The Impact Of Climate Change On The São Francisco River’s Hydroelectric Production

Pieter de Jong Doutorado em Engenharia Industrial (PEI - UFBA ); Mestrado em Engenharia Industrial (PEI - UFBA ); Bacharel em Engenharia Elétrica (Universidade Monash).

Universidade Federal da , Salvador-BA, 2017 Background São Francisco basin (SF) & the •The NE region´s hydroelectric Northeast region of Brazil potential is saturated. •The Northeast (NE) is driest part of Brazil and suffers from droughts. Northeast •The drought in the NE from 2012-2017 region is the worst in recorded history. SF •2014-2017 the São Francisco streamflow has been the lowest on record. •Climate change will negatively impact rainfall in the NE and hydroelectric availability.

de Jong Pieter, PEI, Politécnica, UFBA, Salvador – 2017. Schematic of hydroelectric plants along the São Francisco River

Northeast region Luiz Gonzaga/ Itaparica Apolônio Sales Sobradinho Paulo Alfonso Xingó

APOLÔNIO SALES /

Três Marias

Hydroelectric Dam Source: ONS

de Jong Pieter, PEI, Politécnica, UFBA, Salvador – 2017. Hydroelectric plants Apolônio Sales Sobradinho

Paulo Alfonso

Nov 2015 & 2017 the NE’s hydroelectric availability dropped to only 5% of its total storage capacity. Itaparica

Xingó NE Reservoir Volume Levels / Stored Energy 2005-2017 (percentage of the total capacity) Disponibilidade hidroeléctrica mensal

Source: ONS - Operador Nacional do Sistema Elétrico, 2017 Sources of the Northeast’s Electricity Fontes de energia elétrica para o Nordeste

30%

25%

25%

• “Thermal” electricity generation in the NE is from fossil fuels & biomass. “Imported” consists mostly of hydro from other Brazilian regions. • From 2005-2007 hydro was responsible for more than 87% of the NE’s electricity supply, however, by 2016 this figure dropped to 25%. • Source: ONS (2016).

de Jong Pieter, PEI, Politécnica, UFBA, Salvador – 2017. Objective, method and materials • IPCCC model rainfall projections are compared to the linear trend-line of historical average rainfall extrapolated until 2100. • Various studies that predict the percentage of average rainfall reduction by 2100, considering different IPCCC models, are examined. • Historical data from 1961-2017 for the average monthly rainfall within the São Francisco basin area was provided by CPTEC/INPE (Centro de Previsão de Tempo e Estudos Climáticos / Instituto Nacional de Pesquisas Espaciais). • Historical monthly streamflow data for the São Francisco River is sourced from the ONS (Operador Nacional do Sistema Elétrico). • The elasticity factor, which is the reduction in streamflow relative to the decline in rainfall during periods of low rainfall, is estimated from 2013-2016.

de Jong Pieter, PEI, Politécnica, UFBA, Salvador – 2017. Projected changes in precipitation until 2071-2100

Northeast region

Source: Marengo, 2008 and Marengo & Ambrizzi et al, 2007. • Climate change will cause reduced rainfall (up 60% reductions in the semi-arid NE). Projected temperature increases until 2071-2100

Northeast region

Source: Marengo, 2008 and Marengo & Ambrizzi et al, 2007. • Climate change will cause higher temperatures and wind speeds. Projected climate change for Brazilian river basins

Source: Marengo et al (2012) considering the IPCC B2 emission scenario. IPCC Projections of rainfall in the São Francisco basin • According to Tanajura et al (2010) rainfall could decline 25–50% in the in semi-arid areas of Bahia and up to 80% in coastal areas by 2100. • According to Marengo et al (2012) annual rainfall in the São Francisco basin will decrease by 35% by 2100 relative to the 1961–1990 baseline considering the B2 (≈RCP6.0) reduced emissions scenario. • Considering the A2 (RCP8.5) high emission scenario the predicted rainfall reduction could be 40-60% (Marengo, 2008). • However, the linear trend-lines of historical rainfall and streamflow (see next slide) show that they have already declined by 25% and 33%, respectively, relative to the 1961–1990 baseline. • This show that the IPCC models have large uncertainties and could have underestimated the decline in rainfall in the region.

de Jong Pieter, PEI, Politécnica, UFBA, Salvador – 2017. São Francisco (SF) basin Monthly Rainfall and Streamflow 1961-2017 (2 year rolling average)

Source: ONS (2017) and CPTEC / INPE (2017).

de Jong Pieter, PEI, Politécnica, UFBA, Salvador – 2017. Extrapolating Rainfall and Streamflow relative to long term averages Table 1 (based on linear trend-lines) Rainfall Rainfall reduction relative to 1961-90 avg: Year (mm) (mm) (%) 1961-90 90.2 Decrease per year 0.58 0.65% 1995 80.0 Decrease in 20 years 10.2 11.3% 2016 67.8 Decrease in 41 years 22.4 24.9% 2030 59.6 Decrease in 55 years 30.6 33.9% 2050 47.9 Decrease in 75 years 42.3 46.9% 2085 27.5 Decrease in 110 years 62.7 69.5% 2100 18.7 Decrease in 125 years 71.5 79.2%

Rainfall reduction relative to 1961-90 avg: Streamflow Streamflow reduction relative to 1931-90 avg: Year (MWavg) (MWavg) (%) 1931-90 8500 Decrease per year 67 0.79% 1995 7144 Decrease in 20 years 1356 16.0% 2016 5735 Decrease in 41 years 2765 32.5% 2030 4796 Decrease in 55 years 3704 43.6% 2050 3454 Decrease in 75 years 5046 59.4% 2085 1106 Decrease in 110 years 7394 87.0% 2100 100 Decrease in 125 years 8401 98.8% de Jong Pieter, PEI, Politécnica, UFBA, Salvador – 2017. Annual rainfall (trend-lines) in the São Francisco basin 1961-2080

Annual rainfall has been below its 1961-1990 long-term average every year since 1997.

Source: CPTEC / INPE (2017) and and Marengo et al (2012).

de Jong Pieter, PEI, Politécnica, UFBA, Salvador – 2017. Linear projections for the São Francisco basin • Linearly extrapolating rainfall reduction in the São Francisco basin from 1961-1990 to 2071-2100 (110 years) (as per the table in the previous slide) would see a reduction of approximately 70%. • Due to non-linear hydrologic processes, reductions in streamflow can actually be amplified 2 to 3 times relative to reductions in rainfall (SAFT et al, 2015 and TIMBAL et al, 2015). • This elasticity factor is due to a larger percentage of rainfall being lost to infiltration, irrigation and reservoir evaporation. • It was estimate that irrigation and reservoir evaporation equate to 7% and 11%, respectively, of the São Francisco’s long term average streamflow (1931-1991). • However, as a result of the drought, irrigation and reservoir evaporation equated to 20% and 30%, respectively, of the São Francisco’s streamflow in 2015. de Jong Pieter, PEI, Politécnica, UFBA, Salvador – 2017. São Francisco monthly rainfall and streamflow 2005-2017 vs long term averages (streamflow 1931 to 1990 & rainfall 1961-1990)

Source: ONS (2017) and CPTEC / INPE (2017).

• The slope of the respective trend-lines demonstrate that the drop in the São Francisco’s streamflow is more extreme when compared to the decline in rainfall. E.g. 2013-2016. • From June to November streamflow in the São Francisco reservoirs is at its lowest level which are typically the months that wind energy generation is at its highest.

de Jong Pieter, PEI, Politécnica, UFBA, Salvador – 2017. São Francisco monthly rainfall, streamflow, hydro storage and hydro generation 1996-2017 (12 month rolling averages)

Streamflow in the São Francisco basin dropped below 3000MWavg and is continuing to drop. Source: ONS (2017) and CPTEC / INPE (2017).

de Jong Pieter, PEI, Politécnica, UFBA, Salvador – 2017. Elasticity factor for the São Francisco basin • During the last 4 years average rainfall dropped 34% to below 60mm per month. As a result, since 2015 the average streamflow in the São Francisco dropped by 60% below its baseline and is continuing to drop. The average streamflow for 2017 is predicted to be 75% below the baseline average. • Therefore, an elasticity factor of 1.7-2 is assumed for the basin

Rainfall and streamflow under non-drought conditions • Linearly extrapolating the long term average monthly rainfall would see it drop to only 60mm by 2030, which is a reduction of 34% relative to the 1961-1990 baseline (see table). • Applying the elasticity factor means that by 2030, the long term average streamflow could actually drop approximately 60% compared to the baseline average (1931-1990).

de Jong Pieter, PEI, Politécnica, UFBA, Salvador – 2017. Summary of results • Linearly extrapolating rainfall data to 2050 sees a 47% decline in average rainfall and an 80% reduction in average streamflow. • This means the rainfall estimates for 2100 based on IPCC B2 and A2 projections could actually eventuate as early as 2030 and 2050, respectively. • In short, the deficits in rainfall and streamflow experienced during the last 4 years could become the norm by 2030. • Moreover, by 2030 the reduced streamflow in the São Francisco will only allow hydroelectricity to generate approximately 10- 15% of the NE’s electricity demand. • Therefore, in the coming decades the São Francisco river will be more important for urban water supplies, farming and irrigation.

de Jong Pieter, PEI, Politécnica, UFBA, Salvador – 2017. Jucazinho Reservoir full – Jucazinho Reservoir – Pernambuco 2017 PRINCIPAL REFERENCES • ANEEL – Agência Nacional da Energia Elétrica, BIG – Banco de Informações de Geração; 2014. Available at: http://www.aneel.gov.br/aplicacoes/capacidadebrasil/capacidadebrasil.asp accessed on 14/03/2017. • CPTEC / INPE - Centro de Previsão de Tempo e Estudos Climáticos / Instituto Nacional de Pesquisas Espaciais; 2017. • de Jong P, Kiperstok A, Torres E A. Economic and environmental analysis of electricity generation technologies in Brazil. Renewable and Sustainable Energy Reviews 2015; Accepted for publication. • de Jong P, Sánchez A S, Esquerre K, Kalid R A, Torres E A. Solar and wind energy production in relation to the electricity load curve and hydroelectricity in the Northeast region of Brazil. Renewable and Sustainable Energy Reviews 2013; 23: 526–535. • Lucena A F P, Szklo A S, Schaeffer R, Dutra R M. The vulnerability of wind power to climate change in Brazil. Renewable Energy 2010; 35: 904-12. • Lucena A F P de, SZKLO A S, SCHAEFFER R, SOUZA R R de, BORBA B S M C, COSTA I V L da, JÚNIOR A O P, CUNHA S H F da, The vulnerability of renewable energy to climate change in Brazil. Energy Policy 2009; 37: 879-889. • Marengo J A. Regional Climate Change Scenarios for -The CREAS project. Conference on Climate Change and Official Statistics, Oslo Norway, 2008. • Marengo J A, Chou S C, Kay G, Alves L M, Pesquero J F, Soares W R, et al. 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 Paraná River basins. Climate Dynamics 2012; 38: 1829–1848. • Marengo J A, Jones R, Alvesa L M, Valverde M C. Future change of temperature and precipitation extremes in South America as derived from the PRECIS regional climate modeling system. International Journal Of Climatology 2009; 29: 2241–2255. • ONS – Operador Nacional do Sistema Elétrico. RE-3-128-2008 PEL 2009-2010; 2008. • ONS – Operador Nacional do Sistema Elétrico. 2ª Revisão Quadrimestral das Projeções da demanda de energia elétrica do Sistema Interligado Nacional 2014-2018. EPE - Empresa Pesquisa Energética/ONS; 2014. • ONS – Operador Nacional do Sistema Elétrico. Históricas da Operação - Geração de Energia; 2017. Available at: http://www.ons.org.br/historico/geracao_energia.aspx Accessed on 14/04/2017. • Tanajura C A S, Genz F, Araujo H. Mudanças climáticas e recursos hídricos na Bahia: validação da simulação do clima presente do HadRM3P e comparação com os cenários A2 e B2 para 2070-2100. Revista Brasileira de Meteorologia ed. online; 2010. vol. 3. p. 345-58. THANK YOU

Pieter de Jong.

Contact email: [email protected]

Universidade Federal da Bahia, Salvador-BA, Brazil. Escola Politécnica, Rua Aristides Novis, n 2, 6 Andar - Federaçáo, CEP 40.210-630.