4.1. The September 10th -15th 2019 event: rainfall distribution and intensity and runoff
Between September 10th and 14th, 2019, a cut-off low coming from the polar jet stream crossed over the southeast of the Iberian Peninsula. At the same time, a storm located over Algeria sent warm and humid air through the Mediterranean, which, together with orographic forcing, produced the torrential rains (Amengual 2022). Rainfall was concentrated mainly on September 12th and 13th -especially the first day-, due to the passage of successive storms. Figure 2 and Table 2 show the total distribution of rainfall that occurred during the event. Rainfall increased from the west (< 100 mm) towards the coast (between 250 and 335 mm), where the highest intensities were recorded, especially in the central and southern basins. The estimated average rainfall calculated is 208 mm, and the estimated volume is 266 Mm3. The maximum rainfall accumulated in 24 hours was very high: La Puebla, 269.7 mm; San Javier, 216.6 mm and Pozo Estrecho, 206.6 mm. The maximum intensities surpassed 180 mm/h.
The rainfall accumulated in 24 hours, the high intensities and the structure of the three storms, accelerated the runoff processes, generating a very fast response time. According to press reports, Los Alcázares were flooded at 9 in the morning of the 13th of September, which gives a response time of between 1 and 3 hours, which is quite common in Mediterranean ephemeral rivers (Camarasa and Segura 2001), particularly considering that during the third episode the basin was saturated by the preceding rains.
The hyetograms of the Rambla del Albujón (Fig. 3) show three different phases:
- First peak: Occurs between 20:45 on the 11th, and 4:05 on the 12th, with a maximum intensity of 33.6 mm/h recorded in the Rambla del Albujón, without effective runoff production (Table 2 and Fig. 3).
- Second phase: Between 15:40 and 16:45 on September 12th. Higher values in the western part. Maximum recorded in Fuente Álamo (53.6 mm/h at 16:10), with a slight production of runoff.
- Third phase: The highest intensities. Between 22:40 p.m. (12th ) and 7:00 (13th ). Maximum intensity of 182.8 mm/h recorded at 00:50 a.m. in La Puebla, producing the flow peak.
The hydrographs (Fig. 3) show the formation of the flash flood with an increasing flow between observatories 8 and 9 and the lamination of the flood at observatories 10 and 13, since from the first observatory onwards the channel overflowed, and part of the flood flow was dispersed.
|
Rain meter
|
Total rainfall
|
Accumulated rainfall 24 hours
|
Maximum intensity (1 hour)
|
Maximum intensity (day/hour)
|
Table 2
Rainfall characteristics in the different observatories.
1
|
Majal Blanco (AMETSE)
|
200.4
|
123.2
|
88.8
|
13th Sep. 2:15
|
2
|
Relojero (AMETSE)
|
211.9
|
137.6
|
51.6
|
13th Sep. 7:05
|
3
|
Canal Campo Cartagena
|
252.8
|
181.6
|
123.6
|
13th Sep. 16:55
|
4
|
San Javier (El Mirador)
|
218.4
|
172.6
|
128.7
|
13th Sep. 2:10
|
5
|
Murcia/San Javier II
|
220.8
|
160.0
|
141.6
|
13th Sep. 4:00
|
6
|
San Javier (La Manga)
|
334.8
|
216.6
|
192.0
|
13th Sep. 4:00
|
7
|
La Murta
|
165.2
|
84.6
|
204.0
|
14th Sep. 4:05
|
8
|
Fuente Álamo (Rambla Albujón)
|
148
|
76.0
|
53.6
|
13th Sep. 5:40
|
9
|
El Estrecho (Rambla Albujón)
|
163.4
|
101.4
|
43.2
|
13th Sep. 5:40
|
10
|
Rambla del Albujón
|
186.4
|
136.4
|
62.4
|
13th Sep. 3:25
|
11
|
Torre Pacheco
|
240.6
|
199.0
|
110.4
|
13th Sep. 1:15
|
12
|
Pozo Estrecho (Rambla Albujón)
|
255.2
|
206.6
|
110.4
|
13th Sep. 1:15
|
13
|
La Puebla (Rambla Albujón)
|
310
|
269.7
|
182.8
|
13th Sep. 0:50
|
14
|
Mazarrón/Las Torres
|
50.4
|
22.6
|
21.6
|
13th Sep. 2:50
|
15
|
Los Patojos (Rambla Benipila)
|
123.2
|
97.5
|
63.6
|
13th Sep. 5:05
|
4.2. Analysis of flood controlling factors
4.2.1 Geomorphology of the alluvial fan apron system
The alluvial apron system of the Menor Sea (Fig. 4) is made of different coalescent fans formed throughout the Quaternary by ephemeral streams in arid and semiarid conditions (Harvey 1996). The fans have slopes between 5 and 15%, slightly higher in areas with neotectonics. Its age decreases from upwards to downwards. Its genesis is due to different types of flows: a) debris flow, with highly heterometric materials in proximal areas; b) water flows (sheet floods and braided streams), in the central part of the fans; and c) overflows and low energy flows, in the distal areas, with fine materials (Conesa 2006).
The dating of these alluvial fans is complex. Somoza et al. (1989) distinguished 4 prograding systems -with different levels in each group-, corresponding to the Lower, Middle and Upper Pleistocene, and Holocene, all of them crusted, except the last one. The GEODE geological map (Marín et al. 2009), distinguished 6 levels of fans (Fig. 4): the first generation belongs to the Lower Pleistocene; the second to the Middle Pleistocene; third, fourth and fifth to the Upper Pleistocene; and sixth to the Holocene. The arrangement, the type of profile and the tendency to aggregation or dissection are clearly conditioned by neotectonics, which has caused the tilting, overlapping, progradation or incision of the different fan systems (Somoza et al. 1989).
The complex topography of this set of overlapping fans has a determining influence on the response of the Menor Sea basin to floods. There is an alternation of topographically concave and convex sectors that has a direct influence on the distribution of the drainage network, and therefore, on its behaviour during flash flood events. This leads us to distinguish different subunits in the alluvial apron system, which follow an iterative sequence from upstream to downstream (Figs. 4 and 5):
- Upper and Middle Pleistocene fans. Convex topography, with signs of neotectonics, covered with calcareous crusts and strongly dissected by the divergent drainage network.
- Interfan areas, located between these alluvial fans. They are depressed areas where a convergent channel network develops, which generates a new generation of prograding fans. These interfan areas constitute the basin area of the next fringe of younger fans (Lower Pleistocene).
- Lower Pleistocene fans. In this way, in the next generation of fans, the topography is reversed, and gives place to a new fans-interfans sequence. Older fans are susceptible to trenching, by distal or interfan-induced dissection by an incising fan-surface wash, according to the model proposed by Harvey (1996).
- Holocene fans. Prograding, with a braided drainage network and poorly defined channels. Active fans and with an intensive occupation that has altered the topography, which makes them difficult to identify and delimitate.
The arrangement of the different levels of fans implies different sedimentary behaviours. Fan-head trenching as well as inter-fan channel incision induces the dissection of the different levels of Pleistocene fans. In contrast, the Holocene fans are in a prograding phase, with sedimentation processes associated to braided ill-defined drainage networks (flat-bottomed ravines). Thus, as the incision of the fluvial network decreases from upwards to downwards, flood hazard increases in the same direction, being maximum in Holocene deposits.
4.2.2. Fluvial network: natural forms and human activity
The drainage network of the study area has complex characteristics. There are two types of channels: incised and flat-bottomed ravines (Fig. 6). Incised channels appear in areas with steep slopes, on Triassic and Miocene materials, with a certain hardness and consolidation (grey conglomerates, quartzites and other metamorphic rocks). They have clear “v” sections in the headwaters area, forming very marked concavities in the contour lines. Because of their pronounced incision, they have not been occupied by anthropic uses (Fig. 6). Upon reaching the foothills, the longitudinal profile of the ravines could be interrupted. Nevertheless, some of them are capable to continue through the next (downstream) fan unit. Longitudinal profiles can be concave (as they run through Pleistocene fans) indicating a high degree of maturity (Fig. 7), or straight (Holocene fans), when they are recent or are strongly anthropized (Fig. 8).
Flat-bottomed ravines, locally known as cañadas, predominate on soft materials, especially marl or Quaternary deposits (gravels, silts and sands), in some cases crusted (Figs. 5 and 6). They have been markedly anthropized. Traditionally, farmers used them to build terraces and to plant tree crops to take advantage of the soil humidity. They are characterized by a U-shaped cross section, and by the absence of fluvial forms and sediments. From the genetic point of view, they can be current channels or paleochannels. They are partially incised in the headwaters, but this incision significantly decreases when the slope decreases (Fig. 9). Downstream, they have a braided morphology, and the hierarchy of the network is blurred. There, the changing ravine courses and the unclear boundaries of the watersheds cause a high level of uncertainty and danger during floods (Figs. 9 and 10).
These natural drainage network formed by flat-bottomed and incised ravines has been altered by anthropic action since ancient times, through three techniques: terracing; irrigation with flood waters; and diversion and/or conversion of ravines into streets in urban areas:
a) Terracing is a Mediterranean traditional technique to retain water and sediment on the slopes with rainfed crops (Morales 2021). In the plots of these terraces, ridges were built that protruded between 30 and 50 cm, in order to retain rainwater, to reduce erosion and to increase infiltration. Particularly important is the terracing of the flat-bottomed ravines, which constituted a very dense network in the study area, appreciable in the 1956 orthophoto due to the presence of tree crops (Figure 10). Currently, most of these plots small have been levelled to create larger plots (Garcia-Ayllon and Radke 2021), so it is more difficult to identify the channels. Despite this, these systems still form drainage routes that are activated during floods (Figure 10). The dismantling of these historical semi-artificial drainage systems terraces has altered the traditional hydro-sedimentary balance, resulting in an increase of runoff and sediment discharge (Moreno de las Heras et al. 2019).
b) The construction of boqueras and sangradores is another technique for taking advantage of flood waters. Boqueras are weirs that divert water from the ravine beds to the terraces, and sangradores are spillways that divert excess water from one terrace to another (Hernández and Morales 2013; Marco-Molina et al.2021; Fansa and Pérez Cueva 2021; Morales 2021) (Figure 11). In the study area, Morales (2021) estimated that in mid-20th century, the area of rainfed crops on terraces and/or intermittently irrigated by boqueras amounted to 80,000 ha. However, in the second half of the 20th century, these ancestral techniques were abandoned due to the rural exodus and the introduction of machinery, which could not operate small terraces and steep slopes. The arrival of water for permanent irrigation from the Tajo-Segura water transfer, after 1979, completely transformed the crop and plot pattern of this area (Morales 2021).
c) The rapid urban growth without sound planning instruments (Pérez et al. 2017; Caballero 2017; Gil-Guirado et al. 2022) has led to the incorporation of the channels to the urban network (ex. La Rambla del Albujón in Los Alcázares) or/and the construction of artificial river channels to divert floods (ex. The Rambla de la Maraña in Los Alcázares). These waterworks have had very negative consequences during the flood of September 2019 (Figures 12 and 13).
4.3. Land use change
Land use has radically change from 1956 to 2018, from a traditional landscape of rainfed Mediterranean agriculture, to a model of intensive irrigation and tourism development. Only the mountains, where forested areas slightly increased, show some stability during the last 7 decades (Fig. 14 and Table 3).
The most significant change is the transformation of the rainfed areas (some of them with intermittent irrigation with flood waters), into intensive irrigated areas (open-air or greenhouses). The first occupied 75% of the study area in 1956 and only 13.7% in 2018, and the second expanded from 0–51.3% in the same period. Currently, the residual rainfed crops only occupy some flat-bottomed ravines and crusted areas. Urban areas and roads expanded surrounding the Menor Sea in this period, from 0.4–9.2%, occupying 13% of the Holocene fans (37.8% in the Rambla de Cobatillas basin).
Land use change has caused an important soil sealing process, due to the construction of greenhouses, irrigation ponds, roads and buildings. The northwest basins have the highest values of soil sealing. If only the sealed area in Holocene fans is considered, the percentage considerably increases, being higher than 50% in some basins. Moreover, irrigation expansion has dismantled the original plot structure, creating large estates, flattening terraces and disabling boqueras and sangradores.
Table 3
Land use change between 1956 and 2018.
|
1956
|
2018
|
|
ha
|
%
|
ha
|
%
|
Rainfed
|
97,046.0
|
75.0
|
17,504.7
|
13.7
|
Forestry
|
28,199.0
|
21.8
|
28,391.0
|
22.2
|
Irrigated
|
0.0
|
0.0
|
61,349.2
|
47.9
|
Built-up or asphalted
|
578.0
|
0.4
|
11,681.0
|
9.2
|
Greenhouses
|
0.0
|
0.0
|
2,917.2
|
2.3
|
Ponds
|
0.0
|
0.0
|
1,444.6
|
1.1
|
Infrastructures
|
0.0
|
0.0
|
1,544.4
|
1.2
|
Dirt roads
|
2,054.0
|
1.6
|
2,742.5
|
2.1
|
Salt pans and marshes
|
699.0
|
1.2
|
492.4
|
0.4
|
4.4. Flooded area: dimension and factors
The calculation of the ITW index from the Sentinel image has allowed the delimitation of the flooded area at 10:50:31 on the 13th of September (Fig. 15). Despite the poor quality of the data, at the Albujon gauging station, the maximum flow was recorded at 2:00 a.m. on the 13th, so the image of the area overflowed corresponds to the recession limb of the hydrograph and we could assume that the overflow phase had ended. In fact, pixels with water can only be seen in the riverbeds in the lower basin, while several plumes of turbid water can be identified in the Menor Sea.
The characteristics of the flooded area obtained from the Sentinel image are (Fig. 15):
- The total flooded area occupies 50,157,999 ha (18.5% of the study area), adjusting well to the drainage network reconstructed in this work (Fig. 6), but not to the official network (Fig. 1).
- The flooded areas are mainly located in the lower areas of the basins (Holocene fans) of the Rambla del Albujón, Rambla de la Miranda and Rambla de la Maraña. Sheet flooding -typical of active fans governed by a braided ill-defined network (NRC 1996)- occurs in these areas.
- In the upper and middle basins, pixels with water mainly correspond to the flat-bottomed ravines.
- Given the scarce incision of the flat-bottomed ravines in the lower basins, during floods these channels can mutually interconnect, blurring the boundary between the different basins.
- Paleochannels distribution explains the flooding of some areas, such as the Rambla del Albujón, whose paleochannels collect the overflowing water, conveying it towards the Rambla de Miranda (Fig. 15). The hydrographs (Fig. 3) show the flood abatement from gauging 10, the point where the connecting paleochannel between the two ravines begins.
- Infrastructures perpendicular to the drainage network and with low permeability acts as a barrier to overflowing flows. The most prominent case is the AP-7, interrupting the flow towards the Menor Sea (Fig. 15).
- The coastal urbanized areas have suffered the highest impacts, since they interrupt the drainage networks towards the Menor Sea. In Los Alcázares, due to the incorporation of the Rambla de la Maraña into the urban street network and its artificial channelization (Fig. 13). In San Javier, due to the high proportion of the sealed Holocene area (Fig. 14).
4.5. Comparing the 2019 event with official hazard mapping (SNCZI)
The Sentinel-2 image is quite similar to the 50-year return period (T50) official mapping (SNCZI), with a coincidence of 29.7% (Table 4). The similarity with the 100T (26.6%) and 500T (24.7%) is lower, while the similarity with 10T is even lower (21.6%). However, the return period and the magnitude of the September 2019 event are still uncertain, since rainfall data series is too short (since 1987) and there are no long gauging data series (SNCZI mapping is based on simulations that extrapolate data from nearby perennial rivers). The observed differences between the 2019 event and the SNCZI can be explained considering the following facts (Fig. 16):
a) The SNCZI overestimates the danger of flooding in the middle basins of the oldest fans, dominated by through-fan trenching (Harvey 1996). The distribution of the flooded area in these sectors corresponds to the position of flat-bottomed ravines, which should be considered as channels and not as a flood zone (Figure 15).
b) The greatest coincidence between the Sentinel and the SNCZI occurs in the Holocene basins (6th generation fans). In fact, the SNCZI predicts that between 60 and 68% of the flood zone (according to the different TR), is located in the Holocene zone (supplementary material). The Sentinel image shows 70% of the pixels in this sector. The southern ravines (Beal, Matildes, Miedo) have 99% of the flooded area in Holocene fans, while the larger ones, such as Miranda, Albujón or Maraña have between 60 and 65% (supplementary material).
c) The observed barrier effect of the motorway is higher in Sentinel images than in SNCZI T10 and T50 mapping.
Table 4
Comparison between the flooded area extracted from the Sentinel image and the different return periods estimated by SNCZI flood hazard maps.
|
SNCZI (ha)
|
SNCZI (%)
|
Sentinel (ha)
|
Sentinel (%)
|
Coincidence (ha)
|
Coincidence (%)
|
T10
|
4,769.2
|
39.8
|
4,637.1
|
38.7
|
2,590.2
|
21.6
|
T50
|
7,043.9
|
49.4
|
2,983.8
|
20.9
|
4,243.5
|
29.7
|
T100
|
10,324.3
|
58.8
|
2,557.9
|
14.6
|
4,669.4
|
26.6
|
T500
|
13,291.6
|
64.8
|
2,167.6
|
10.6
|
5,059.7
|
24.7
|