Introduction
⌅Worldwide, agricultural and livestock zones have been developed near floodplains to facilitate access to water resources (Hartnett & Nash, 2017HARTNETT, M.; NASH, S.: (2017). “High-resolution flood modeling of urban areas using MSN_Flood-NC-ND”DOI: https://doi.org/10.1016/j.wse.2017.10.003 ). However, these developments also contribute to natural disasters by significantly altering the river’s morphological pattern, which modifies the hydrodynamic conditions of the riverbed. As a consequence, severe flooding problems occur, affecting agricultural areas, causing loss of human lives, and generating economic and environmental damage (Zhu et al., 2019ZHU, X.; DAI, Q.; HAN, D.; ZHUO, L.; ZHU, S.; ZHANG, S.: “Modeling the high-resolution dynamic exposure to flooding in a city region”. Hydrology and Earth System Sciences, 23(8), 3353-3372, 2019, DOI: https://doi.org/10.5194/HESS-23-3353-2019 ).
In particular, the Chillón River basin in Peru faces various agricultural and livestock challenges due to the lack of preventive planning in response to climate change phenomena, such as El Niño, which exacerbates the frequency of floods. The Carabayllo area is largely characterized by the agricultural sector, with an irrigated area exceeding 3,000 hectares. This area accounts for 45% of Lima Metropolitana's horticultural production, with 52 species and varieties of vegetables. Meanwhile, the livestock sector focuses on cattle, goat, pig, and poultry farming (Galagarza et al., 2021GALAGARZA, O. A.; RAMIREZ-HERNANDEZ, A.; OLIVER, H. F.; ÁLVAREZ RODRÍGUEZ, M. V.; VALDEZ ORTIZ, M. D. C.; PACHARI VERA, E.; CERECEDA, Y., DIAZ-VALENCIA, Y. K.; DEERING, A. J.: “Occurrence of chemical contaminants in peruvian produce: a food-safety perspective”. Foods, 10(7), 1461, 2021. DOI: https://doi.org/10.3390/FOODS10071461/S1 ).
Disaster risk management and mitigation of natural disasters are crucial factors for effectively addressing floods and their adverse effects. Spatiotemporal modeling of natural disasters mainly employs disaster system simulation methods, including hydrological and hydraulic models capable of simulating river flooding. In particular, HEC-RAS is recognized for its working range and accuracy in addressing uncertainty and flood impact (Kim & Cho, 2019KIM, J., & CHO, H.: “Scenario-based urban flood forecast with flood inundation map”. Tropical Cyclone Research and Review, 8(1), 27-34, 2019, DOI: https://doi.org/10.1016/J.TCRR.2019.07.003 ; Mokhtar et al., 2018MOKHTAR, E. S.; PRADHAN, B.; GHAZALI, A. H.; SHAFRI, H. Z. M.: “Assessing flood inundation mapping through estimated discharge using GIS and HEC-RAS model”. Arabian Journal of Geosciences, 11(21), 1-20, 2018, DOI: https://doi.org/10.1007/S12517-018-4040-2/METRICS ; Pinos et al., 2019PINOS, J.; TIMBE, L.; TIMBE, E.: “Evaluation of 1D hydraulic models for the simulation of mountain fluvial floods: a case study of the Santa Bárbara River in Ecuador” 2019, DOI: https://doi.org/10.2166/wpt.2019.018 ).
Additionally, various methodologies exist worldwide for preventing and reducing the risk of natural disasters. However, recent studies propose new multi-criteria evaluation methods based on the tools used in these assessments. Volonté et al. (2021VOLONTÉ, A.; GONZÁLEZ, M. A.; GIL, V.: “Gestión del riesgo y territorio fluvial. El caso del arroyo San Bernardo”. Boletín Geográfico, 43(1), 2021, https://revele.uncoma.edu.ar/index.php/geografia/article/view/3246 ) consider historical, ecological, and geomorphological factors. López-Martínez (2019)LÓPEZ-MARTÍNEZ, F.: “Análisis y diagnóstico de la vulnerabilidad general al riesgo derivado de los procesos de inundación fluvial en el litoral mediterráneo peninsular” 2019. https://dialnet.unirioja.es/servlet/tesis?codigo=289643&info=resumen&idioma=ENG quantifies the influence of different causal agents that shape and condition vulnerability levels in their analysis and diagnosis. Meanwhile, Vega-Ochoa (2020)VEGA-OCHOA, D. A.: “Análisis bajo un enfoque sustentable del riesgo por inundaciones en sistemas fluviales: “Río Tunjuelo (Bogotá, Colombia)” (2020), https://repository.udistrital.edu.co/items/a44311fe-1df2-4f3b-bc72-231ade4e2600 proposes a local land-use planning methodology with a sustainable approach to river systems. However, an essential aspect that prevention methodologies must integrate is the estimation of direct economic losses, defined as the temporary disruption of economic flows in the affected area. In this regard, the scientific literature cites analytical models to determine the true economic costs of floods, which must be adapted to each country’s circumstances.
The Chillón River is both a source of water and life and a latent threat to the agricultural sector that develops along its banks when it carries peak flows. Therefore, the purpose of this study is to define the spatial limits of agricultural risk and its direct economic impact due to flooding at different probability levels in the Chillón River in the district of Carabayllo, Peru.
Materials and methods
⌅The study focused on the Chillón River as the population of interest, with a sample of 21 km of the river located in the Carabayllo district, which has a population density of 315 inhabitants/km² (INEI, 2017INEI: “Resultados Definitivos de los Censos Nacionales 2017- Censos Nacionales 2017”, 2017. https://censo2017.inei.gob.pe/resultados-definitivos-de-los-censos-nacionales-2017/ ). The topographic data was obtained from a study conducted by CENEPRED using a DJI MAVIC 2 PRO drone with a 0.1m x 0.1m grid. Source: https://sigrid.cenepred.gob.pe/sigridv3/drones
Regarding the geographic database for the urban area, the study used current land-use data from the Global Land Cover by National Mapping Organizations (GLCNMO). Source: (Kobayashi et al., 2017). Agricultural areas were estimated using the satellite tool provided by the Ministry of Agricultural Development and Irrigation (MIDAGRI). Source: https://minagri-geoespacial.users.earthengine.app/view/pastosv2. Meanwhile, the economic information for the study area was extracted from the World Bank. Source: https://datahelpdesk.worldbank.org/knowledgebase/articles/906519
For hydrological factors, a 55-year historical discharge series from 1964 to 2018 was used, compiled by the National Meteorology and Hydrology Service of Peru (SENAMHI) and the National Water Authority (ANA). The HEC-RAS numerical model version 6.2 was applied for one-dimensional (1D) hydraulic simulation. Seven flood scenario models were proposed with hydrometeorological probabilities of 0.1% (1,000 years), 0.2% (500 years), 0.5% (200 years), 1% (100 years), 2% (50 years), 5% (20 years), and 10% (10 years).
The river section geometry was delineated using the 0.1m x 0.1m grid with the RAS Mapper interface linked to ArcGIS Pro 2.0, considering the most representative cross-sections to minimize model bias.
The boundary conditions of the river section were set as normal, with uniform and constant flow. Meanwhile, Manning’s n coefficient was calibrated for low and high flows based on the model’s statistically rigorous performance.
The spatial limits were determined based on flood risk along the margins of the Chillón River. A specific methodology was developed, as shown in Figure 1, to assess flood risk. The risk analysis flowchart was structured around three dependent variables: hazard, vulnerability, and direct economic impact. The latter introduced a new approach to measuring direct economic effects induced by floods, integrating aspects of the CENEPRED and CEPAL methodologies.
The estimation of direct economic impact was analyzed primarily based on water depth in the flooded area, the fraction of the flooded area, and the probability of the event-factors related to water exceeding the river’s capacity and flood protection standards. Finally, direct economic losses were estimated based on flood extent according to probabilistic scenarios. These direct economic losses were defined in terms of disruptions to the agricultural and livestock sectors (Tanoue et al., 2020TANOUE, M.; TAGUCHI, R.; NAKATA, S.; WATANABE, S.; FUJIMORI, S.; HIRABAYASHI, Y.: “Estimation of Direct and Indirect Economic Losses Caused by a Flood With Long-Lasting Inundation: Application to the 2011 Thailand Flood”. Water Resources Research, 56(5), 2020, DOI: https://doi.org/10.1029/2019WR026092 ) as shown in the following equation:
Where: Loos represents economic losses (USD), Dflood is the number of flood days, P is the daily production value (USD/day), and Drecov is the recovery period after the river flood (days), defined as Drecov = α × Dflood, where α is the recovery parameter ranging from 2.0 to 11.5 (Tanoue et al., 2020TANOUE, M.; TAGUCHI, R.; NAKATA, S.; WATANABE, S.; FUJIMORI, S.; HIRABAYASHI, Y.: “Estimation of Direct and Indirect Economic Losses Caused by a Flood With Long-Lasting Inundation: Application to the 2011 Thailand Flood”. Water Resources Research, 56(5), 2020, DOI: https://doi.org/10.1029/2019WR026092 ). The α parameter depends on disaster experience, mitigation measures, and policies. P represents the relationship between the GDP of the agricultural and livestock sectors and the GDP of the services sector, estimated at 0.864 based on the Central Reserve Bank of Peru's analysis at the time of the study.
Results and discussion
⌅As a result of the hydrological analysis, peak flows ranging from 40.84 to 102.17 m³/s were obtained for the best-fit Log Pearson III function. Meanwhile, the hydraulic model validation, comparing observed and simulated water surface levels for Manning’s n, showed high accuracy with an RMSE of 0.018 to 0.023 and an NSE of 0.92 to 0.962 for calibrated Manning’s n values ranging from 0.035 to 0.040.
The spatial risk limits are shown in Figure 2. However, the most significant return periods were 20 and 100 years, which allowed for precise estimates of flooded areas of 0.24 km² and 0.37 km², respectively. For other probabilities of peak flow occurrence, overlapping of spatial risk limits was observed.
The possible causes of this spatial risk limit overlap are due to the river channel width, which reaches up to 100 meters in certain sections, as well as the natural riverbank height, which ranges from 1 to 3 meters. These factors facilitate the evacuation of peak flows but increase flood frequency for return periods of 20 years. Figure 3 illustrates some characteristics of the studied river section, where fluvial terraces and a natural slope composed of sandy gravel with rounded particles, prone to erosion, worsen the river’s hydraulic conditions, affecting agricultural zones and nearby buildings along the riverbanks.
The risk and vulnerability analysis of the study area is presented in Table 1. The conditioning factor value was higher than the triggering factor value, primarily due to current land use. However, both factors reflected the same weight concerning the susceptibility of the geographical area. Consequently, the hazard value obtained was 0.173, interpreted as unfavorable geographic conditions for agricultural activities in these zones. Therefore, the area is classified under the category of frequent occurrence susceptibility, according to (Román & Orozco, 2019ROMÁN, A. Q.; OROZCO, J. J. Z.: “Zonificación de procesos de ladera e inundaciones a partir de un análisis morfométrico en la cuenca alta del río General, Costa Rica”. Investigaciones Geográficas, 99, 2448-7279, 2019, DOI: https://doi.org/10.14350/RIG.59843 ).
Similarly, vulnerability was assessed by considering social and economic impacts, as shown in Table 1. The obtained vulnerability value was 0.23, interpreted as high exposure for a fragile population engaged in the agricultural sector. In line with López-Martínez (2019)LÓPEZ-MARTÍNEZ, F.: “Análisis y diagnóstico de la vulnerabilidad general al riesgo derivado de los procesos de inundación fluvial en el litoral mediterráneo peninsular” 2019. https://dialnet.unirioja.es/servlet/tesis?codigo=289643&info=resumen&idioma=ENG , who highlights the integration of natural and societal relationships in risk management based on specific needs, greater weight is given to vulnerability due to exposure, fragility, and resilience in the evaluation. Overall, the resulting risk value was 0.13, representing a very high level. However, Reyna-García et al. (2020)REYNA-GARCÍA, A. E.; MOREIRA-MOREIRA, D. E.; BONILLA-PONCE, A. N.; PISCO PALACIOS, J. A.; MACÍAS MERA, C. J.; REYNA GARCÍA, A. E.; MOREIRA MOREIRA, D. E.; BONILLA PONCE, A. N.; PISCO PALACIOS, J. A.; MACÍAS MERA, C. J.: “Asentamientos humanos en zonas susceptibles a riesgos por inundación y deslizamiento de la ciudad de Portoviejo”. Revista San Gregorio, 43, 109-123, 2020, DOI: https://doi.org/10.36097/RSAN.V1I43.1413 emphasize the importance of resource interaction in determining hazard and vulnerability. Additionally, Alcocer-Yamanaka et al. (2016)ALCOCER-YAMANAKA, V. H.; RODRÍGUEZ-VARELA, J. M.; BOURGUETT-ORTIZ, V. J.; LLAGUNO-GUILBERTO, O. J.; ALBORNOZ-GÓNGORA, P. M.; ALCOCER-YAMANAKA, V. H.; RODRÍGUEZ-VARELA, J. M.; BOURGUETT-ORTIZ, V. J.; LLAGUNO-GUILBERTO, O. J.; ALBORNOZ-GÓNGORA, P. M.: “Metodología para la generación de mapas de riesgo por inundación en zonas urbanas”. Tecnología y Ciencias Del Agua, 7(5), 33-55. 2016, http://www.scielo.org.mx/scielo.php?script=sci_arttext&pid=S2007-24222016000500033&lng=es&nrm=iso&tlng=es argue that such information is valuable for defining safer agricultural zones.
| Susceptibility of the geographic area | ||||
| Conditioning factor | Triggering factor | Value | ||
| Value | Weight | Valor | Weight | |
| 0,194 | 0,5 | 0,152 | 0,5 | 0,173 |
| Vulnerability | ||||
| Social dimension | Economic dimension | Value | ||
| Value | Weight | Value | Weight | |
| 0,262 | 0,713 | 0,149 | 0,287 | 0,230 |
Since the most significant return periods were 20 and 1,000 years, the direct economic impact analysis of river floods was conducted only for these scenarios. Figure 4 presents the results for the 20-year scenario, showing that the agricultural sector is the most affected, with 94% of flooded areas, leading to direct economic losses exceeding USD 5,000. The most affected crops include celery, parsley, broccoli, lettuce, and cabbage. On the other hand, the livestock sector is affected by only 6% of the flooded areas, resulting in minimal losses of just USD 40. Specifically, livestock losses are concentrated in stables, poultry farms, and pig farming. Thus, the minimum estimated cost for agricultural and livestock recovery is USD 5,169.14.
In comparison, for the 1,000 year scenario shown in Figure 5, the probability of occurrence is low, and the flooded areas increase by only 1% in both sectors. However, economic losses rise by 26% in the agricultural sector and 46% in the livestock sector.
As a result of the floods caused by peak river flows, the study identified 514 affected residents and 185 impacted homes in the 20-year scenario. In contrast, for the 1,000 year scenario, these numbers rise to 1,105 families and 335 homes.
Conclusions
⌅This study enabled the delineation of two spatial risk boundaries for the occurrence of peak flows with return periods of 20 and 100 years, which result in flood areas of 0.24 km² and 0.37 km², respectively.
A flood risk of 0.13 was determined for the agricultural sector, indicating both total and partial deterioration in the yield of cultivated crops. Additionally, there were impacts on stables, poultry farms, and pig farming operations in the region, as well as failures in irrigation and drainage infrastructure and access roads.
The identification of the riverbed characteristics, composed of sandy gravel, highlights that during peak flow events, the unevenness of riverbanks is exacerbated, leading to fluvial erosion, sediment transport, and flooding in agricultural areas.
The study demonstrated that the agricultural sector experienced economic losses of USD 5,169.14 and USD 6,509.04 for flooded areas of 3.96 ha and 5.03 ha in the 20-year and 100-year scenarios, respectively.
This research establishes a new approach for delineating safe zones for agricultural producers to protect against natural risks. Furthermore, for future studies, it is recommended to classify risk based on cultivated agricultural areas and categorize the livestock sector accordingly.