Hydrodynamical Assessment of the Recent Droughts at Gallinas River in San Luis Potosí México and Its Impact on the Waterfall Tamul

: During the dry season of the years 2016–2020, the Gallinas River in San Luis Potosí State, Mexico, experienced a completeoss of its downstream ﬂow. These events impacted the Tamul waterfall, a tourist attraction with economicosses for the region. To investigate the causes, this research focuses on identifying the ﬂow variations in different river sections using the EFDC model under different scenarios to determine the causes of ﬂowosses resulting in the disappearance of the waterfall. To set up the conditions, measurements of ﬂow and speed, photogrammetry, bathymetry, and digital elevation modeling were necessary. The EFDC model was calibrated based on data acquired from measurement campaigns from 2017 to 2018. Five scenarios were established with different inﬂow boundary conditions: 1.5, 30, 60, and 1000 m 3 /s. According to the modeling results, it can be inferred that the mostikely reason for the ﬂow variations in the river is the clandestine water extraction and the inﬂuence of the karst geomorphology of the river that would generate speciﬁc inﬁltrations.


Introduction
Scarcity of water is a scientific reality that humankind must face in the coming decades, becoming a significant issue to sustainable development. Besides the potential economic impact of water scarcity in the society [1], it is estimated that the available freshwater resources are also directly affected by climate change [2], even in historically water abundant regions [3]. Urbanization and climate change are exacerbating water scarcity under different socioeconomic and climate change scenarios; the global urban population facing water scarcity is projected to increase from 933 million (one-third of the global urban population) in 2016 to 1.693-2.373 billion people (one-third to nearly half of the global urban population) in 2050 [4]. The American continent, particularly Mexico, has already faced the impact of water scarcity within its territory. Insufficient water is a decisive factor for towns and entire regions to be abandoned by the population in a formative stage, who seek opportunities elsewhere. This situation frequently exacerbates the social effects of water scarcity [5].
It is estimated that the future droughts that Mexico will face will be more frequent and prolonged due to climate change, constituting one of the hydro-meteorological phenomena with the most significant impact on the population and the economy derived from water scarcity, as food shortage, hygiene problems, among others [6].
During the last decade, several studies have proven the severity of droughts and the lack of access to freshwater for the population along the country [7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22]. A portion of Mexico that has received constant attention from the research point of view is the US-Mexico border. In order to illustrate, some towns as San Ignacio, Sonora is experiencing water resource pressures and increasingly visible impacts of climate change and climate variability, which have had significant gender impacts, with ripple effects on socioeconomic relations in the community [22]. In this context, the transboundary water governance in the frame of the NAFTA treaty are explored [17], while seeking to combat water scarcity along the border using renewable-powered desalinization processes. Moreover, other regions of the country face water scarcity problems [7].
In the northern part of the country, one of the most internationally recognized attractions is the waterfall of Tamul, located in Huasteca Potosina. The waterfall is formed when the Río Gallinas falls from a height of 105 m over the tributary of the Río Santa María. Environmentalists and tourism service providers have been reporting to the National Water Commission (CONAGUA https://www.gob.mx/conagua/en, accessed on 1 March 2020) of Mexico for years that the waterfall dries up during the dry season.
In April 2016, 2017, and again in 2019, during the low water season, most of the flow downstream in the Gallinas River was lost, causing the waterfall to dry up; there was no record before these dates. Until that time, no study explained the causes of the disappearance of flow in the Tamul waterfall.
The disappearance of the Tamul waterfall has directly affected the economy due to it having become an essential part of the economic income in the Huasteca region of San Luis Potosi. The inhabitants of the region attribute it to the sugar mill in the area which, during the last decade, went from processing 17,000 hectares of cane to 30,000 hectares each harvest season.
The watershed that supplies the Gallinas River has a high production of sugarcane, which has a water footprint of 203 cubic meters per ton. In addition, there has been uncommonly low rainfall in the region, clandestine intakes in the river, natural factors such as infiltration, among other possible causes of the temporary interruption of the flow. The lack of information about the reasons for the disappearance of the waterfall has resulted in the absence of actions to maintain the waterfall flow. In this work, we intend to identify the flow variations in different sections of the Gallinas River to determine alternative solutions to preserve the tourist attraction of the Tamul waterfall.
The main objective of this work is to model the hydrodynamics of the Gallinas River numerically to evaluate temporally and spatially flow losses. To carry out accurate modeling, it is necessary to collect the hydrological and hydraulic conditions of the Gallinas River and identify variations in river flow. The EFDC model was used to carry out the numerical simulations because it is a well-proven code to predict hydrodynamics, solving the momentum and free surface equations considering the continuity and mass balance equations [23].
The EFDC model simulates hydrodynamic processes in surface waters, solves the equations of motion and continuity, models the circulation and transport in complex environments, with which it is possible to know the behavior of the channel and identify the causes of the loss of flow in the river during the low water season from January to April. The purpose of the simulations is to identify the points of the significant upsets of flow during the flow in the Gallinas River until it reaches the Tamul waterfall and identify the leading causes for the drying up of the waterfall during the low water season mentioned above, and to rule out or corroborate the hypothesis that the variations in the flow of the Gallinas River are mainly due to natural hydrological processes of infiltration in the rock stratum. In the river section selected to be modeled there are many infiltrations in the soil, and during the dry season, the flow almost disappears. Additionally, the section has the presence of human settlements.
It is necessary to mention that drought events can be dealt with from different perspectives, for example, a statistical approach taking into account several parameters and their frequency (such as rain) and applying soft computing techniques to try to determine the behavior [24], even the use of neural networks can be used to forecast droughts [25]; In fact, neuro-fuzzy methods have proven to be precise in the prediction of drought with a criterium of 80% [26]. Although these methods have proven their efficacy, even in predicting sunspots [27] or determining temperature patterns in Mexico City for Global Warming Assessment [28], hydrodynamic modeling is more suitable because we want to know what is happening specifically with the river flow instead of the watershed or the region. This paper is organized as follows, Section 2 describes the study area where the River Gallinas and the Waterfall Tamul is located, considering the watershed, rain, bathymetry, topobathymetry, hydrograms and the numerical model used. Section 3 introduces the model setup and describes the measurement campaigns necessary to feed the model. Section 4 presents the simulation results under the established scenarios. Finally, the conclusions are given.

Description of the Study Area
The region under study in this work is located in the Huasteca region, in San Luis Potosí state, Mexico, with an altitude of 195 masl and belongs to the RH-26 Pánuco hydrographic basin. The zone under study is bounded between the coordinates 99 • 15 W, 21 • 59 N, and 99 • 14 W, 21 • 53 N (see Figure 1). Located within the municipality of Tamasopo, Gallinas river flows downstream for more than 100 km, up to the Tamul cascade, located in the Aquismón municipality, with a waterfall of 105 m. After the Tamul cascade, the river is known as Tampaón [29], which downstream contributes to the formation of the Pánuco River, which is one of the five main rivers of Mexico ending at the Gulf of Mexico.
The element that has given shape and life to the Huasteca is water, which has been the biological, economic, and social engine of the communities settled in the Sierra Madre Oriental and the great plains, being a reason for union and dispute among its inhabitants, as well as of ritual encounters related to agricultural cycles, fertility, and natural disasters. Moreover, touristic activities have accompanied the evolution of the local economics of the region [30]. Nowadays, the region that embodies the Gallinas river has several natural resources, with a multi-faceted economy [31]. Ninety percent of the local economic activities are tourism, farming, and cattle raising. Along the Gallinas River, which is under study, fishing and commerce are also important economic activities. These activities have changed the hydrographic features of the river, modifying its hydrology and its water quality. The delimitation of the basin is depicted in Figure 1, where the nozzle is located at the Tamul cascade. In this Figure, the main channel or primary collector is blue colored, while the wedGallinas sub-basin delimitation is red. The basin is large-intermediate, and it possesses a type B hypsometric curve, which means that it is a basin in equilibrium (maturity phase). The main features of the basin can be observed in Table 1.
The weather is warm during most of the year, ranging from cool to cold between November and February. Within the region, rainfall averages between 1200 and 2000 mm per year. Although such amount is considered, in quantity, very good for the development of the different agricultural activities, it turns out that since 80% of them occur in the months of June to September, there are periods of drought between January and April that cause significant risks for human activities, livestock, and agriculture [32]. For instance, [33] finds out, through the Water poverty index for 2010 with 1100 mm of annual precipitation, that the socio-economic activities in the Río Valles sub-basin depend on 55.33% of rainfall water stored in La Lajilla dam, which recharges only in such a wet season. The development of tourist activity is highly dependent on hydro-meteorological conditions, and it presents a high vulnerability to the impacts of climate change. Although the phenomenon of flow loss occurs during the dry season, it is necessary to include the analysis of precipitations in the study. A period of 55 years was considered, from 1 January 1961, to 31 December 2016. The precipitation data were acquired from the 19 weather stations within a radius of 20 km from the watershed, with data available in the period. Data were provided by National Water Commission (CONAGUA https: //www.gob.mx/conagua/en, accesed on 1 March 2020) of Mexico, a Mexican government agency. The location of weather stations is depicted in Figure 1. Regarding the temperature, the values present a slight rise as they approach the present, which, combined with low rainfall, can generate atypical conditions regarding water availability.
In Figure 2a, the average monthly precipitation values for the Gallinas sub-basin for the period under study are shown. The mean precipitation corresponding to the Gallinas sub-basin was calculated based on Thiessen polygons, yielding a value of 1537.65 mm/year. Furthermore, precipitations were determined for different return periods through the "Incomplete Gamma" method can be observed in Table 2. The Intensity-Duration-Return Period curves (I-D-Tr) computed for a representative weather station are presented in Figure 2b, which features return periods of 2, 5, 10, 25, 50, and 100 years.

Samples Collection Bathymetry Measurement
Due to the dimensions and configuration of the section of the river understudy, a broader zigzag route is made with points captured centimeters away from each other and without a second pass in the longitudinal direction. The sweep was performed with a length of more than 450 m on the zigzag, at a speed of 0.427 m/s. Bathymetry measurement campaigns were also helpful in detecting extraction points for agricultural use, which turned not have a formal structure such as a lift station.
As the current topographic information is only found in Digital Elevation Models with a maximum resolution of 15 m, which is insufficient for the elaboration of a hydrodynamic model, a topographic survey of the areas adjacent to the main course of Gallinas river is required.
For this study, the topography is required up to 0.5 m of precision. Based on the available times and the extension of the study area, the topographic survey was carried out by means of photogrammetry with UAV images. Flights were performed with a VANT Phantom 3 UAV, flying at 70 m to obtain a spatial resolution of ∼5 cm in orthomosaic and up to 50 cm in the Digital Elevation Model (DEM). The topographic survey is divided into eight different flights, which can be observed in Figure 3. As a result, a total of 2350 images were obtained. To adequately geolocate the aerial obtained images for this research, checkpoints measured in the field with differential GPS are used. The position correction of each image is carried out by Post-processing. Sixteen checkpoints were established, which are shown in Figure 1. For reference, the Level Bench point was taken in the Vertical Geodetic Station (BN) with the name 5112, being part of the Passive National Geodesic Network of the Mexican government agency INEGI (Instituto Nacional de Estadística, Geografía e Información, or the National Institute of Statistics, Geography, and Information).

Digital Elevation Model
For this research, a uniform roughness is applied throughout the model with a value of 0.02 (roughness coefficient (s/m (1/3) ), adjusted during calibration to modify the velocities in the river flow. For this case, we have two boundary conditions at the beginning of the modeling, the inflow represented by a hydrogram and distributed in the cells present at the birth of the model over the river area and an open outflow condition to the south. In this case, the inflow covers 35 cells.
The hydrogram was carried out based on the hourly monitoring of the flow rate and the patterns of consumption variation in drinking water recommended by CONAGUA, considering the withdrawals for domestic use as flow variation. For the open boundary in the southern part of the model, an output equal to the 62 cells that make up the end of the model was assigned.
Photogrammetry was performed through Agisoft PhotoScan Professional to obtain a 3D model with texture. Two Digital Elevation Models were generated, both of surface and terrain. Each DEM is given as a raster image in Tiff format at a resolution of 0.0709391 × 0.0709391 m for each pixel. After that, employing the ArcGIS software, the res-olution was reduced to a value of 0.5 × 0.5 m to speed up data processing during the generation of the meshing in modeling. Both DEMs can be observed in Figure 4.

Numerical Model
To perform the numerical modeling and assessment, many numerical models could be used in the field of hydrodynamic study of reservoirs for predictions of speed, temperature, and salinity in natural water bodies and patterns of circulation and dispersion. For this purpose, we can find the MIKE DHI, DELFT 3D, RMA-4, GEMSS (Generalized Environmental Modeling System for Surface waters), CWR-ELCOM, and UNTRIM, among others. The most recognized and with the most significant number of users have their own well-detailed interfaces, such as EFDC, FVCOM (VISIT), ROMS, MOHID; Even models programmed for specific applications, for flooding modeling [34,35], flows through vegetation [36], metal transport in rivers [37] and thermal plumes dispersion [38].
We used the EE Modeling System, a hydrodynamic and environmental modeling system, to conduct technical and environmental studies of fresh and saltwater bodies. The software enables engineers and scientists to create accurate scientific models of aquatic environments efficiently. Its main component is the EFDC (Environmental Fluid Dynamics Code). EFDC is a 3D model developed at the end of the 1980s for water quality studies in coastal areas and estuaries; it can be coupled to various quality models such as WASP7 and CEQUAL_ICM and also internally includes its quality model called HEM3D. This model also includes Lagrangian models for plumes dispersion (similar to JETLAG), which allows a near-field model to be coupled with a far-field model. EFDC is itself the 3D hydrodynamic model adopted and recommended by the U.S. Environmental Protection Agency (EPA) for water quality studies. Since its inception, the EFDC model has been systematically improved. Historically, the EPA was in charge of making the improvements, of giving workshops for their use and dissemination. After 2007, this responsibility was delegated to other environmental consulting companies (TetraTech, DS-Intl), which have been in charge of updating, disseminating, training, and generating free access tools for the pre and post-process. Nowadays, Dynamics Solutions is the company that has this task and has developed a graphical interface, which allows the pre and post-process, called EFDC Explorer. EFDC Explorer is a Microsoft Windows-based preprocessor, and postprocessor for the three-dimensional EFDC initially developed at the Virginia Institute of Marine Sciences [39,40].
The governing equations of the EFDC model include the Navier-Stokes for flow, the advection equations-diffusion for salinity, temperature, colorants, toxins, water quality components, and suspended sediment transport. They are discretized with the finite difference method in an explicit scheme. It should be noted that the hydrodynamic phenomena under study are characterized by horizontal length scales with a magnitude more significant than the vertical length scales. Thus, for this work, EFDC solves the shallow water differential equation) using a finite difference scheme and a Mellor-Yamada turbulence model of order 2.5 [41].

Model Setup
To perform in situ measurements of flow and speed, a RiverRay ADCP device over a Hydro board was used to transmit data in real-time to a laptop. The device, from Teledyne RD Instruments Water Resources, allows a range of flow profiles up to 60 m 3 /s as well as up to 100 m of depth for bottom tracking, with a 1% error rate under uniform temperature and salinity conditions. Measurements were geolocalised through an XGPS160 GPS device. The software to process data was WinRiver II.
The domestic use of water is carried out mainly by underground extraction, leaving a small percentage by direct extraction from the river, which is negligible in this modeling. The other use of water recognized in the study area is agriculture, which is used for sugarcane crops on the banks of the river. These crops use direct extraction from the river employing pumping for irrigation. During the dry season, it is necessary to carry out irrigation to cover the crop's water demand. The rainy season is from June to September, while the dry season is from October to May.
Since the extraction of water for agricultural irrigation in the site is not constant or predictable according to the existing free demand regime, it is not easy to quantify it, so for the modeling, the generation of the model with the data measured in campaigns is considered adequate.
Four measurement campaigns were performed based on different periods of the years; the control points' locations are shown in Figure 6.

First Measurement Campaign
Taking into account the differences between wet and dry seasons, the first measurement campaign was performed in March 2017 (dry season). The flow rate on the Gallinas river was obtained from six different points along the river. Point six is within a short distance of the Tamul waterfall.
For this campaign, each measurement was carried out between 3 to 5 transects per section to determine the flow values in each section of the river.

Second Measurement Campaign
The flow rate in the Gallinas River was obtained at nine different points along the river and the channels that reach the Gallinas River. The first three points are located on the tributaries that discharge into the Tamasopo stream, while the other points are in locations close to those that were measured in March 2017. The measurement sites were selected based on flow monitoring in the first measurement campaign of March 2017.
The second measurement campaign to obtain the flow rates in the Gallinas river was carried out in January 2018.
Multiple variations of flows were presented due to the presence of rains in the days before the campaign, which is reflected as an increase in speed in sections where the river narrows and making measurement difficult due to the instability that the hydro board may present, as is the case of the measure at point 9.

Third Measurement Campaign
The third measurement campaign was performed on 17 and 18 June and consisted of a total of eight measurement points.

Fourth Measurement Campaign
Measurement campaign four was performed on 9-11 August and it consisted of a total of 18 points, which were selected at constant distances along the river to locate the sections where flow losses occur in the river. This campaign focused on monitoring the flows measured during the third campaign, continuing with flow measurements up the Tamul waterfall.

Model Calibration
For this research, the flow and velocity time series analysis was required to calibrate the model. Seven control points were used to perform the records; their location are shown in Figure 7. In Figure 8, graphs of time (days) vs. velocity (m/s) recorded in the seven checkpoints for a specific period of time. Such checkpoints are used to calibrate flows passing the defined control points. Analogous, the flow rates are shown in Figure 9.       The flows measured in the third campaign in June 2018 were used for the model's calibration process. Simulations were carried out considering only the inflow conditions measured; nevertheless, the results do not match the sections' measurements (uncalibrated results).
Based on the results of the uncalibrated results, many simulations were performed with flow correction in specific points. Figure 7 shows the measurement points (with the prefix "P") and the introduction of flow variations (with the prefix "V") in the length of the modeled section of the Gallinas River.
Adding a flow loss or contribution near the points the flow was measured allows adjusting the results until they match the velocities and flow rates measured in the control points.
Negative flows can be due to various factors, such as flow deviations. In contrast, positive flows could represent the arrival of flow from tributaries to the main channel, the phreatic level's contribution, or springs generated by the same fractures of the rocky stratum.
A statistical analysis of the flows calculated by the model and the flows measured in the field is carried out. The square root of the mean square error, also known as RMSE, is used for this analysis. In Tables 3 and 4, the RMSE values for the June 2018 campaign before and after calibration, as well as the RMSE values for velocity and flow with the calibrated model, are shown. To perform the model's calibration, data from the measurement campaign of August 2018 were used. For this validation, the hydrogram shown in Figure 11 was introduced as an inlet, which corresponds to August 2018. The flow is more significant than that presented in June due to the rains. It must be remarked that only the inlet boundary condition is altered for validation purposes.   The RMSE value for checkpoints is around 0.034, which is the order of those values obtained in the calibrated model. Thus, the validation of the model is adequate, and it reliably represents the current hydrodynamic conditions in the Gallinas River.
The longitudinal profile of the river was also determined, showing the elevation of the land surface and the water surface in meters above sea level. Figure 12 shows the elevations of the water and ground surface collected in the field campaign of 18 June, obtained directly from the measurements.
The interpretation of the variations in elevation of the land surface suggests that the riverbed is conformed by ponds, as can be observed in Figure 13.

Simulations Results
To simulate different flow scenarios along the Gallinas River, flow rises of 30 m 3 /s, 60 m 3 /s, and 1000 m 3 /s were applied, representing 10 and 20 times a typical flow in the Gallinas River, which turns to be higher than the two years return period. Moreover, the inlet flow of the model was reduced by half, which is equivalent to 1.5 m 3 /s. After that, seven pumping conditions distributed along the river at different times of the day are introduced in the modeling. Table 5 shows the initial flow rates used in every scenario, and Figure 14 shows the initial free surface conditions for all the scenarios. Table 5. Initial water flow for each scenario.

Scenario
Inlet Condition

Scenario 1: Inflow of 30 m 3 /s
Scenario 1 simulates a flow boundary condition at 30 m 3 /s, which represents an inflow to the model of 10 times higher than the normal flow in the Gallinas River, for the section located in the Jabalí community.
This scenario represents the increased inflow due to heavy precipitation in the Gallinas sub-basin, expressed as flow in the model at 00:00 h on 9 August. Figure 15a-

Scenario 3: Inflow of 1000 m 3 /s
Scenario 3 presents a flow inlet condition of 1000 m 3 /s, which generates an overflow in the river near the El Jabalí community. Such flooding has been recorded on only one occasion. Therefore it can be considered an atypical event, which is covered within a return period of hundreds of years.
The flow is entered at 00:00 on 9 August and is only modeled for 45 min due to the flow stability reaching fast. The initial conditions at 00:00 h are the same as in scenario one, and the results can be observed in Figure 17a-c.
In Figure 17b, it can be seen that the flood was generated instantaneously in the first section of the segment, reaching the peak of the flow in the middle of the Section 2.
In Figure 17c, the level in Section 1 stabilizes, while Section 3 reaches the peak of the flow. Figure 17d, shows the peak flow reaches Section 4, where the model ends.
Although the flow is substantial, the river tolerates the effects of the increased flow.

Scenario 5: Pumping for Irrigation
In Scenario 5, a simulation is performed with the hydrogram of the flow measured on 17 June and 7 boundary conditions representing the previously identified field pumping for agricultural irrigation. Such a scenario represents the end of a risk ban after the end of the holiday season, even when the flow in the river is higher than the one that normally occurs in the month of April.
Assuming that the average monthly water requirement of sugarcane is 100 mm and irrigation is carried out in two fortnightly periods, it corresponds to 500 m 3 for each hectare of cultivation. Considering a total of 16 hectares as the taxable area for each pump, 8000 m 3 are required for irrigation monthly.
During the measurement campaigns, the seven located pumps had a 10 in pipe, yielding a capacity to extract up to 1,000,000 L per hour. Depending on the pump capacity and the assigned water requirement per pump, it would be necessary to keep the pump running for 8 h to meet the demand. The location of the pumps is shown in Figure 19, while the allocation of irrigation periods per pump is carried out randomly between 5:00 am and 9:00 pm, and can be observed in Table 6. The initial conditions of the model are the same as in the previous scenarios. The results of the simulation can be observed in Figure 20.

Conclusions
The Gallinas River sub-basin is a complex and sizeable hydrological system that requires detailed analysis. The lack of precise information regarding the types of soil, its uses, and its vegetation is a widespread problem in Mexico, where it is difficult to determine an adequate runoff coefficient for its study.
Another essential characteristic is the absence of reliable meteorological information, in addition to the frequency with which missing data is presented in the records. This problem can be partially resolved with statistical methods, but these solutions are still less accurate than the direct measurement of data in the field.
Field visits to determine hydraulic parameters in a river are of utmost importance in the modeling process. Moreover, for hydrodynamic modeling projects, the margin of error present in surveys with drones is negligible, unlike a study for cadastre, construction, etc.
Although the information collected from the bottom of the water by a drone is of good precision, it is always advisable to perform the bathymetry measurement to corroborate this information since the sun's reflection, turbidity, and the type of flow can alter the data measured by photogrammetry.
Temporal and spatial variations in the flow of a river are common. Still, their magnitude can vary due to natural hydrological, meteorological, edaphological, and geological conditions and anthropogenic conditions such as domestic, industrial, agricultural, and recreational activities. In the Gallinas river, various natural and anthropogenic conditions intervene in the flow variations.
All the modeling scenarios established can not be determined the complete flow loss; this suggests that the river exits clandestine water extraction and the influence of the karst geomorphology of the river that would generate specific infiltrations.
The flow in the Gallinas River presents significant variations along its channel, which can not be entirely attributed to the geological characteristics of the area, despite the geological environment that predominates of the karst type. Karst is distinguished by its high probability of generating infiltrations in cracks in the rock. This means that the variations caused by infiltration may change according to the season of the year due to changes in the phreatic level and the direction of the flow lines.
The hydrodynamic model carried out in the Gallinas river showed that the flow behavior presents variations reproduced in the calibration and validation conditions. Therefore, these variations are constant over time during the same season and under similar conditions. This means that critical flow variations have to be due to external factors like extraction.
The different scenarios in the hydrodynamic model show results of possible conditions that occur in nature. For example, the first three scenarios show the capacity of the Gallinas River to assimilate increased inflow, being 30 m 3 /s and 60 m 3 /s typical flows more minor than a return period of 5 years. Scenario 3 with a flow of 1000 m 3 /s represents an atypical scenario in return periods greater than 100 years. The river level reached extreme levels that were only recorded once and which may not reappear for hundreds of years.
Scenario 4 represents a decrease in flow to half the average flow, that is, to 1.5 m 3 /s. This scenario shows how the flow in the Gallinas River is not interrupted only by the reduction inflow.
Scenario 5 represents an extreme pumping condition, which is a drastic event that explains how the river flow becomes temporarily interrupted even when the water surface level shows a large volume of stagnant water in the river and not a disappearance of the river.
One of the main points observed is that the river section under study is made up of small reservoirs; the simulation shows to guarantee that there is continuity in the river flow, the reservoirs must first be filled. The pumps are mainly located in the intermediate part of the reservoirs, which meets the necessary conditions to carry out extraction pumping; when pumping the water, it causes the reservoirs to empty, preventing the river from flowing. One of the recommendations necessary to guarantee that the Tamul waterfall does not dry up is to organize the communities and establish pumping periods in which the water level in the ponds is observed.
According to the results, the most likely reason for the flow loss is a high infiltration due to its karstic nature. Therefore, it is recommended to study groundwater to reduce these filtrations without altering the underground ecosystems.
The process to determine the causes of flow loss could be used to determine the effects of extreme weather events on the rivers of our plan, such as droughts, and extreme rains, in any region of the world; in this sense, hydrodynamic models are a valuable tool for determining many hydrodynamic processes, such as the assimilation capacity of a river [23], heavy metals transport [37] or to determine the transport and distribution of any contaminant [42].
Finally, it is necessary to keep in mind that the numerical models are limited to the information collected in the field that is provided to them; for this reason, is required to use the most precise equipment, such as the use of drones, to be able to delimit the river and a more realistic topography, taking into account that in many cases it is impossible to access the affluent due to vegetation. Moreover when the current is very shallow the drone allows take the photographs to provide accurate bathymetry values, so it will only be necessary to perform a bathymetry, using equipment such as the ADCP, where there are areas in the river that the photography does not detect the bottom. This decreases the time of the measurement campaigns and increases the accuracy of the topobatimetric data collected.  Acknowledgments: This work was partially supported by projects 20210925, 20211518, and 20211789, as well as grants provided by SIP/Instituto Politécnico Nacional.

Conflicts of Interest:
The authors declare that there are no conflict of interest concerning the publication of this article.