See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/349869618 Can hydrological drought be efficiently predicted by conceptual rainfall-runoff models with global data products? Article  in  Journal of Natural Resources and Development · February 2021 CITATIONS 0 READS 141 7 authors, including: Some of the authors of this publication are also working on these related projects: Development of functional eco-hydrological indicators to assess the connectivity between slopes and wetlands in a wet Páramo micro-catchment View project LUCCi - Land Use and Climate Change Interactions in Central Vietnam View project Nelson Venegas Warsaw University of Life Sciences - SGGW 6 PUBLICATIONS   1 CITATION    SEE PROFILE Christian Birkel University of Costa Rica 162 PUBLICATIONS   2,442 CITATIONS    SEE PROFILE Juan Diego Giraldo-Osorio Pontificia Universidad Javeriana 42 PUBLICATIONS   317 CITATIONS    SEE PROFILE All content following this page was uploaded by Nelson Venegas on 07 March 2021. The user has requested enhancement of the downloaded file. https://www.researchgate.net/publication/349869618_Can_hydrological_drought_be_efficiently_predicted_by_conceptual_rainfall-runoff_models_with_global_data_products?enrichId=rgreq-3c855277c5aef8cec34c3e9e0f6bb1b0-XXX&enrichSource=Y292ZXJQYWdlOzM0OTg2OTYxODtBUzo5OTg3NjQyNzc5NDQzMjJAMTYxNTEzNTM2ODY3OA%3D%3D&el=1_x_2&_esc=publicationCoverPdf https://www.researchgate.net/publication/349869618_Can_hydrological_drought_be_efficiently_predicted_by_conceptual_rainfall-runoff_models_with_global_data_products?enrichId=rgreq-3c855277c5aef8cec34c3e9e0f6bb1b0-XXX&enrichSource=Y292ZXJQYWdlOzM0OTg2OTYxODtBUzo5OTg3NjQyNzc5NDQzMjJAMTYxNTEzNTM2ODY3OA%3D%3D&el=1_x_3&_esc=publicationCoverPdf https://www.researchgate.net/project/Development-of-functional-eco-hydrological-indicators-to-assess-the-connectivity-between-slopes-and-wetlands-in-a-wet-Paramo-micro-catchment?enrichId=rgreq-3c855277c5aef8cec34c3e9e0f6bb1b0-XXX&enrichSource=Y292ZXJQYWdlOzM0OTg2OTYxODtBUzo5OTg3NjQyNzc5NDQzMjJAMTYxNTEzNTM2ODY3OA%3D%3D&el=1_x_9&_esc=publicationCoverPdf https://www.researchgate.net/project/LUCCi-Land-Use-and-Climate-Change-Interactions-in-Central-Vietnam-2?enrichId=rgreq-3c855277c5aef8cec34c3e9e0f6bb1b0-XXX&enrichSource=Y292ZXJQYWdlOzM0OTg2OTYxODtBUzo5OTg3NjQyNzc5NDQzMjJAMTYxNTEzNTM2ODY3OA%3D%3D&el=1_x_9&_esc=publicationCoverPdf https://www.researchgate.net/?enrichId=rgreq-3c855277c5aef8cec34c3e9e0f6bb1b0-XXX&enrichSource=Y292ZXJQYWdlOzM0OTg2OTYxODtBUzo5OTg3NjQyNzc5NDQzMjJAMTYxNTEzNTM2ODY3OA%3D%3D&el=1_x_1&_esc=publicationCoverPdf https://www.researchgate.net/profile/Nelson-Venegas-2?enrichId=rgreq-3c855277c5aef8cec34c3e9e0f6bb1b0-XXX&enrichSource=Y292ZXJQYWdlOzM0OTg2OTYxODtBUzo5OTg3NjQyNzc5NDQzMjJAMTYxNTEzNTM2ODY3OA%3D%3D&el=1_x_4&_esc=publicationCoverPdf https://www.researchgate.net/profile/Nelson-Venegas-2?enrichId=rgreq-3c855277c5aef8cec34c3e9e0f6bb1b0-XXX&enrichSource=Y292ZXJQYWdlOzM0OTg2OTYxODtBUzo5OTg3NjQyNzc5NDQzMjJAMTYxNTEzNTM2ODY3OA%3D%3D&el=1_x_5&_esc=publicationCoverPdf https://www.researchgate.net/institution/Warsaw-University-of-Life-Sciences-SGGW?enrichId=rgreq-3c855277c5aef8cec34c3e9e0f6bb1b0-XXX&enrichSource=Y292ZXJQYWdlOzM0OTg2OTYxODtBUzo5OTg3NjQyNzc5NDQzMjJAMTYxNTEzNTM2ODY3OA%3D%3D&el=1_x_6&_esc=publicationCoverPdf https://www.researchgate.net/profile/Nelson-Venegas-2?enrichId=rgreq-3c855277c5aef8cec34c3e9e0f6bb1b0-XXX&enrichSource=Y292ZXJQYWdlOzM0OTg2OTYxODtBUzo5OTg3NjQyNzc5NDQzMjJAMTYxNTEzNTM2ODY3OA%3D%3D&el=1_x_7&_esc=publicationCoverPdf https://www.researchgate.net/profile/Christian-Birkel?enrichId=rgreq-3c855277c5aef8cec34c3e9e0f6bb1b0-XXX&enrichSource=Y292ZXJQYWdlOzM0OTg2OTYxODtBUzo5OTg3NjQyNzc5NDQzMjJAMTYxNTEzNTM2ODY3OA%3D%3D&el=1_x_4&_esc=publicationCoverPdf https://www.researchgate.net/profile/Christian-Birkel?enrichId=rgreq-3c855277c5aef8cec34c3e9e0f6bb1b0-XXX&enrichSource=Y292ZXJQYWdlOzM0OTg2OTYxODtBUzo5OTg3NjQyNzc5NDQzMjJAMTYxNTEzNTM2ODY3OA%3D%3D&el=1_x_5&_esc=publicationCoverPdf https://www.researchgate.net/institution/University-of-Costa-Rica?enrichId=rgreq-3c855277c5aef8cec34c3e9e0f6bb1b0-XXX&enrichSource=Y292ZXJQYWdlOzM0OTg2OTYxODtBUzo5OTg3NjQyNzc5NDQzMjJAMTYxNTEzNTM2ODY3OA%3D%3D&el=1_x_6&_esc=publicationCoverPdf https://www.researchgate.net/profile/Christian-Birkel?enrichId=rgreq-3c855277c5aef8cec34c3e9e0f6bb1b0-XXX&enrichSource=Y292ZXJQYWdlOzM0OTg2OTYxODtBUzo5OTg3NjQyNzc5NDQzMjJAMTYxNTEzNTM2ODY3OA%3D%3D&el=1_x_7&_esc=publicationCoverPdf https://www.researchgate.net/profile/Juan-Diego-Giraldo-Osorio?enrichId=rgreq-3c855277c5aef8cec34c3e9e0f6bb1b0-XXX&enrichSource=Y292ZXJQYWdlOzM0OTg2OTYxODtBUzo5OTg3NjQyNzc5NDQzMjJAMTYxNTEzNTM2ODY3OA%3D%3D&el=1_x_4&_esc=publicationCoverPdf https://www.researchgate.net/profile/Juan-Diego-Giraldo-Osorio?enrichId=rgreq-3c855277c5aef8cec34c3e9e0f6bb1b0-XXX&enrichSource=Y292ZXJQYWdlOzM0OTg2OTYxODtBUzo5OTg3NjQyNzc5NDQzMjJAMTYxNTEzNTM2ODY3OA%3D%3D&el=1_x_5&_esc=publicationCoverPdf https://www.researchgate.net/institution/Pontificia_Universidad_Javeriana?enrichId=rgreq-3c855277c5aef8cec34c3e9e0f6bb1b0-XXX&enrichSource=Y292ZXJQYWdlOzM0OTg2OTYxODtBUzo5OTg3NjQyNzc5NDQzMjJAMTYxNTEzNTM2ODY3OA%3D%3D&el=1_x_6&_esc=publicationCoverPdf https://www.researchgate.net/profile/Juan-Diego-Giraldo-Osorio?enrichId=rgreq-3c855277c5aef8cec34c3e9e0f6bb1b0-XXX&enrichSource=Y292ZXJQYWdlOzM0OTg2OTYxODtBUzo5OTg3NjQyNzc5NDQzMjJAMTYxNTEzNTM2ODY3OA%3D%3D&el=1_x_7&_esc=publicationCoverPdf https://www.researchgate.net/profile/Nelson-Venegas-2?enrichId=rgreq-3c855277c5aef8cec34c3e9e0f6bb1b0-XXX&enrichSource=Y292ZXJQYWdlOzM0OTg2OTYxODtBUzo5OTg3NjQyNzc5NDQzMjJAMTYxNTEzNTM2ODY3OA%3D%3D&el=1_x_10&_esc=publicationCoverPdf Journal of Natural Resources and Development 2021; 02: 01 - 18DOI number: 10.5027/jnrd.v11i0.02 1 JOURNAL OF NATURAL RESOURCES AND DEVELOPMENT Can hydrological drought be efficiently predicted by conceptual rainfall-runoff models with global data products? Nelson Venegas-Cordero1, 6, Christian Birkel1,2, Juan Diego Giraldo-Osorio3, Alicia Correa-Barahona1, Ana María Duran-Quesada4, Rafael Arce-Mesen1, Alexandra Nauditt5. 1 Department of Geography and Water and Global Change Observatory, University of Costa Rica, San José, Costa Rica. 2 Northern Rivers Institute, University of Aberdeen, Aberdeen, AB24 3UF, UK. 3 Grupo de Investigación Ciencia e Ingeniería del Agua y el Ambiente, Departamento de Ingeniería Civil, Facultad de Ingeniería, Pontificia Universidad Javeriana, Bogotá, Colombia. 4 Atmospheric, Oceanic and Planetary Physics Department & Climate System Observation Laboratory, School of Physics and Center for Geophysical Research, University of Costa Rica, San José, Costa Rica. 5 Institute for Technology and Resources Management in the Tropics and Subtropics, Cologne University of Applied Sciences, Cologne, Germany. 6 Department of Hydrology, Meteorology and Water Management, Warsaw University of Life Sciences, Nowoursynowska 166, 02-787, Warsaw, Poland Corresponding author: nelson_venegas@sggw.edu.pl Article history Received 17.06.2020 Accepted 29.10.2020 Published 27.02.2021 Keywords Hydrological Drought Tropics HBV-Light Global Precipitation Products Regional Climate Mode Costa Rica. Abstract The seasonally-dry tropics of northern Costa Rica are characterized by recurrent drought events with negative socio-economic impacts on a vulnerable population. Scarce hydroclimatic obser- vational data constraints reasonable water management and often results in water scarcity issues. This study analyses hydrological drought situations using freely available Global Precipitation Products (GPP) and a Regional Climate Model (RCM) that drive a relatively simple, semi-distributed rainfall-runoff model (HBV-Light). Firstly, the GPP and observed rainfall were used to calibrate the model simulating streamflow dynamics. Secondly, drought detection and estimates of drought du- ration, intensity and severity were determined with a daily variable threshold approach. Thirdly, we developed future hydrological drought scenarios based on a RCM. Generally, the GPP CHIRPS (Cli- mate Hazards Group InfraRed Precipitation with Station Data) resulted in the best streamflow sim- ulations (KGE > 0.6) compared with the model driven by observed rainfall (KGE > 0.7). CHIRPS also correctly identified the observed streamflow drought periods of 1994, 1997-1998 and 2001-2002. Average observed streamflow drought severity was 27.9 mm compared to the CHIRPS-derived se- verity of 20.6 mm (error estimate of ±7.3 mm). The model in combination with global data can be successfully used to identify drought periods and their duration, but model uncertainty currently prevents from forecasting streamflow deficit with volume errors below 50%. The future hydrolog- ical drought scenario showed more severe drought periods between 2039-2041 and 2042-2043. This study responds to the need for drought assessments in the seasonally-dry tropics with scarce observations as a tool for adaptation to climate change and water resource management. Research article Journal of Natural Resources and Development 2021; 02: 01 - 18DOI number: 10.5027/jnrd.v11i0.02 2 1 Introduction Drought is a recurrent extreme climate event characterized by be- low-normal precipitation (van Loon, 2015), that can occur global- ly, in contrast to the permanent aridity that prevails in arid zones (Dai, 2011). Droughts are mainly classified into meteorological, agricultural, hydrological, and socio-economic events (Van Loon, 2015). Meteorological drought is related to precipitation deficit during a prolonged period, while the agricultural drought is the reduction of water availability in the soil for vegetation use, re- sulting in crop losses (Van Loon, 2015). Hydrological drought is defined as the deficit of streamflow and/or volume of surface-or groundwater levels (Marcos, 2001). Finally, the socio-econom- ic drought characterizes the impacts by each previous types of drought (Van Loon, 2015). Drought analysis suffers from inaccessible and unreliable hydro- meteorological in-situ observations particularly in tropical regions where hydrological droughts are recurrent events and are expect- ed to become more frequent in the future (Nauditt et al., 2017). Drought research has so far been less focused in tropical regions than in other geographical areas (Nauditt & Ribbe, 2017), limiting the prospects of a robust monitoring (Ndehedehe et al., 2016), and constraining the understanding of spatio-temporal dynamics of drought occurrence at different spatial and temporal scales. To partly overcome this data scarcity, freely available Global Precip- itation Products (GPP) provide spatially and temporally precipi- tation estimates in relation to the sparse in situ measurements. GPP together with Regional Climate Models (RCM) that forecast weather conditions are useful input information for rainfall-run- off models. This assembly can help to characterize and forecast drought events (Van Loon, 2015), with the aim to improve water resources management and to reduce their related socio-eco- nomic impacts. Many studies have evaluated the performance of GPP (Behrangi et al., 2011; Thiemig et al., 2012; Adler et al., 2001; Shen et al., 2010) and their ability to generate accurate precipi- tation estimates for hydrological modelling (Artan et al., 2007; Liechti et al., 2012; Maggioni & Massari, 2018). Zambrano-Bigiarini et al., (2017) highlighted the opportunities of using GPP for hydro- logical applications such as modelling and evaluation of droughts and floods. Moreover, conceptual hydrological models are consid- ered the main tools for drought assessment (Van Loon, 2015). In a context of data scarcity, such as in Central America, the use of such technologies to monitor drought‐related variables, forecast and quantify drought impacts (AghaKouchak et al., 2015) should be perceived as an opportunity to improve the understanding of droughts at different spatial and temporal scales. Central America is climatologically influenced by the Caribbean Sea and the Eastern tropical Pacific Ocean weather patterns re- sulting in a pronounced seasonality of precipitation and flow re- gimes. This seasonality presents periods of water deficit during the dry season, that in a context of climatic variability is often con- sidered as drought. The region has historically been affected by droughts with negative consequences for the agricultural sector, impacting the food security of the population (Calvo-Solano et al., 2018; Quesada-Montano et al., 2018). One of the most affected locations is found across the Central American Dry Corridor (CADC), where the dry season lasts on average 5 months. This area is inhab- ited by 50% of the total population and 67% of the rural population living in poverty (Gotlieb et al., 2019) and socio-economic vulner- ability (Calvo-Solano et al., 2018). One of the main concerns is that the unregulated water use for industry and irrigation, affects this population particularly during the dry season (Guzmán, 2013). In northern Costa Rica – a part of the CADC - recent droughts resulted in drinking water supply issues and economic losses (Birkel, 2006). Our study site, the Tempisque catchment located in the CADC is one of the most vulnerable regions to climate change in the country (Hi- dalgo & Alfaro, 2012) and often affected by hydrological droughts. However, the aim of this research is to evaluate past and near-future hydrological drought in this catchment, using GPP, RCM and the rel- atively parsimonious hydrological model HBV-Light. The specific objectives are to: I Analyze the performance of GPP used as input to the HBV- Light model in a gauged Tempisque River sub-catchment. II Perform a hydrological drought analysis calculating severity, duration, and intensity for a historic period that covered ob- served and documented extreme droughts. III Analyze the predictive capability of the calibrated model to estimate future hydrological drought scenarios using climate projections of precipitation from a Regional Climate Model for the Central American and Caribbean domain. 2 Methods 2.1 Study site: The Tempisque catchment The Tempisque-Bebedero catchment (5 400 km2) is located in the Guanacaste province, north-west Costa Rica (Figure 1), extending from the Central Volcanic cordillera to the Gulf of Nicoya (Pacific Ocean) (Monge & Gómez, 2007). This catchment with mean ele- vation of 260 m.a.s.l. and a mean slope of 6° exhibits rivers with a low transport capacity particularly in the lowlands and flood- plains. Their tributaries originate in the foothills of the Guanacaste Volcanic Range, which has maximum altitudes of around 1900 m.a.s.l. (Barrantes, 2010). Although forests and wetlands were the original land cover of the catchment, forests have been partly re- placed by crops, pastures and urban areas. The extension of pro- tected areas has increased over the past twenty years and today, the only intensification of agricultural activities is related to cash crops such as sugar cane, rice and melon (Mateo-Vega, 2001). The study catchment is geologically composed of Tertiary and Quater- nary materials, with a predominance of young Quaternary volcan- ic rocks (Guzmán, 2013). The Tempisque sub-catchment used for modelling purposes is de- fined by the discharge gauging station La Guardia (Figure 1, right Journal of Natural Resources and Development 2021; 02: 01 - 18DOI number: 10.5027/jnrd.v11i0.02 3 Figure 1: The right panel shows the topographic representation and drainage network of the gauged Tempisque subcatchment, Guanacaste, Costa Rica. The left panel shows a CHIRPS V2.0 interpolated mean annual precipitation (1981-2017) map for the Tempisque-Bebedero catchment. The green points indicate meteorological stations and black point shows the streamflow gauging station La Guardia. The bottom panel represents the monthly precipitation (blue bars) and streamflow (dotted red line) regime of the Tempisque sub-catchment for 1993 to 2003. panel) and drains an area of 990 km2 (Table 1) (Guzmán, 2013). Here, twelve soil subgroups have been identified, most of them part of two taxonomic orders: entisols and inceptisols (56% and 34% of the catchment area, respectively) (Guzmán, 2013; ITCR, 2004). A seasonal tropical climate with a clear cycle of dry and wet peri- ods dominates the Tempisque catchment (Birkel et al., 2017). The Caribbean climate influences the upper part of the catchment close to the continental divide with consistent rain and high rel- ative humidity throughout the year (Figure 1). The Pacific Ocean influence is associated with rainfall events mainly between May and November and a marked dry season from December to April (Duran-Quesada et al., 2010). The dry season presents low to zero precipitation (Figure 1, bottom panel) with maximum values in September and October. The monthly temperature varies be- tween 30.2°C and 24.6°C throughout the year (Mateo-Vega, 2001). The streamflow regime closely follows the climatic variability as shown in Figure 1, bottom panel. Journal of Natural Resources and Development 2021; 02: 01 - 18DOI number: 10.5027/jnrd.v11i0.02 4 2.2 Data 2.2.1 Ground-based and Global Precipitation data sources The Center for Geophysical Research (CIGEFI) of the University of Costa Rica provided daily in situ precipitation data from 14 meteor- ological stations. Daily streamflow time series (1993-2003) from the La Guardia station were provided by the National Electricity Insti- tute (ICE). The potential evapotranspiration (PET) was derived from the Moderate Resolution Imaging Spectroradiometer (MODIS;Kan- niah et al., 2009), and spatially averaged to represent the mean catchment PET. The spatio-temporal resolution and available data of the GPP products CHIRPS v2.0, MSWEPv2.1, PGFv3 and TRMM- 3B42v7 are summarized in Table 2 and are briefly described below. All data sets were spatially averaged for the Tempisque catchment and for the period from 1993 to 2003 for model analysis. The study period was selected according to the most reliable streamflow data record. The benchmark model used a single best precipita- tion station close to the streamflow gauging station (Figure 1) after initial tests with different spatial interpolations of all available 14 stations that performed worse in hydrological modelling. CHIRPS v2.0 The Climate Hazards Group InfraRed Precipitation with Station Data (https://data.chc.ucsb.edu/products/CHIRPS-2.0/) version 2.0 has a relatively high spatial resolution of 0.05° and covers from 50°S to 50°N. The product uses an “intelligent interpolation” approach, which works with anomalies in high-resolution climatology. It also contains a daily, pentadal and monthly precipitation estimate from 1981 to present and was designed primarily for monitoring agri- cultural drought and global change (Funk et al., 2015). MSWEPv2.1 The Multi-Source Weighted-Ensemble Precipitation (http://www. gloh2o.org) version 2.1 provides daily precipitation data from 1979 to 2018, with a spatial resolution of 0.10°. It was designed to analyze and understand differences in precipitation, as well as hydroclimatic processes to improve the performance of hydrolog- ical models (Beck et al., 2019). PGFv3 The PGFv3 (http://hydrology.princeton.edu/data/pgf/v3/0.25deg/ daily/) as an improved global precipitation product from Prince- ton University, that presents information combined through a re-analysis of data from various products, such as from the Nation- al Center for Environmental Prediction for Atmospheric Research (NCEP-NCAR) and the TRMM mission. The spatial resolution is 0.25° within a time window from 1979 to 2010 (Zambrano- Bigia- rini et al., 2017). Descriptor Units Tempisque Precipitation Min mm/ year 1517 Max mm/ year 3881 Mean mm/ year 1930 Area Total km2 990 Geology - - Tertiary and Quaternary materials (Quaternary volcanic rocks) Land cover - - Forest, wetland, crops, pastures and urban areas Soils - - Taxonomic orders: Entisols and inceptisols Elevation Min m.a.s.l. 18 Max m.a.s.l. 1916 Mean m.a.s.l 332 Product Spatial Resolution Temporal Resolution Time period CHIRPS v2.0 0.05° Daily 1981- Current MSWEPv2.1 0.10° Daily 1979-2018 PGFv3 0.25° Daily 1979-2010 TRMM-3B42v7 0.25° Daily 1998-2019 Table 1: Precipitation and physical characteristics of the Tempisque subcatchment. Table 2: Summary of the tested daily Global Precipitation Products. Journal of Natural Resources and Development 2021; 02: 01 - 18DOI number: 10.5027/jnrd.v11i0.02 5 TRMM-3B42v7 The Tropical Rainfall Measuring Mission (TRMM) global precip- itation product (https://giovanni.gsfc.nasa.gov/giovanni/#ser- vice=MpAn&starttime=&endtime=&data=TRMM_3B42_Dai- ly_7_precipitation) provides an estimate of the quasi-global precipitation of a wide variety of modern global products that relate to satellite-borne rainfall. It presents a spatial resolution of 0.25° with an extension of 50°N-S from 1998 to present (Huffman et al., 2010). The Regional Climate Model (RCM) projections The Regional Climate Model (RCM) has a spatial domain covering northern Central America and the Caribbean, and uses the coupled system ROM, which consists of a regional atmosphere model cou- pled to a global ocean and is based on the CORDEX (Coordinate Regional Downscaling Experiment) spatial domain (Cabos et al., 2018). REMO, is the Regional atmosphere Model with prognosis of the surface pressure, horizontal wind components, rainfall, temper- ature, water vapor, and cloud ice (Sein et al., 2015). In ROM, REMO is coupled with the global ocean model, MPIOM that includes sea-ice, marine biogeochemistry modules and a global hydrological dis- charge model and was dynamically downscaled matching the cli- matic features of Central America (Cabos et al., 2018). Therefore, we directly used the ROM near-future climate projections as a possible future scenario for hydrological modelling until 2045. After different tests with regional climate models detected that none was suitable for calibrating the hydrological model. How- ever, the in situ precipitation from a meteorological station in Tempisque sub-catchment was used to compute the benchmark calibration model used for further evaluations. 2.2.2 Hydrological modelling and drought assessment The conceptual rainfall-runoff model HBV-Light The HBV-Light model (https://www.geo.uzh.ch/en/units/h2k/Servic- es/HBV-Model/HBV-Download.html) is a conceptual, semi-distributed model, widely used in hydrological forecast and water balance stud- ies (Paredes et al., 2014). The model structure (Figure 2) incorporates the most important runoff generation processes with relatively few parameters (see Seibert & Vis, 2012 and Table 3 for more details). The model uses a soil water redistribution reservoir (3 calibrated parame- ters: FC, LP and BETA) that determines the volume of water recharging (PERC) the two runoff generation reservoirs from which streamflow is generated. Runoff is generated from an upper, quicker (ALPHA and K1) and a lower, slower (K2) reservoir. A simple triangular algorithm with one parameter (MAXBAS) is used for streamflow routing. The previous- ly described GPP were used as model input and tested against an in-si- tu observational benchmark, in order to determine the most suitable product for drought analyses. A one-year warm-up period from 01 Jan- uary 1993 to 01 January 1994 was used, allowing the model reservoirs to evolve from initial conditions (Montalván, 2017). Figure 2: Structure of the conceptual hydrological model HBV-Light with calibrated parameters in red according to Seibert & Vis (2012). Journal of Natural Resources and Development 2021; 02: 01 - 18DOI number: 10.5027/jnrd.v11i0.02 6 Model calibration After a one-year warm-up period (1993-1994), the model was cal- ibrated for in situ and all GPP data from 1994-2003. Due to the relatively short streamflow record, we did not apply a split-sample test for validation. The calibration process was carried in three steps: I Calibration using in situ precipitation data as a benchmark for model evaluation. II The benchmark calibration was refined using the posterior parameter ranges (5th and 95th percentiles) as parameter in- tervals in a second calibration. III Each GPP was used as input for calibration using the parame- ter intervals established in the second step. The 100 best pa- rameters were subsequently used for streamflow simulation for each independently calibrated product, to calculate per- formance statistics and to indicate the parameter uncertainty in form of the 5th/95th percentiles of simulated streamflow from the 100 best parameter sets (Nauditt et al., 2016). From the calibration of each global precipitation product and the benchmark, we analyzed the relationships between the cali- brated parameters attempting to detect changes and co-linearity between parameters as an indicator of limited sensitivity. Finally, the global product with the highest logarithmic Nash-Sutcliffe (LogReff) performance criterion was selected for the hydrological drought analysis. This calibration process was carried out using a Monte Carlo pa- rameter sampling strategy over 1 million iterations each (West- erberg & Birkel, 2015). The initial parameter values (Table 3) es- tablished by Birkel et al., (2012) in similar climate conditions were used. We additionally evaluated model performance using the Nash-Sutcliffe criterion (Reff), the modified Kling-Gupta efficiency (KGE) of Kling et al. (2012) and the Volumetric Error (VE). Table 3: Initial parameter ranges and constrained parameter range after the first calibration step of HBV-Light in the Tempisque sub-catchment according to Birkel et al. (2012). Parameter Initial lower Limit Initial upper Limit 5th Percentile 95th Percentile Soil Moisture Routine FC 0.1 1500 32.7 1390.3 LP 0.6 0.6 0.6 0.6 BETA 1 10 1.0 3.7 Response Routine PERC 0 10 2.9 9.8 Alpha 0 1 0.1 0.2 K1 0.05 0.9 0.1 0.2 K2 0.0001 0.1 0.002 0.01 PCALT 0 50 8.1 48.7 Routine Routing MAXBAS 1 4 1.1 2.0 2.2.3 Hydrological drought detection and simulation A hydrological drought event occurs when precipitation, soil moisture, water storage or discharge are below a defined thresh- old (Van Loon & Van Lanen, 2012). A hydrological drought can be characterized by its duration (length of a dry spell in days), severi- ty (deficit volume in mm) and intensity (deficit volume in mm per time unit) (Van Loon et al., 2014). • Duration (d) = days of drought below threshold. • Severity (s) = ∑ ( d(t) * Δt ); where d(t) is the deviation from the threshold and Δt represents the time difference. • Intensity (i) = s / d; where s is the severity in mm and d is the drought duration in days. In this study, a dynamic threshold level was used, based on the hydroclimatic seasonality (dry and rainy season) of the catchment, with a high probability of shifts during the seasons and the per- sistence of droughts caused by a deficit of precipitation in the dry and/or rainy season. The threshold level (Figure 3) was assumed as a dynamic 30-day moving window of the recorded 80th percentile from streamflow measurements, which is based on previous studies and the dy- namic nature of the study streamflow regime (e.g. Van Loon & Van Lanen, 2012). A variable threshold can be chosen when seasonal patterns need to be taken into account (Van Loon, 2015). Accord- ing to Hisdal et al., (2001) and Van Loon & Van Lanen, (2012) the Journal of Natural Resources and Development 2021; 02: 01 - 18DOI number: 10.5027/jnrd.v11i0.02 7 recommended method for capturing such hydroclimatic charac- teristics is the minimum monthly moving threshold. The analysis was carried out for the observed streamflow and the simulation results from 1994 to 2003 (1993 was the model warming up) using the statistical language R according to Van Loon, (2015). 2.2.4 Future hydrological drought scenario Future hydrological drought periods were derived following the same methodology of the dynamic threshold level as described before. However, streamflows were simulated driving the calibrat- ed benchmark HBV-Light model with the projected climate varia- bles from ROM as input data. The precipitation data from ROM and the simulated streamflow data were divided in two equal periods (24 years each) for comparison, the historic period from 1994 to 2019 and the projection from 2020 to 2045. Finally, the possible future hydrological droughts scenarios were characterized based on their intensity, severity and duration. Figure 3: Representation of the daily threshold level method with an arbitrary hydrograph. Figure 4: Observed streamflow time series (black dots) against the simulated best-fit streamflow generated with the benchmark in situ data (red line). The 5th and 95th percentiles from the best 100 results indicate parameter uncertainty (salmon color). Journal of Natural Resources and Development 2021; 02: 01 - 18DOI number: 10.5027/jnrd.v11i0.02 8 3 Results 3.1. Streamflow modelling in the seasonally-dry tropics with HBV-light The model was capable of simulating the streamflow time series maintaining the climatic seasonality (Figure 4). The model was trained to match low flows using LogReff as calibration criterion and reproduced the major drought period recorded in 1998. De- spite a more systematic underestimation of peak flows, the mod- el was able to match the largest recorded peak flow in 2000. The best-fit Reff resulted in 0.59 for the benchmark model and a LogR- eff of 0.68. The model also did not exhibit problems in reproducing the re-wetting period after the dry season. In general, the model seemed a well-balanced representation of the streamflow dynam- ics of the Tempisque River over the study period from 1994 to 2003. In comparison to the benchmark model, the calibration of each global precipitation product is presented in Table 4. The global products CHIRPS, TRMM and MSWEP resulted in best-fit KGE val- ues of around 0.5. The PGF product showed a lower performance. All products resulted in acceptable volumetric errors with a max- imum of 0.9. CHIRPS was identified as the most suitable global precipitation product for drought simulation due the highest LogReff value (0.65 – 0.5, table 4) compared with the results of the other products (TRMM, 0.65- 0.5; MSEWP, 0.3-0.003 and PGF, 0.3 -0.01). Furthermore, CHIRPS has the longer record with respect to the similar TRMM performance and was subsequently used for hydrological drought simulation. Table 4: Evaluation of the 100 best HBV-Light calibrated model parameter sets using the in-situ station data benchmark and the Global Precipitation Products used for calibration. The CHIRPS product with the best LogReff rating (0.65-0.5) is highlighted in bold. Parameters Calibrated Products Maximum Value Minimum Value Reff Benchmark 0.6 0.5 LogReff 0.7 0.6 KGE 0.8 0.6 Volumetric Error 0.9 0.9 Reff CHIRPS 0.5 0.4 LogReff 0.65 0.5 KGE 0.5 0.4 Volumetric Error 0.9 0.9 Reff TRMM 0.5 0.4 LogReff 0.6 0.5 KGE 0.5 0.4 Volumetric Error 0.9 0.8 Reff MSWEP 0.3 0.1 LogReff 0.3 0.003 KGE 0.5 0.3 Volumetric Error 0.9 0.7 Reff PGF 0.1 -0.2 LogReff 0.3 -0.01 KGE 0.01 -0.2 Volumetric Error 0.9 0.5 The streamflow simulation based on CHIRPS as input data were compared visually to the benchmark simulation in Figure 5. Both precipitation inputs showed relatively similar low flow simula- tions. However, CHIRPS did underestimate the high flows. Journal of Natural Resources and Development 2021; 02: 01 - 18DOI number: 10.5027/jnrd.v11i0.02 9 Figure 5: Visual comparison of the simulated best-fit streamflow time series from in situ benchmark precipitation input (red line) and from calibration with CHIRPS data (blue line). The black dots represent the observed streamflow. Table 5: Hydrological drought parameters for the upper Tempisque river: duration, severity and maximum intensity using the observed streamflow, and simulations with in situ station data and with CHIRPS (1994 to 2003). The latter overall average drought parameters are compared to the major drought events with an absolute error estimate. Hydrological drought parameters for the upper Tempisque river Average Max Max Max Average Max Duration Duration Intensity Avg. Intensity Severity Severity (days) (days) (mm/day) (mm/day) (mm) (mm) Observed Flow 91.4 307 0.9 2.2 27.9 108.1 Simulated Flow CHIRPS 78.5 358 0.7 1.6 20.6 147.2 Simulated Flow in situ Station 99.8 329 0.6 1.3 25.4 99.8 Major hydrological drought periods with absolute error estimates Period Average Duration (days) Average Severity (mm) Max Intensity Average (mm/day) Absolute Error Average Duration (days) Absolute Error Average Severity (mm) Absolute Error Max Intensity Average (mm/day) Observed Flow 1994 27.3 9.8 0.7 Simulated in situ 26.8 9.1 0.5 -0.5 -0.7 -0.2 Simulated CHIRPS 48 21 0.9 20.7 11.3 0.3 Observed Flow 1997 9.9 7.1 0.6 --- --- --- Simulated in situ 18.9 9.3 0.7 8.9 2.3 0.2 Simulated CHIRPS 7.4 3.4 0.3 -2.5 -3.6 -0.2 Observed Flow 1998 10.5 2.3 0.3 --- --- --- Simulated in situ 23.4 4.1 0.3 12.9 1.8 0.04 Simulated CHIRPS 9 1.9 0.3 -1.5 -0.4 -0.01 Observed Flow 2001 - 2002 4.8 1.2 0.3 --- --- --- Simulated in situ 23.1 4.7 0.3 18.2 3.4 -0.05 Simulated CHIRPS 19.8 4.2 0.5 15 2.9 0.2 Journal of Natural Resources and Development 2021; 02: 01 - 18DOI number: 10.5027/jnrd.v11i0.02 10 Figure 6: Logarithmic scale representation of the variable threshold level for drought detection (red shaded area below the threshold in blue) for (A) the observed streamflow time series (black line), (B) the CHIRPSv2.0 simulated flow (black line) and (C) the simulated flow (black line) by in situ data. Journal of Natural Resources and Development 2021; 02: 01 - 18DOI number: 10.5027/jnrd.v11i0.02 11 3.2 Hydrological drought analysis The hydrological drought duration, severity and intensity are pre- sented in Table 5. These parameters were derived from observed flow as a benchmark and from the HBV-Light best-fit streamflow simulations using CHIRPS and in situ precipitation observations. In Table 5, the observed flow showed an average drought dura- tion of 91.4 days for the 1994-2003 study period with a maximum duration of 307 days and a maximum intensity of 0.9 mm/day. We detected an overestimation of simulated drought severi- ty compared to the streamflow derived severity. The maximum severity with CHIRPS resulted in 147 mm compared to only 108 mm from observed streamflow. The simulated drought intensity resulted closer to the observed benchmark of 1.6 mm/day. Gen- erally, the simulated results showed acceptable estimates of hy- drological drought parameters in comparison to observed values. In Table 5, we show the absolute error of simulated hydrological drought parameters with respect to the observed benchmark for the major drought events of 1994, 1997, 1998 and 2001 to 2002. The model simulations using in situ and CHIRPS data correctly identified the major observed drought periods during 1994, 1997- 1998 and 2001-2002. The panel A of Figure 6 shows the hydrolog- ical drought detection from observed streamflow. In comparison, the hydrological drought analysis with CHIRPS resulted in slightly longer drought duration and larger maximum severity than ob- served flow (Figure 6A). The in-situ precipitation input simulated drought (Figure 6C) particularly well and captured the minimum flow of the major drought 1997-1998. 3.3 Future hydrological drought The projected precipitation data from the Regional Climate Model ROM resulted in an overall trend of decreasing precipitation for the period from 2020 to 2045 compared to 1994 to 2019 (Figure 7A). Mostly decreasing precipitation was projected for July corre- sponding to the mid-summer drought phenomenon. The climate projections were used to simulate a future streamflow scenario with the calibrated HBV-light model (Figure 7B). The scenario re- sulted in mainly lower streamflows at the beginning of the rainy season from April to June and for July, while the peak rainy sea- son from August until October exhibited similar mean monthly streamflows. The projected daily streamflow simulations from 2020 to 2045 for hydrological drought analysis are presented in Figure 8B in rela- tion to the baseline from 1994 to 2019 (Figure 8A). The hydrological drought analysis using the variable threshold level showed the main periods of streamflow deficit (Figure 8). Droughts described by the duration, severity and intensity for each period are summarized in Table 6. Generally, the hydrological droughts detected between 2020 to 2045 increased in terms of the number of events, duration and severity compared to the baseline period 1994-2019. The latter increase was mainly caused by the projected lower precipitations. The major future drought periods detected were 2020-2021, 2021-2022, 2025-2027, 2033-2034. The largest drought event was identified for 2039-2041 and 2042-2043 resulting in a severity exceeding 125 mm of deficit volume in both cases (Table 6) as well as the duration indicating higher persistence and multi-year droughts (duration > 700 days). Figure 7: A) Projected monthly precipitation data by ROM for 2020 to 2045 compared to the baseline period of 1994-2019. B) Simulated monthly streamflow regime using ROM as input to the calibrated HBV-Light model for 2020 to 2045 with respect to the observed baseline period. Journal of Natural Resources and Development 2021; 02: 01 - 18DOI number: 10.5027/jnrd.v11i0.02 12 Figure 8: A) The drought analysis using the simulated best-fit baseline is compared to B) the projected future streamflow with identical threshold levels for 2020 to 2045 on a logarithmic scale. Journal of Natural Resources and Development 2021; 02: 01 - 18DOI number: 10.5027/jnrd.v11i0.02 13 Table 6: Projection of hydrological drought (2020-2045) with respect to the best-fit simulated baseline from 1994-2019. Simulated hydrological drought events (1994 - 2019) Start Date End Date Duration (Days) Severity (mm) Max Intensity (mm/day) 22/8/1994 2/10/1994 42 11.0 0.4 20/8/1997 24/5/1998 278 48.5 0.5 25/10/1998 28/3/1999 155 13.9 0.2 9/10/2002 28/4/2003 202 21.2 0.2 21/3/2004 20/5/2004 61 3.4 0.2 28/7/2008 5/8/2009 374 77.2 0.5 3/8/2012 14/9/2012 43 10.1 0.4 11/5/2013 19/6/2013 40 4.8 0.2 28/6/2013 16/8/2013 50 9.2 0.4 10/9/2014 24/4/2015 227 16.9 0.2 16/6/2015 28/7/2015 43 7.0 0.3 27/5/2018 26/7/2018 61 5.8 0.2 15/5/2019 28/7/2019 75 8.0 0.2 Projected hydrological drought events (2020 - 2045) 6/7/2020 5/10/2020 92 40.2 0.7 15/10/2020 20/6/2021 249 32.9 0.3 24/6/2021 21/10/2021 120 44.3 0.6 30/11/2021 28/5/2022 180 17.6 0.3 25/4/2023 24/6/2023 61 8.1 0.3 16/6/2025 2/10/2025 109 39.7 0.7 5/12/2025 25/5/2026 172 16.3 0.2 20/7/2026 30/10/2026 103 45.3 0.6 2/11/2026 25/5/2027 205 38.2 0.4 16/1/2033 8/5/2033 113 2.3 0.1 1/11/2033 17/5/2034 198 6.8 0.2 12/10/2037 28/3/2038 168 17.8 0.2 16/10/2038 6/5/2039 203 38.7 0.4 9/5/2039 25/6/2041 779 215.0 0.8 6/5/2042 8/6/2043 399 127.9 0.7 Journal of Natural Resources and Development 2021; 02: 01 - 18DOI number: 10.5027/jnrd.v11i0.02 14 4. Discussion 4.1. The performance of Global Precipitation Products ap- plied to simulate streamflow with HBV-Light The HBV-Light model simulations driven by GPP in the Tempis- que subcatchment (Table 4) showed potential in the data scarcity context of the seasonally-dry tropics in Costa Rica. A similar con- cept has been applied in other tropical regions with limited data, where GPP were evaluated with and without bias corrections and used for streamflow simulations via HBV-Light (Koutsouris et al., 2017). Zambrano-Bigiarini et al. (2017) highlighted the great opportunity that the use of GPP offers in conceptual hydrologi- cal modelling. However, other authors reported that streamflow simulations strongly depend on the product type (Bitew & Ge- bremichael, 2011) and that the global product must be careful- ly evaluated (Zambrano-Bigiarini et al., 2017). Despite notorious data scarcity in Latin America (Birkel et al., 2019), we were in a position to use constrained historic daily rainfall and streamflow records (1993 to 2003) to benchmark the global product driven model against station data. The CHIRPSv2.0 product (Funk et al., 2015) showed together with the TRMM product the best skill in reproducing observed streamflow dynamics similar to Zambra- no-Bigiarini et al. (2017) who evaluated these products against the Chilean rainfall station network. We selected the CHIRPSv2.0 product for hydrological drought analysis over TRMM despite sim- ilar model performance due to the longer data record availability. Despite the challenge to achieve a good model performance in the tropics (Strauch et al., 2017; Zubieta et al., 2015) even with cor- rected GPP input data (Strauch et al., 2017), our modelling showed a reasonably good performance (KGE = 0.6). The model succeed- ed in simulating low flows particularly over the dry season (e.g. Figure 5), when hydrological models usually perform better for wetter periods (Pilgrim et al., 1988; Lidén & Harlin 2000; Nauditt et al., 2016). However, an initial quality check of the CHIRPSv2.0 data against the in-situ observations did apart from volumetric differ- ences not show any temporal bias. Hence, the relatively accept- able simulations using the global precipitation product allowing for a more in-depth hydrological drought analysis. 4.2. Simulating hydrological drought with Global Precipi- tation Products Clearly, the hydrological drought analysis will crucially depend on the model performance (van Loon & van Lanen, 2012) and we can- not expect the GPP to outperform the overall better simulations with in situ data. Nonetheless, the model simulations correctly identified the observed drought periods of 1994, 1997-1998 and 2001-2002 with the highest simulated severities, intensities and durations. These periods were related to strong El Niño events (2002-2003) and very strong El Niño (1997-1998) events emphasiz- ing the potential drought behavior during such warm sea surface temperature (SST) episodes (Wolter & Timlin, 2011; NOAA, 2017; Muñoz-Jiménez et al., 2018). The hydrological droughts simulat- ed by CHIRPSv2.0 (Table 5) were benchmarked against an aver- age duration of 91 days during the study period 1994-2003, with a maximum duration of 307 days and with maximum intensities of 0.96 mm/day using observed streamflow as a benchmark. The model’s absolute errors showed despite a variability from event to event a tendency of a slightly overestimated drought duration and under-estimated drought severity. In direct comparison, the CHIRPSv2.0-driven model showed less errors predicting drought duration compared to the deficit volume (severity) with errors up to 50 %. Similar to Van Loon and Van Lanen (2012), our analysis identified the rain deficit as one of the main causes of hydrologi- cal drought in regions with pronounced wet and dry seasons. Our results, additionally pointed in the same direction compared to findings from Hoyos et al., (2019). They found that rain and subse- quently groundwater recharge strongly controls the occurrence of droughts by simulating synthetic scenarios in the seasonal tropics of Colombia. 4.3. Predictive capability of the model to estimate future hydrological drought The RCM model ROM simulated precipitation regime revealed a slight decrease in precipitation from 2020-2045 (Figure 7A), com- pared to precipitation for the period 1994-2019. Similar to climate simulations from Neelin et al. (2006), where precipitation trends showed a drying trend for the Caribbean and Central American region. Moreover, the ROM model previously showed to match the intra-annual climatic behavior of rainfall in the Central Ameri- can Dry Corridor (Cabos et al., 2019) with a characteristic bimodal distribution with two rainfall maxima in June and September/Oc- tober (Taylor & Alfaro, 2005; Hidalgo et al., 2019). The precipitation decreases used as input to the calibrated HBV-Light model subse- quently resulted in less future streamflow as a possible future sce- nario. Water deficit for particular months are often evidenced in future scenarios (Avilés et al., 2020), despite undeniable uncertain- ty in such projections. Here, such a deficit was identified around the first rainfall peak in May/June (Figure 7B). Comparable results were presented for a case study in Cuba, where future projections suggested a regional decrease of about 38 % of the mean annual precipitation and up to 61 % in streamflow (Montecelos-Zamora et al., 2018). According to Hidalgo and Alfaro (2012), changes in the magnitude and the seasonal cycle of precipitations most like- ly affect Central American ecosystem dynamics and water availa- bility with most importantly, an increase in water scarcity in the region. Central America will experience a drought intensification (Imbach et al., 2018; Rauscher et al., 2008), as supported by our simulated future droughts. As shown in Table 6, increases in dura- tion and severity were evident, which was also found for a larger scale study by Hidalgo et al., (2013). Reduced precipitation and strengthened mid-summer drought may have consequences for farming and energy production (Imbach et al., 2018), and accord- ing to Quesada-Montano et al., (2018), projecting hydrological droughts is necessary for an improved water management in the socio-economically vulnerable region of Central America. Finally, this study might pave the way towards an early drought warning system, which is fundamental to proactive decision making and disaster preparedness (AghaKouchak et al., 2015). Journal of Natural Resources and Development 2021; 02: 01 - 18DOI number: 10.5027/jnrd.v11i0.02 15 5. Conclusions We used and evaluated GPP as input to the conceptual hydrolog- ical model HBV-Light for hydrological drought analysis in the sea- sonally-dry and data scarce tropics of Costa Rica. The use of GPP was encouraged due to limited information for modelling. We found that the CHIRPSv2.0 product showed the best skill to rep- resent the Tempisque streamflow compared to in situ benchmark precipitation data. The hydrological model driven by CHIRPSv2.0 successfully simulated observed streamflow variability and subse- quently allowed detection of hydrological droughts. A relatively successful simulation of historic streamflow also gave confidence in projecting potential future hydrological drought behavior. We specifically conclude that: • Global Precipitation Products can provide an alternative source of input data for hydrological modelling if carefully evaluated against in situ data. • The CHIRPSv2.0 product showed the best skill in simulating low flows in the Tempisque catchment at a relatively high spa- tial resolution with daily data. • The conceptual model HBV-Light was able to reproduce ob- served streamflow with the minimum parameterization (8 cal- ibrated parameters) and resulted in a maximum KGE of 0.8. • The model driven by CHIRPSv2.0 data correctly detected the major drought events. • The model driven by CHIRPSv2.0 data also correctly simulated the drought duration, but care has to be taken in using the simulated drought severity. • The simulated drought severity suffered from errors of 50% overestimating the volume deficits. • According to the results obtained future hydrological droughts will increase in duration and severity. Our novel drought research in the tropics provides some alterna- tives in the implementation of hydrological models based on GPPs as a tool for drought management in areas where the amount and timing of precipitation is closely related to hazards and economic sustainability. Acknowledgements The authors would like to gratefully acknowledge the support by the DAAD-funded TropiSeca project, UCREA-B8276, project C0603 and the Water and Global Change Observatory (OACG, ED- 3319) at the University of Costa Rica. Journal of Natural Resources and Development 2021; 02: 01 - 18DOI number: 10.5027/jnrd.v11i0.02 16 References Adler, R., Kidd, C., Petty, G., Morissey, M & Goodman, H. (2001). Inter- comparison of Global Precipitation Products: The Third Precipita- tion Intercomparison Project (PIP-3). Bull. Amer. Meteor. Soc., 82, 1377–1396. https://doi.org/10.1175/1520-0477(2001)082<1377:I- OGPPT>2.3.CO;2 AghaKouchak, A., Farahmand, A., Melton, F., Teixeira, J., Anderson, M., Wardlow, B., & Hain, C. (2015). Remote sensing of drought: pro- gress, challenges and opportunities. Rev. Geophys. 53 (2), 452–480. Artan, G., Gadain, H., Smith, J., Asante, K., Bandaragoda, C & Verdin J. (2007). Adequacy of satellite derived rainfall data for stream flow modeling. Nat Hazards 43, 167–185. https://doi.org/10.1007/ s11069-007-9121-6 Avilés, A., Palacios, K., Pacheco, J., Jiménez, S., Zhiña, D., & Delgado, O. (2020). Sensitivity exploration of water balance in scenarios of fu- ture changes: a case study in an Andean regulated river basin, The- or Appl Climatol, doi:10.1007/s00704-020-03219-y. Barrantes, G. (2010). Evaluation of the Environmental Water Service in the Tempisque River Basin (Costa Rica) and its application to tariff ad- justment. St. Thomas University. LEBRET Magazine - no. 2. Division of Economic and Administrative Sciences. Bucaramanga, Colombia . Beck, H., Wood, E., Pan, M., & Fisher, C. (2019). MSWEP V2 Global 3-Hourly 0.1 Precipitation: Methodology and Quantitative Assessment. De- partment of Civil and Environmental Engineering, Princeton Uni- versity, Princeton, New Jersey. Behrangi, A., Khakbaz, B., Jaw, T. C., AghaKouchak, A., Hsu, K., & Sorooshi- an, S. (2011). Hydrologic evaluation of satellite precipitation prod- ucts over a mid-size basin, Journal of Hydrology, 397(3), 225–237, doi:10.1016/j.jhydrol.2010.11.043. Beven, K & Westerberg, I. (2011). On red herrings and real herrings: disin- formation and information in hydrological inference. Hydrological Processes, 25, 1676–1680. doi:10.1002/hyp.v25.10. Birkel, C. (2005). Temporal and Spatial Variability of Drought Indices in Costa Rica. (Master’s Thesis). Institut für Hydrologie, Universität Freiburg. 75. Birkel, C. (2006). Hydrological Drought in Costa Rica Have they become more severe and frequent in recent years? Rev. Reflections 85 (1-2): 107-116, ISSN: 1021-1209 / 2006. Birkel, C., Brenes, A., & Sánchez, R. (2017). The Tempisque-Bebedero catchment system: energy-water-food consensus in the seasonally dry tropics of northwestern Costa Rica. Nexus Research Focus, TH- Köln, University of Applied Sciences. Birkel, C., Soulsby, C., & Tetzlaff, D. (2012). Modelling the impacts of land-cover change on streamflow dynamics of a tropical rainforest headwater catchment. Hydrological Sciences Journal, 57 (8), 1–19. Birkel, C., Moore, GW & Zambrano-Bigiarini, M. (2019). Hydrological Pro- cesses Special Issue “Hydrological processes across climatic and geomorphological gradients of Latin America”. Hydrological Pro- cesses. https://doi.org/10.1002/hyp.13648PREFACE3 Bitew, M & Gebremichael, M. (2011). Assessment of satellite rainfall prod- ucts for streamflow simulation in medium watersheds of the Ethi- opian highlands, [online] Available from: https://core.ac.uk/read- er/26834150 (Accessed 26 May 2020). Cabos, W., Sein, D., Durán-Quesada, A., Liguori, G., Koldunov, N., Martínez López, B., Álvarez, F., Sieck, K., Limareva, N., & Pinto, J. (2018). Dy- namical downscaling of historical climate over CORDEX Central America domain with a regionally coupled atmosphere–ocean model. Climate Dynamics (2019) 52:4305–4328. Calvo-Solano, O., Quesada-Hernández, L., Hidalgo, H., & Gotlieb, Y. (2018). Impacts of drought in the primary sector of the Central Ameri- can Dry Corridor. Agron. Mesoam. 29(3):695-709. ISSN 2215-3608, doi:10.15517/ma.v29i3.30828. Dai, A. (2011) Drought under global warming: a review. Wiley Interdisci- plinary Reviews: Climate Change, 2(1):45–65, doi:10.1002/wcc.81. Durán-Quesada, A., Gimeno, L., Amador, J., & Nieto, R. (2010). Moisture sources for Central America: Identification of moisture sources us- ing a Lagrangian analysis technique. J. Geophys. Res., 115, D05103. Funk, C., Peterson, P., Landsfeld, M., Pedreros, D., Verdin, J., Shukla, S., Husak, G., Rowland, J., Harrison, L., Hoell, A., & Michaelsen, J. (2015). The climate hazards infrared precipitation with stations-a new en- vironmental record for monitoring extremes, Sci. Data, 2, 150066, doi:10.1038/sdata.2015.66.76. Gendzh, N. (2015). Modelling Climate Sensitive River Catchments: The ap- plication of the HBV-Light modeltosevenIrish Catchments. Faculty of Science. Department of Geography. Maynooth University. Gotlieb, Y., Pérez-Briceño, P., Hidalgo, H., & Alfaro, E. (2019). The Central American Dry Corridor: a concensus statement and its backgroud. Journal Yu’am 3(5): 42-51. Guzmán, I. (2013). Base for the Planning of the Surface Water Resource in the Upper and Middle Basin of the Tempisque River, Costa Rica. (PhD thesis). National University, Heredia, Costa Rica. Journal of Natural Resources and Development 2021; 02: 01 - 18DOI number: 10.5027/jnrd.v11i0.02 17 Kanniah, K., Beringer, J., Hutley, L, Tapper, N., & Zhu, X. (2009). Evalua- tion of Collections 4 and 5 of the MODIS Gross Primary Produc- tivity product and algorithm improvement at a tropical savanna site in northern Australia. Remote Sens. Environ. 113, 1808–1822. doi:10.1016/j.rse.2009.04.013. Kling, H., Fuchs, M & Paulin, M. (2012). Runoff conditions in the upper Dan- ube basin under an ensemble of climate change scenarios. Journal of Hydrology. 424–425, pp. 264-277. 10.1016/j.jhydrol.2012.01.011 Hidalgo, H & Alfaro, E. (2012). Some Physical and Socio-economical As- pects of Climate Change in Central America. Progress in Physical Geography. doi: 10.1177/0309133312438906. 36(3), 380 – 399. Hidalgo H., Amador, J., Alfaro, E., & Quesada, B. (2013). Hydrological climate change projections for Central America. Journal of Hydrol- ogy. 495: 94-112. Hidalgo, H., Alfaro, E., Amador, J., & Bastidas, A. (2019). Precursors of qua- si-decadal dry-spells in the Central America Dry Corridor. Clim Dyn 53, 1307–1322. https://doi.org/10.1007/s00382-019-04638-y. Hisdal, H., Stahl, K., Tallaksen, L., & Demuth S. (2001). Have streamflow droughts in Europe become more severe or frequent? Internation- al Journal of Climatology 21(3):317 – 333. doi: 10.1002/joc.619 77. Hoyos, N., Correa-Metrio, A., Jepsen, S. M., Wemple, B., Valencia, S., Marsik, M., Doria, R., Escobar, J., Restrepo, J. C., & Velez, M. I. (2019). Mode- ling Streamflow Response to Persistent Drought in a Coastal Tropi- cal Mountainous Watershed, Sierra Nevada De Santa Marta, Colom- bia, Water, 11(1), 94, doi:10.3390/w11010094. Huffman G., Adler R., Bolvin D., & Nelkin E. (2010). The TRMM Multi-Satel- lite Precipitation Analysis (TMPA). In: Gebremichael M., Hossain F. (eds) Satellite Rainfall Applications for Surface Hydrology. Springer, Dordrecht. Imbach, P., Chou, S. C., Lyra, A., Rodrigues, D., Rodriguez, D., Latinovic, D., Siqueira, G., Silva, A., Garofolo, L., & Georgiou, S. (2018). Future cli- mate change scenarios in Central America at high spatial resolution, PLOS ONE, 13(4), e0193570, doi:10.1371/journal.pone.0193570. IPCC. (1997). In: Watson, R.T., Zinyowera, M.C., Moss, R.H. (Eds.), The Re- gional Impacts of Climate Change: An Assessment of Vulnerability. Cambridge University Press, UK, pp 517. ITCR (Technological Institute of Costa Rica). (2004). Digital Atlas of Costa Rica 2004. Koutsouris, A. J., Seibert, J., & Lyon, S. W. (2017). Utilization of Global Precipitation Datasets in Data Limited Regions: A Case Study of Kilombero Valley, Tanzania, Atmosphere, 8(12), 246, doi:10.3390/ atmos8120246. Lidén, R & Harlin, J. (2000). Analysis of conceptual rainfall–runoff model- ling performance in different climates. Journal of Hydrology, 238 (3–4), 231–247. doi:10.1016/S0022-1694(00)00330-9. Liechti T., Matos J., Boillat J & Schleiss A. (2012). Comparison and eval- uation of satellite derived precipitation products for hydrolog- ical modeling of the Zambezi River Basin. Hydrol Earth Syst Sci 16(2):489–500. doi:10.5194/hess-16-489-2012 Maggioni, V & Massari, C. (2018). On the performance of satellite precipi- tation products in riverine flood modeling: A review, J. Hydrol., 558, 214–224, https://doi.org/10.1016/j.jhydrol.2018.01.039 Mateo-Vega, J. (2001). General characteristics of the Tempisque River, p. 32-72. In: Jiménez, J.A. and E. González (editors). The Tempisque River Basin. Perspectives for integrated management. Organization for Studies Tropical, San Jose, Costa Rica. Marcos, Ó. (2001). Drought: definitions, typologies and methods of quan- tification University Institute of Geography University of Alicante. Geographical Research, No. 26 (2001) pp. 59-80. Monge, J & Gómez, P. (2007). Tempisque: a basin of high ecological di- versity in northwest Costa Rica. Biocenosis Magazine / Vol. 20 (1-2) 2007. Direction of Academic Production, State Distance University. Costa Rica. Montalván, I. (2017). Evaluation of the performance of the Hydrological HBV-Light model in a micro-watershed. Faculty of Chemical Scienc- es, School of Environmental Engineering University of Cuenca, Ecuador. Montecelos-Zamora, Y., Cavazos, T., Kretzschmar, T., Vivoni, E. R., Corzo, G., & Molina-Navarro, E. (2018). Hydrological Modeling of Climate Change Impacts in a Tropical River Basin: A Case Study of the Cauto River, Cuba, Water, 10(9), 1135, doi:10.3390/w10091135. Muñoz‐Jiménez, R., Giraldo‐Osorio, J.D., Brenes‐Torres, I., Avendaño‐ Flores, I., Nauditt, A., Hidalgo‐León, H., & Birkel, C. (2018). Spatial and temporal patterns, trends and teleconnection of cumulative rainfall deficits across Central America. International Journal of Cli- matology; Doi 10.1002/joc.5925. National Ocean and Atmospheric Administration (NOAA). (2017). ENSO cold and warm episodes by season. College Park, MD: Climate Pre- diction Center. Available at: http://www.cpc.ncep.noaa.gov/prod- ucts/analysis_monitoring/ensostuff/ensoyears_ERSSTv3b.shtm (accessed 23rd August 2017). Nauditt, A & Ribbe, L. (2017). Land Use and Climate Change Interactions in Central Vietnam: LUCCi, Springer. Journal of Natural Resources and Development 2021; 02: 01 - 18DOI number: 10.5027/jnrd.v11i0.02 18 Nauditt, A., Firoz, A., Trinh, V., Fink, M., Stolpe, H & Ribbe, L. (2017). Hy- drological Drought Risk Assessment in an Anthropogenically Im- pacted Tropical Catchment, Central Vietnam. In: Nauditt A., Ribbe L. (eds) Land Use and Climate Change Interactions in Central Viet- nam. Water Resources Development and Management. Springer, Singapore. https://doi.org/10.1007/978-981-10-2624-9_14 Nauditt, A., Birkel, C., Soulsby, C., & Ribbe, L. (2016). Conceptual modelling to assess the influence of hydroclimatic variability on runoff pro- cesses in data scarce semi-arid Andean catchments, Hydrological Sciences Journal. Ndehedehe, C., Agutu, N., Okwuashi, O., & Ferreira, V. (2016). Spatio-tem- poral variability of droughts and terrestrial water storage over Lake Chad Basin using independent component analysis. Journal of Hy- drology, 540, 106–128. doi:10.1016/j.jhydrol.2016.05.068. Neelin, J. D., Münnich, M., Su, H., Meyerson, J. E., & Holloway, C. E. (2006). Tropical drying trends in global warming models and observations, PNAS, 103(16), 6110–6115, doi:10.1073/pnas.0601798103. Paredes, J., Solera, A., Andreu, J., & Lerma, N. (2014). EvalHid tool for water resources assessment. Technical Manual v1.1. Water Resources En- gineering Group Universidad Politécnica de Valencia, Spain. Pilgrim, D., Chapman, T., & Doran, D. (1988). Problems of rainfall runoff modelling in arid and semiarid regions. Hydrological Sciences Jour- nal, 33 (4), 379–400. doi:10.1080/02626668809491261. Poméon, T., Diekkrüger, B., Springer, A., Kusche, J., & Eicker, A. (2018). Mul- ti-Objective Validation of SWAT for Sparsely-Gauged West African River Basins—A Remote Sensing Approach. Water 2018, 10, 451. Quesada-Montano, B.; Wetterhall, F.; Westerberg, I.; Hidalgo, H., & Halldin, S. (2019). Characterising droughts in Central America with uncer- tain hydro-meteorological data. Theoretical and Applied Climatol- ogy. Rauscher, S. A., Giorgi, F., Diffenbaugh, N. S., & Seth, A. (2008). Extension and Intensification of the Meso-American mid-summer drought in the twenty-first century, Climate Dynamics, 31, 551–571, doi:10.1007/s00382-007-0359-1. Seibert, J. (2005). HBV light; User’s Manual. Stockholm University. Depart- ment of Physical Geography and Quaternary Geology. Seibert, J & Vis, P. (2012). Teaching hydrological modeling with a us- er-friendly catchment-runoff-model software package. Hydrology Earth System Sciences. 16, 3315–3325. Sein, D., Mikolajewicz, U., Gröger, M., Fast, I., Cabos, W., Pinto, J., Hagemann, S., Semmler, T., Izquierdo, A., & Jacob, D. (2015). Regionally coupled atmosphere-ocean-sea ice-marine biogeochemistry model ROM: 1. Description and validation. Journal of Advances in Modeling Earth Systems. 7:268–304. https://doi.org/10.1002/2014MS000357 . Shen, Y., Xiong A., Wang, Y & Xie, P. (2010). Performance of high‐resolu- tion satellite precipitation products over China. J.Geophys. Res.115: D02114, doi: 10.1029/2009JD012097 Strauch, M., Kumar, R., Eisner, S., Mulligan, M., Reinhardt, J., Santini, W., Vet- ter, T., & Friesen, J. (2017). Adjustment of global precipitation data for enhanced hydrologic modeling of tropical Andean watersheds, Climatic Change, 141(3), 547–560, doi:10.1007/s10584-016-1706-1. Taylor, M & Alfaro, E. (2005) Climate of Central America and the Caribbean. In: Oliver JE (ed) Encyclopedia of world climatology. Springer, Neth- erlands, pp 183–189. Thiemig, V., Rojas, R., Zambrano-Bigiarini, M., Levizzani, V., & De Roo, A. (2012). Validation of Satellite-Based Precipitation Products over Sparsely Gauged African River Basins, J. Hydrometeor., 13(6), 1760– 1783, doi:10.1175/JHM-D-12-032.1. Van Loon, A. (2015). Hydrological drought explained. WIREs Water, 2: 359– 392. doi: 10.1002/wat2.108. Van Loon, A & Van Lanen, J. (2012). A process-based typology of hydro- logical drought. Hydrology Earth System Sciences. 16, 1915–1946. Van Loon, A., Tijdeman, E., Wanders, N., Van Lanen, H., Teuling, A & Uijlen- hoet, R. (2014). How climate seasonality modifies drought duration and deficit, Journal of Geophysical Research: Atmospheres., 119, 4640–4656, doi:10.1002/2013JD020383. Westerberg, I & Birkel, C. (2015). Observational uncertainties in hypothe- sis testing: investigating the hydrological functioning of a tropical catchment. Hydrological Processes. Hydrol. Process. 29, 4863–4879 (2015). Wolter, K & Timlin, M.S. (2011). El Niño/Southern Oscillation behaviour since 1871 as diagnosed in an extended multivariate ENSO index (MEI.ext). International Journal of Climatology, 31(14), 1074–1087. Zambrano-Bigiarini, M., Nauditt, A., Birkel, C., Verbist, K., & Ribbe, L. (2017). Temporal and spatial evaluation of satellite-based rainfall estimates across the complex topographical and climatic gradients of Chile, Hydrol. Earth Syst. Sci., 21, 1295-1320, doi: 10.5194/hess-21-1295. Zubieta, R., Getirana, A., Espinoza, J. C., & Lavado, W. (2015). Impacts of satellite-based precipitation datasets on rainfall–runoff modeling of the Western Amazon basin of Peru and Ecuador, Journal of Hy- drology, 528, 599–612, doi:10.1016/j.jhydrol.2015.06.064. View publication statsView publication stats https://www.researchgate.net/publication/349869618