Please wait a minute...
Journal of Arid Land  2021, Vol. 13 Issue (3): 205-223    DOI: 10.1007/s40333-021-0097-x
Research article     
Monthly and seasonal streamflow forecasting of large dryland catchments in Brazil
Alexandre C COSTA1,*(), Alvson B S ESTACIO2, Francisco de A de SOUZA FILHO2, Iran E LIMA NETO2
1Institute of Engineering and Sustainable Development, University of International Integration of the Afro-Brazilian Lusophony, Redenção, CEP 62.790-970, Brazil
2Department of Hydraulic Engineering and Environment, Federal University of Ceará, Fortaleza, CEP 60.451-970, Brazil
Download: HTML     PDF(627KB)
Export: BibTeX | EndNote (RIS)      

Abstract  

Streamflow forecasting in drylands is challenging. Data is scarce, catchments are highly human-modified and streamflow exhibits strong nonlinear responses to rainfall. The goal of this study was to evaluate the monthly and seasonal streamflow forecasting in two large catchments in the Jaguaribe River Basin in the Brazilian semi-arid area. We adopted four different lead times: one month ahead for monthly scale and two, three and four months ahead for seasonal scale. The gaps of the historic streamflow series were filled up by using rainfall-runoff modelling. Then, time series model techniques were applied, i.e., the locally constant, the locally averaged, the k-nearest-neighbours algorithm (k-NN) and the autoregressive model (AR). The criterion of reliability of the validation results is that the forecast is more skillful than streamflow climatology. Our approach outperformed the streamflow climatology for all monthly streamflows. On average, the former was 25% better than the latter. The seasonal streamflow forecasting (SSF) was also reliable (on average, 20% better than the climatology), failing slightly only for the high flow season of one catchment (6% worse than the climatology). Considering an uncertainty envelope (probabilistic forecasting), which was considerably narrower than the data standard deviation, the streamflow forecasting performance increased by about 50% at both scales. The forecast errors were mainly driven by the streamflow intra-seasonality at monthly scale, while they were by the forecast lead time at seasonal scale. The best-fit and worst-fit time series model were the k-NN approach and the AR model, respectively. The rainfall-runoff modelling outputs played an important role in improving streamflow forecasting for one streamgauge that showed 35% of data gaps. The developed data-driven approach is mathematical and computationally very simple, demands few resources to accomplish its operational implementation and is applicable to other dryland watersheds. Our findings may be part of drought forecasting systems and potentially help allocating water months in advance. Moreover, the developed strategy can serve as a baseline for more complex streamflow forecast systems.



Key wordsnonlinear time series analysis      probabilistic streamflow forecasting      reconstructed streamflow data      dryland hydrology      rainfall-runoff modelling      stochastic dynamical systems     
Received: 27 October 2020      Published: 10 March 2021
Corresponding Authors: Alexandre C COSTA     E-mail: cunhacos@unilab.edu.br
About author: * Alexandre C COSTA (E-mail: cunhacos@unilab.edu.br)
Cite this article:

Alexandre C COSTA, Alvson B S ESTACIO, Francisco de A de SOUZA FILHO, Iran E LIMA NETO. Monthly and seasonal streamflow forecasting of large dryland catchments in Brazil. Journal of Arid Land, 2021, 13(3): 205-223.

URL:

http://jal.xjegi.com/10.1007/s40333-021-0097-x     OR     http://jal.xjegi.com/Y2021/V13/I3/205

Fig. 1 Location of the Brazilian semi-arid area, the State of Ceará (Ceará) and the Jaguaribe River Basin (a), with the main reservoirs and streamgauges (b) used in this study
Combination Training set Validation set
I 1951-1979; 1990-2015 1980-1989
II 1951-1989; 2000-2015 1990-1999
III 1951-1999; 2010-2015 2000-2009
Table 1 Period combinations for training and validation sets, which were used for the applied cross-validation approach, given 65-a (1951-2015) streamflow time series at both IS and SS
Predicand Predictor
March Jan, Feb - - -
April Jan, Feb, Mar Jan, Feb - -
May Jan, Feb, Mar, Apr - Jan, Feb -
June Jan, Feb, Mar, Apr, May - - Jan, Feb
Lead time (month) 1 2 3 4
Table 2 Developed streamflow forecasting in the Jaguaribe River Basin after the combination of predicands, predictors and lead times.
Fig. 2 Average monthly hydrograph (1951-2015) of the Iguatu streamgauge (IS), with 3.7% of gap-filled streamflow data (29 months) using a rainfall-runoff model
Fig. 3 Average monthly hydrograph (1951-2015) of the Salgado streamgauge (SS), with 35.1% of gap-filled streamflow data (278 months) using a rainfall-runoff model
Fig. 4 SRMSE (standardized root mean square error) of the applied time series models in the validation set (1980-2010) for the monthly streamflow forecasting at the IS and SS
Model performance Mar_IS Mar_SS Apr_IS Apr_SS May_IS May_SS Jun_IS Jun_SS
Best-fit model 3-NN 4-NN 7-NN LC 2-NN 2-NN 5-NN 7-NN
SRMSE Validation 0.71 0.67 0.90 0.86 0.72 0.77 0.65 0.71
SRMSE Training 0.93 0.72 0.89 0.76 0.89 0.80 0.73 0.83
Table 3 SRMSE (standardized root mean square error) of the best-fit models in the validation set for the monthly streamflow forecasting at the Iguatu streamgauge (IS) and Salgado streamgauge (SS)
Fig. 5 Monthly deterministic forecasting in the validation set. The predicand is the streamflow in March at the SS. The predictors are the streamflow in January and February. The time series model, which is the best-fit one, is 4-NN. SRMSE (forecast error) in the validation set is 0.67, which means the 4-NN model is 33% better than the mean predictor (climatology). Here, SRMSE is the rms error divided by the SD (standard deviation) of the validation set.
Confidence interval March (92)* April (204)* May (79)* June (12)*
Length SRMSE Length SRMSE Length SRMSE Length SRMSE
33% 39 0.54 49 0.88 12 0.70 2 0.62
50% 72 0.43 93 0.84 31 0.66 3 0.59
66% 128 0.26 195 0.78 55 0.62 6 0.53
Table 4a Monthly probabilistic forecasting of the validation set at the IS
Confidence interval March (85)* April (123)* May (68)* June (14)*
Length SRMSE Length SRMSE Length SRMSE Length SRMSE
33% 26 0.58 32 0.78 12 0.73 4 0.66
50% 60 0.47 58 0.72 31 0.68 7 0.64
66% 89 0.40 101 0.67 63 0.60 13 0.59
Table 4b Monthly probabilistic forecasting of the validation set at the IS
Fig. 6 Monthly probabilistic forecasting in the validation set (30 a) with a confidence interval of 50%, which defines the upper and lower bound forecast. The predicand is the streamflow in May at the IS. The predictors are the streamflow in January, February, March and April. The time series model is 2-NN. SRMSE in the validation set is 0.66, which means the stochastic approach is comparatively 34% better than the mean predictor (climatology).
Fig. 7 SRMSE of the applied time series models in the validation set (1980-2010) for the seasonal streamflow forecasting (SSF) in April (Apr), May and June (Jun) at the IS and SS
Model performance Apr_IS Apr_SS May_IS May_SS Jun_IS Jun_SS
Best-fit model 6-NN 7-NN LA 7-NN 7-NN 6-NN
SRMSE validation 0.78 1.05 0.80 1.06 0.87 0.94
SRMSE trainning 0.95 1.03 1.60 1.03 1.04 1.00
Table 5 SRMSE of the best-fit models in the validation set for the seasonal streamflow forecasting (SSF) in April, May and June at the IS and SS
Fig. 8 Seasonal deterministic forecasting in the validation set (30 a). The predicand is the streamflow in April at the IS. The predictors are the streamflow in January and February. The time series model, which is the best-fit one, is 6-NN. SRMSE in the validation set is 0.78, which means the 6-NN model is 22% better than the mean predictor (climatology).
Confidence interval April (204)* May (79)* June (12)*
Length SRMSE Length SRMSE Length SRMSE
33% 81 0.67 40 0.56 5 0.78
50% 139 0.60 73 0.46 10 0.71
66% 235 0.48 98 0.38 16 0.60
Table 6a Seasonal probabilistic forecasting of the validation set at the IS
Confidence interval April (123)* May (68)* June (14)*
Length SRMSE Length SRMSE Length SRMSE
33% 54 0.95 24 0.93 5 0.86
50% 110 0.80 37 0.89 11 0.78
66% 164 0.67 58 0.83 18 0.71
Table 6b Seasonal probabilistic forecasting of the validation set at the SS
[1]   Araújo J A de A. 1990. Dams in Northeastern Brazil: an experience in the semi-arid region (2nd ed). Fortaleza: DNOCS, 328.
[2]   Bennett J C, Wang Q J, Robertson D E, et al. 2017. Assessment of an ensemble seasonal streamflow forecasting system for Australia. Hydrology and Earth System Sciences, 21:6007-6030.
[3]   Block P J, Souza Filho F A, Sun L, et al. 2009. A streamflow forecasting framework using multiple climate and hydrological models. Journal of the American Water Resources Association, 45(4):828-843.
[4]   Collischon W, Tucci C E M, Clarke R T, et al. 2007. Medium-range reservoir inflow predictions based on quantitative precipitation forecasts. Journal of Hydrology, 344(1-2):112-122.
[5]   Costa A C, Bronstert A, de Araújo J C. 2012a. A channel transmission losses model for different dryland rivers. Hydrology and Earth System Sciences, 16(4):1111-1135.
[6]   Costa A C, Bronstert A, Kneis D. 2012b. Probabilistic flood forecasting for a mountainous headwater catchment using a nonparametric stochastic dynamic approach. Hydrological Sciences Journal, 57(1):10-25.
[7]   Costa A C, Foerster S, de Araújo J C, et al. 2013. Analysis of channel transmission losses in a dryland river reach in north-eastern Brazil using streamflow series, groundwater level series and multi-temporal satellite data. Hydrological Processes, 27(7):1046-1060.
[8]   Crochemore L, Ramos M H, Pappenberger F. 2016. Bias correcting precipitation forecasts to improve the skill of seasonal streamflow forecasts. Hydrology and Earth System Sciences, 20(9):3601-3618.
[9]   Delgado J M, Voss S, Bürger G, et al. 2018. Seasonal drought prediction for semiarid northeast Brazil: Verification of six hydro-meteorological forecast products. Hydrology and Earth System Sciences, 22(9):5041-5056.
[10]   Fioreze A P, Bubel A P M, Callou A É P, et al. 2012. The water issue in the Northeast Brasília: Ministry of Science and Technology. [2020-01-24]. http://livroaberto.ibict.br/handle/1/669.
[11]   Formiga-Johnsson R M, Kemper K. 2005. Institutional and Policy Analysis of River Basin Management: The Jaguaribe River Basin, Ceará, Brazil. New York: Social Science Research Network, 42.
[12]   Frischkorn H, Santiago M F, de Araújo J C. 2003. Water resources of Ceará and Piauí. In: Gaiser T, Krol M, Frischkorn H, et al, Global Change and Regional Impacts. Berlin: Springer-Verlag, 87-94.
[13]   FUNCEME (Foundation for Meteorology and Water Resources of the State of Ceará). 2008. Mapping of the surface of the water bodies in Brazil. Fortaleza: FUNCEME, 108.
[14]   Güntner A, Bronstert A. 2004. Representation of landscape variability and lateral redistribution processes for large-scale hydrological modelling in semi-arid areas. Journal of Hydrology, 297(1-4):136-161.
[15]   Güntner A, Krol M S, de Araújo J C, et al. 2004. Simple water balance modelling of surface reservoir systems in a large data-scarce semiarid region. Hydrological Sciences Journal, 49(5):901-918.
[16]   Gutierrez A P A, Engle N L, de Nys E, et al. 2014. Drought preparedness in Brazil. Weather and Climate Extremes, 3:95-106.
[17]   Hastenrath S, Heller L. 1977. Dynamics of climatic hazards in northeast Brazil. Quarterly Journal of the Royal Meteorological Society, 103(435):77-92.
[18]   He Z, Wen X, Liu H, et al. 2014. A comparative study of artificial neural network, adaptive neuro fuzzy inference system and support vector machine for forecasting river flow in the semiarid mountain region. Journal of Hydrology, 509:379-386.
[19]   Kalra A, Miller W P, Lamb K W, et al. 2012. Using large-scale climatic patterns for improving long lead time streamflow forecasts for Gunnison and San Juan River Basins. Hydrological Processes, 27(11):1543-1559.
[20]   Kalra A, Li L, Li X, et al. 2013. Improving streamflow forecast lead time using oceanic-atmospheric oscillations for Kaidu River Basin, Xinjiang, China. Journal of Hydrologic Engineering, 18(8):1031-1040.
[21]   Kantz H, Schreiber T. 2004. Nonlinear Time Series Analysis (2nd ed). Cambridge: Cambridge University Press, 388.
[22]   Kirchner J W. 2009. Catchments as simple dynamical systems: Catchment characterization, rainfall-runoff modelling, and doing hydrology backward. Water Resources Research, 45(2):1-34.
[23]   Klemeš V. 1986. Operational testing of hydrological simulation models. Hydrological Sciences Journal, 31(1):13-24.
[24]   Koutsoyiannis D. 2005. Hydrologic persistence and the hurst phenomenon. In: Lehr J H, Keely J. The Encyclopedia of Water. New York: John Wiley & Sons, Inc., 27.
[25]   Kwon H H, Souza Filho F A, Sun L, et al. 2012. Uncertainty assessment of hydrologic model and climate forecast model in Northern Brazil. Hydrological Processes, 26(25):3875-3885.
[26]   Lopes J C, Braga J B F, Conejo J L. 1981. Hydrological simulation: Applications of a simplified model. In: III Brazilian Symposium for Water Resources. Fortaleza: ABRH, 42-62.
[27]   Machado C J F, Santiago M M F, Mendonça L A R, et al. 2007. Hydrogeochemical and flow modeling of aquitard percolation in the cariri valley-northeast Brazil. Aquatic Geochemistry, 13:187-196.
[28]   Mamede G L, Araújo N A M, Schneider C M, et al. 2012. Overspill avalanching in a dense reservoir network. PNAS, 109(19):7191-7195.
[29]   Moradkhani H, Meier M. 2010. Long-lead water supply forecast using large-scale predictors and independent component analysis. Journal of Hydrology Engineering, 15(10):744-762.
[30]   Moura A D, Shukla J. 1981. On the dynamics of droughts in northeast Brazil: Observations, theory, and numerical experiments with a general circulation model. Journal of the Atmospheric Sciences, 38(12):2653-2675.
[31]   Nobre P, Shukla J. 1996. Variations of sea surface temperature, wind stress, and rainfall over the tropical Atlantic and South America. Journal of Climate, 9(10):2464-2479.
[32]   Nunes C M. 2012. Project of integration of the San Francisco River with the watersheds in the Northeast - PISF. In: The water issue in the Northeast. Brasília: Ministry of Science and Technology. [2020-01-24]. http://livroaberto.ibict.br/handle/1/669.
[33]   Perreti C, Munch S, Sugihara G. 2013. Model-free forecasting outperforms the correct mechanistic model for simulated and experimental data. PNAS, 110(13):5253-5257.
pmid: 23440207
[34]   Philander S G. 1990. El Niño, La Niña, and the Southern Oscillation. International Geophysics Series (vol. 46), San Diego: Academic Press, 293
[35]   Pilz T, Voss S, Vormoor K, et al. 2019. Seasonal drought prediction for semiarid northeast Brazil: what is the added value of a process-based hydrological model? Hydrology and Earth System Sciences, 23:1951-1971.
[36]   Porporato A, Ridolfi L. 2001. Multivariate nonlinear prediction of river flows. Journal of Hydrology, 248(1-4):109-122.
[37]   Rao V B, Hada K. 1990. Characteristics of rainfall over Brazil: annual variations and connections with southern oscillation. Theoretical and Applied Climatology, 42:81-91.
[38]   Robertson D E, Wang Q J. 2012. A Bayesian approach to predictor selection for seasonal streamflow forecasting. Journal of Hydrometeorology, 13(1):155-171.
[39]   Rodrigues R R, Haarsma R J, Campos E J D, et al. 2011. The impacts of inter-El Nino variability on the Tropical Atlantic and Northeast Brazil climate. Journal of Climate, 24(13):3402-3422.
[40]   Seibert M, Merz B, Apel H. 2017. Seasonal forecasting of hydrological drought in the Limpopo Basin: a comparison of statistical methods. Hydrology and Earth System Sciences, 21:1611-1629.
[41]   Shukla S, Lettenmaier D P. 2011. Seasonal hydrologic prediction in the United States: understanding the role of initial hydrologic conditions and seasonal climate forecast skill. Hydrology and Earth System Sciences, 15:3529-3538.
[42]   Sittichok K, Gado D A, Seidou O, et al. 2016. Statistical seasonal rainfall and streamflow forecasting for the Sirba watershed, West Africa, using sea-surface temperatures. Hydrological Sciences Journal, 61(5):805-815.
[43]   Sivakumar B, Singh V P. 2012. Hydrologic system complexity and nonlinear dynamic concepts for a catchment classification framework. Hydrology and Earth System Sciences, 16:4119-4131.
[44]   Souza Filho F A, Lall U. 2003. Seasonal to interannual ensemble streamflow forecasts for Ceara, Brazil: Applications of a multivariate, semiparametric algorithm. Water Resources Research, 39(11):1-13.
[45]   SUDENE ( Superintendência de Desenvolvimento do Nordeste ). 1980. Plan for the integrated utilization of the water resources in Northeastern Brazil. Recife: SUDENE, 712.
[46]   Takens F. 1980. Detecting strange attractors in turbulence. In: Rang D, and Young L S. Lecture notes in mathematics. Berlin: Springer, 366-381.
[47]   Tongal H. 2020. Comparison of local and global approximators in multivariate chaotic forecasting of daily streamflow. Hydrological Sciences Journal, 65(7):1129-1144.
[48]   Trambauer P, Werner M, Winsemius H C, et al. 2015. Hydrological drought forecasting and skill assessment for the Limpopo River basin, southern Africa. Hydrology and Earth System Sciences, 19:1695-1711.
[49]   Uvo C B, Repelli C A, Zebiak S E, et al. 1998. The relationships between tropical Pacific and Atlantic SST and northeast Brazil monthly precipitation. Journal of Climate, 11(4):551-562.
[50]   Vrugt J A, ter Braak C J F, Clark M P, et al. 2008. Treatment of input uncertainty in hydrologic modeling: Doing hydrology backward with Markov chain Monte Carlo simulation. Water Resources Research, 44(12):1-15.
[51]   Vrugt J A, ter Braak C J F, Diks C G H, et al. 2009. Accelerating Markov chain Monte Carlo simulation by differential evolution with self-adaptive randomized subspace sampling. International Journal of Nonlinear Sciences and Numerical Simulation, 10(3):273-290.
[52]   Vrugt J A. 2016. Markov chain Monte Carlo simulation using the DREAM software package: Theory, concepts, and MATLAB implementation. Environmental Modelling and Software, 75:273-316.
[53]   Werner P C, Gerstengarbe F W. 2003. The climate of Piauí and Ceará. In: Gaiser T, Krol M, Frischkorn H, et al. Global Change and Regional Impacts. Berlin: Springer-Verlag, 81-86.
[54]   Wilks D S. 2005. Statistical Methods in the Atmospheric Sciences. Burlington: Academic Press, 704.
[55]   Wu C L, Chau K W, Li Y S. 2009. Predicting monthly streamflow using data-driven models coupled with data-preprocessing techniques. Water Resources Research, 45(8), W08432, doi: 10.1029/2007WR006737.
[56]   Wu C L, Chau K W. 2010. Data-driven models for monthly streamflow time series prediction. Engineering Applications of Artificial Intelligence, 23(8):1350-1367.
[57]   Yassen Z M, Jaafar O, Deo R C, et al. 2016. Stream-flow forecasting using extreme learning machines: A case study in a semi-arid region in Iraq. Journal of Hydrology, 542:603-614.
[58]   Yossef N C, Winsemius H, Weerts A, et al. 2013. Skill of a global seasonal streamflow forecasting system, relative roles of initial conditions and meteorological forcing. Water Resources Research, 49(8):4687-4699.
[59]   Yuan X. 2016. An experimental seasonal hydrological forecasting system over the Yellow River basin - Part 2: The added value from climate forecast models. Hydrology and Earth System Sciences, 20:2453-2466.
[60]   Yuan X, Ma F, Wang L, et al. 2016a. An experimental seasonal hydrological forecasting system over the Yellow River basin - Part 1: Understanding the role of initial hydrological conditions. Hydrology and Earth System Sciences, 20:2437-2451.
No related articles found!