectories

With regard to the long-term sustainability of flow, indicated by MWL, all projections are consistent with the possibility of low flows being significantly lower in the scenarios without regional LUCC changes (Fig. 3e). Most of these changes were statistically significant. The impacts were reduced by the inclusion of forest conversion to pasture (Fig. 3f): only M1, M3

124

Climatic Change (2015) 129:117–129

and M4 showed significant reduction of MWL, while HD2 showed significant increment (Table S.8 in the Supplementary Material). These results indicate that land use conversion, which is associated with the reduction of transpiration, partially counterbalance the reduction of rainfall and increase potential evaporation projected in most climate change scenarios (Supplementary Material S.10). In general, there was no consensus among projections concerning the signals in peak discharges indicated by the MWH statistic, both under climate change scenarios and considering the conversion of forest to pasture (Fig. 3g–h). Most of the statistically significant changes resulted in decreasing MWH. Figure 4 shows the mean monthly discharge simulations for the Fazenda Vista Alegre station. Most of the scenarios in the simulations without deforestation (Fig. 4a, c, and e) suggested a reduction of discharge along the whole year. However, CSR projections produced higher than the baseline discharges in 2011–2040, while IPS projections discharges were higher in 2041–2070 and 2070–2099. Comparing the projections in Puerto Siles (Fig. 2a, c and e) and Fazenda Vista Alegre (Fig. 4a, c, and e), HD2 projections have the most different behavior (particularly for the period 2071–2099) among all the other climate models, associated with changes in the spatial distribution of rainfall. When LUCC are introduced in the simulations (Fig. 4b, d, and f), IPS, CSR and MR5 models indicated higher than the baseline discharges, with a larger variability among models both in signal and along time-slices. This result differs with the impacts predicted for the upper

Fig. 4 Comparison between averaged monthly discharge of all climate models for future projections (the periods from 2011–2040, 2041–2070 and 2071–2099) and the mean monthly discharge of all models for the period from 1970–1990 (referred to as the base line) in the Fazenda Vista Alegre station. The left side does not include projected deforestation trajectories; the right side includes projected deforestation trajectories

Climatic Change (2015) 129:117–129

125

gauging station of Puerto Siles, where the inclusion of deforestation in the simulations primarily affected the dry season discharge (Fig. 2b, d, and f), whereas in Fazenda Vista Alegre (Fig. 2b, d, and f), the impacts occurred over the entire year. The differences in the impact of LUCC in the upper and lower Madeira can be explained by the more vigorous landuse conversion projected for the lower basin (details in the Supplementary Material S.11), which affect the water balance more significantly in the Brazilian side of the basin. In addition, dry season flows in Fazenda Vista Alegre were, in the case of the M2, MR5, and IPS projections, lower than the dry season flows of the unaltered vegetation scenarios, which appear to contradict the fact that LUCC scenarios are associated with reduced evaporation. This result is due to the occurrence of faster flows in LUCC scenarios compared to the CC scenarios (because of the higher wet season discharges), which cause a more rapid recession. Since the wave propagation used by the hydrological model is purely kinematic, this result should be considered with caution because Amazon River tributaries suffer backwater effects close to the river main stem (Meade et al. 1991). In Fazenda Vista Alegre, most of the models indicated decrease in QSM of up to 23 % in M4 scenario (Fig. 5a), which were significant for M2, M3 and M4 simulations. Conversely, CSR, IPS and HD2 simulations suggested an increase of up to 17 %, with only the HD2 projection showing significant increment for 2071–2099. If the LUCC trajectories are included, the signal is reversed. Except for M2 and M4, the remaining projections showed an increased QSM of up to 36 % in the CSR scenario (Fig. 5b), implying a more flashy response of the basin at this station due to LUCC (Yilmaz et al. 2008). This is directly related to higher wet season flows associated with LUCC (Fig. 4b, d, and f) compared with the scenarios that include only CC (Fig. 4a, c, and e). In Fazenda Vista Alegre, SEASON signature suggested mixed signals among climate models (Fig. 5c). Only the increment indicated by IPS for 2041–2070 was statistically significant. Major differences are introduced by LUCC changes, and most of projections indicated an increasing seasonality of up to 58 % (HD2) (Fig. 5d). In general, projections agree in that the low discharges, indicated by MWL, could be statistically significantly reduced by the scenarios that do not include regional LUCC changes (Fig. 5e). The impacts were mixed under the vegetation conversion scenario, with the MWL signature reduced in the case of the M1, M3, M4 and in some time-slices for other projections. In addition, projections of IPS for 2071–2099 showed a statistically significant increase in MWL (Fig. 5f). Finally, the majority of the models suggest a decreased of MWH in the projections of CC scenarios (Fig. 5g). After the inclusion of LUCC changes (Fig. 5h), the signals are consistent in almost all simulations, and indicated a statistically significant increase in peak discharges. 4.4 Effect of climate and land use changes on energy potential The averaged energy potential estimated from the hydrological model simulations using all the climate models was 2359 MW for the period from 1970–1990 at the Santo Antonio dam location. This estimation is higher than the firm energy capacity of 2218 MW projected for the plant, because firm energy estimations are based on the hydrological critical period (June 1949 through November 1956). In terms of the impacts of the CC scenario, there is no consensus among models. The EtaINPE model (projections M1-M4) indicated consistent reductions (up to −45 % by 2099) of energy potential, whereas the remaining projections showed less important effects with mixed signals. For instance, the CSR projection suggests an increase of 12 % for the period from 2011–2040. Including LUCC in the projections, reductions in the M1-M4 integrations are

126

Climatic Change (2015) 129:117–129

Fig. 5 Variability between modeled and simulated signatures of the flow duration curve—FDC in the Fazenda Vista Alegre station: slope of the FDC at the medium range—QSM (a–b); differences between wet and dry season discharges—SEASON (c–d); low flow segment of the FDC—MWL (e–f); and high flow segment of the FDC—MWH (g–h); Simulated discharges are the result of hydrological model simulations using data as input for all climate models using data for the period from 1970–1990. Left side panels do not include projected deforestation trajectories; right side include projected deforestation trajectories

reduced, whereas the other projections indicate increases of up to 38 % (IPS) for the period from 2071–2099.

Climatic Change (2015) 129:117–129

127

Decreasing energy potential in the M1-4 scenarios was related to the reduction of discharges during the whole year due to CC effects, which are largely counterbalanced by LUCC effects. In MR5, CSR, IPS and HD2 projections, on the other hand, wet and dry season discharges changes due to CC have opposite signals, which resulted in lower than the baseline annual discharges and, then, in the energy potential. Under LUCC effects, both, higher and lower discharges increased, resulting in a more consistent change in the energy potential in those projections. Supplementary Material S.12 contains detailed information about the projections. It is clear that a consensus has not been reached between the models regarding energy potential. This result can be explained by the fact that energy potential estimations are based on average discharges, whereas the projections consensus regarding the impact in the upper Madeira Basin indicates the reduction of minimum discharges, which only marginally affects the average discharges.

5 Conclusions We evaluated the impacts of climate and land use change on flow regime of the Madeira River basin by considering climate change outputs from climate models with and without projected deforestation trajectories from a land-use and land-cover change model. The MHD-INPE hydrological model was successful in simulating observed river discharges in 19 sub-basins of the Madeira River. When the hydrological model used biascorrected climate model simulations as the input, the results indicated consistency with both the mean discharges and four FDC signatures that represent the entire hydrological regime. We conclude that the hydrological model forced with the climate model data adequately represented the observed hydrological regime. In general, changes induced by climate change were statistically significant in few FDC signatures. Significant changes were verified mainly in the high and low flow segments. The number of projected FDC signatures with significant change increased when LUCC is considered, especially the seasonality and variability in medium range flows. The different scenarios generated for the Puerto Siles gauge-station under climate change showed consensus in terms of the MWL statistics, which indicates a reduction of long-term sustainability of flow. If land-use and land-cover changes are introduced (mostly forest conversion to pasture), the scenarios conclude that there will be less intra-seasonal and seasonal variability in discharge related to an increase of soil moisture in the dry season. With regard to the Fazenda Vista Alegre gauge-station, the projections of climate change consistently suggest a reduction of lower discharges. When forest conversion to pasture is included in the analysis, the reduction of the impacts in lower discharges persists and coexists with increased higher discharges in most of the projections. The decline in projected discharge suggested by some models driven by climate change are consistent with those observed by Stickler et al. (2013). In general, it is possible to conclude that climate change scenarios affect extreme discharges of the Madeira River basin. Although forest conversion to pasture reduces these impacts in the upper Madeira basin, the effects in the lower Madeira basin are enhanced in the high discharges. These signals are a consequence of the more vigorous land-use conversion on the Brazilian side of the basin. Moreover, impacts are more significant in extreme discharges, which suggest that previous studies, based on monthly or even annual discharges, are probably underestimating the intensity of those impacts. We recognize that the results show variability between models, which is caused by uncertainties inherent in each of the model chains used to study the impacts. We used state

128

Climatic Change (2015) 129:117–129

of the art data models for climate and land-use and land-cover changes and verified their performance for current climate conditions using observations. In addition to the uncertainties resulting from our incomplete knowledge of the function of earth systems, the projections include other sources of uncertainty, such as economic policies and deforestation trajectories among others. Consequently, these conclusions should be analyzed accordingly. Acknowledgments The authors acknowledge ‘Rede GEOMA’, ‘Rede Clima’, ‘Conselho Nacional de Desenvolvimento Científico e Tecnológico’ (CNPq), HYBAM and the CMIP5 modelling groups.

References 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. doi:10.1016/j.jhydrol.2009.02.043 Collischonn B, Collischonn W, Tucci C (2008) Daily hydrological modeling in the Amazon basin using TRMM rainfall estimates. J Hydrol (Amsterdam), p. 207 Duan Q, Sorooshian S, Gupta V (1994) Optimal use of the SCE – UA global optimization method for calibrating watershed models. Journal of Hydrology, Vol 158:265–284. doi:10.1016/0022-1694(94)90057-4 Espinoza Villar JC (2009) Spatio-temporal rainfall variability in the Amazon basin countries (Brazil, Peru, Bolivia, Colombia, and Ecuador). International Journal of Climatology 29:1574–1594. doi:10.1002/joc.1791 Geist HJ, McConnell W, Lambin EF, Moran E, Alves D, Rudel T (2006) Causes and trajectories of land-use/ cover change. In: Lambin EF, Geist H (eds) Land-use and land-cover change, local process and global impacts. Springer, Berlin Heidelberg. doi:10.1007/3-540-32202-7 Guimberteau M et al (2012) Discharge simulation in the sub-basins of the Amazon using ORCHIDEE forced by new datasets. Hydrol Earth Syst Sci 16:911–35 Ley R, Casper MC, Hellebrand H, Merz R (2011) Catchment classification by runoff behaviour with self-organizing maps (SOM). Hydrology and Earth System Sciences 8:3047–3083. doi:10.5194/hessd-8-3047-2011 Marengo JA, Nobre CA, Tomasella J, Oyama MD, de Oliveira GS, de Oliveira R, Camargo H, Alves LM, Brown IF (2008) The drought of Amazonia in 2005. Journal of Climate 21:495–516. doi:10.1175/ 2007JCLI1600.1 Marengo JA, Chou S, Kay G, Alves LM, Pesquero J, Soares W, Santos DC, Lyra A, Sueiro G, Betts R, Betts R, Chagas D, Gomes J, Bustamante J, Tavares P (2011) Development of regional future climate change scenarios in South America using the Eta CPTEC/HadCM3 climate change projections: climatology and regional analyses for the Amazon, São Francisco and the Parana River basins. Climate Dynamics 23:1–20. doi:10.1007/s00382-011-1155-5 Marengo JA, Borma LS, Rodriguez DA, Pinho P, Soares WR, Alves LM (2013) Recent extremes of drought and flooding in Amazonia: vulnerabilities and human adaptation. American Journal of Climate Change 02:87– 96. doi:10.4236/ajcc.2013.22009 Meade R, Rayol J, Conceição S, Natividade J (1991) Backwater effects in the Amazon river basin of Brazil. Environmental Geology and Water Sciences 18:105–114. doi:10.1007/BF01704664 Molinier M, Guyot JL, Oliveira E, Guimarães V (1996) Les regimes hydrologiques de l’Amazone et de ses affluents. In L’hydrologie tropicale: géoscience et outil pour le développement. IAHS Publishers: Mai, Paris; 209–222 Moriasi DN, Arnold JG, Van Liew MW, Binger RL, Harmel RD, Veith TL (2007) Model evaluation guidelines for systematic quantification of accuracy in watershed simulations. Transactions of the American Society of Agricultural and Biological Engineers 50(3):885–900 Programa de Aceleração do Crescimento—PAC (2013) Ministério do Planejamento. Governo Federal. Ribeiro Neto A, Silva RCV, Collischonn W, Tucci CE (2008) Simulação na Bacia Amazônica com Dados Limitados: Rio Madeira. Revista Brasileira de Recursos Hídricos 13:47–58 Stickler CM, et al (2013) Dependence of hydropower energy on forests in the Amazon Basin at local and regional scales. [S.l.]. 2013. to the NWS distributed hydrologic model. Water Resour Res 44:W09417. doi:10.1029/ 2007WR006716 Tomasella J, Rodriguez DA (2014) A numerical approach for modelling sub-grid variability in tropical catchment. Hydrol Process - Submitted 2014

Climatic Change (2015) 129:117–129

129

Tomasella J, Borma LS, Marengo JA, Rodriguez DA, Cuartas LA, Nobre CA, Prado MCR (2011) The droughts of 1996–1997 and 2004–2005 in Amazonia: hydrological response in the river main-stem. Hydrol Process (25) 1228–1242. doi:10.1002/hyp.7889 Trancoso R, Carneiro Filho A, Tomasella J, Schietti J, Forsberg BR, Miller RP (2009) Deforestation and conservation in major watersheds of the Brazilian Amazon. Environmental Conservation 36:277–288. doi: 10.1017/S0376892909990373 Yilmaz KK, Gupta HV, Wagener T (2008) A process-based diagnostic approach to model evaluation: application to the NWS distributed hydrologic model. Water Resources Res 44, W09417. doi:10.1029/2007WR006716