Remote sensing-aided rainfall–runoff modeling in the tropics of Costa Rica

Abstract

Abstract. Streamflow simulation across the tropics is limited by the lack of data to calibrate and validate large-scale hydrological models. Here, we applied the process-based, conceptual HYPE (Hydrological Predictions for the Environment) model to quantitatively assess Costa Rica's water resources at a national scale. Data scarcity was compensated for by using adjusted global topography and remotely sensed climate products to force, calibrate, and independently evaluate the model. We used a global temperature product and bias-corrected precipitation from Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) as model forcings. Daily streamflow from 13 gauges for the period 1990–2003 and monthly Moderate Resolution Imaging Spectroradiometer (MODIS) potential evapotranspiration (PET) and actual evapotranspiration (AET) for the period 2000–2014 were used to calibrate and evaluate the model applying four different model configurations (M1, M2, M3, M4). The calibration consisted of step-wise parameter constraints preserving the best parameter sets from previous simulations in an attempt to balance the variable data availability and time periods. The model configurations were independently evaluated using hydrological signatures such as the baseflow index, runoff coefficient, and aridity index, among others. Results suggested that a two-step calibration using monthly and daily streamflow (M2) was a better option than calibrating only with daily streamflow (M1), with similar mean Kling–Gupta efficiency (KGE ∼ 0.53) for daily streamflow time series, but with improvements to reproduce the flow duration curves, with a median root mean squared error (RMSE) of 0.42 for M2 and a median RMSE of 1.15 for M1. Additionally, including AET (M3 and M4) in the calibration statistically improved the simulated water balance and better matched hydrological signatures, with a mean KGE of 0.49 for KGE in M3–M4, in comparison to M1–M2 with mean KGE < 0.3. Furthermore, Kruskal–Wallis and Mann–Whitney statistical tests support a similar model performance for M3 and M4, suggesting that monthly PET-AET and daily streamflow (M3) represents an appropriate calibration sequence for regional modeling. Such a large-scale hydrological model has the potential to be used operationally across the humid tropics informing decision-making at relatively high spatial and temporal resolution.

Tracer‐aided ecohydrological modelling across climate, land cover, and topographical gradients in the tropics

“…This configuration was still able to reproduce the topography‐dominated (STARRtropics uses the TWI to generate and route flow similar to the original TOPMODEL approach of Beven & Kirkby, 1979) discharge response to rainfall events (Figures 7 and 8) and also the associated tracer mixing and transport at a greater computational efficiency, but at the expense of lower performance at more dynamic headwater sites such as, for example, the San Lorencito experimental catchment simulated by Dehaspe et al (2018) and Correa et al (2020). Discharge simulations at the outlet of the San Carlos catchment were similar to other conceptual modellings, such as the regionally calibrated HYPE model for Costa Rica (Arciniega‐Esparza et al, 2022), with a KGE and CC of 0.61 and 0.91.…”

“…Therefore, the largest source of uncertainty and challenge in modelling was the distributed input data to drive the model (e.g., underestimation of peak rainfall events by global products as demonstrated by Arciniega‐Esparza et al, 2022). Nonetheless, few such attempts to use climate model output as drivers for tracer‐aided hydrological modelling have been made in the past (e.g., Delavau et al, 2017 for a large catchment in Canada), but in the future, this might be the only feasible approach to generate large‐scale ecohydrological simulations of water partitioning, particularly in scarce data regions.…”

“…Daily precipitation from the bias‐corrected global product CHIRPSv2 (Funk et al, 2015) generated by Arciniega‐Esparza et al (2022) for Costa Rica was used as forcing, which was checked against the Soltis Center station data (Figure 1a) with a good agreement ( R 2 > 0.8). This already corrected precipitation product matches the overall water balance of the San Carlos and closely agrees with the rainfall regimes of the modelled IsoRSM precipitation output but was preferred due to the higher spatial resolution of CHIRPS (~5 × 5 km vs. ~10 × 10 km) to capture the topographical precipitation gradient and the observed daily rainfall‐runoff response.…”

Dry season plant water sourcing in contrasting tropical ecosystems of Costa Rica

“…Water source contribution indicated a moderate uptake from the past hydrological year (38.6%) biassed toward the second rainfall maxima between September and October (19.5%). However, since the dry season in northwestern Costa Rica spans from mid‐November to May (~5–6 months) with a notable decrease in soil moisture, plants also revealed a moderate dependency on surface water (33.1%), which can be interpreted as groundwater contribution (deeper routing and competition) during a common and prolonged baseflow regime under high evaporative conditions (Arciniega‐Esparza et al, 2022; Hund et al, 2021). The most prominent rainfall deficits in this region are experienced during the warm phase of ENSO (Babcock et al, 2016).…”

Rainfall in the Greater and Lesser Antilles: Performance of five gridded datasets on a daily timescale

“…The RGD were evaluated along four characteristics: The KGE score is widely used in hydrology in order to compare observations with model simulations (e.g. Arciniega-Esparza et al, 2022;Mathevet et al, 2020), and is also used the context of RGD comparison with rain gauges series (e.g. Baez-Villanueva et al, 2018;Centella-Artola et al, 2020;Saemian et al, 2021).…”