Flooded areas 1- Processing of information One of the main factors involved in the flood-vegetation association relationship is the flood height. It is possible to modelise the flooded areas for different flood-heights on the Mopti flood gauge by assigning to each of the 14,535 vegetation units its average flood depth. In such a model, "intergrade mosaics" are problematical. We therefore decided to consider that an intergrade mosaic, for example BP/VB, whose constituent associations are set at flood level 7 for BP (flood depth between 4 m and 2.8 m) and at level 6 for VB (flood depth (between 2.8 m and 1.5 m), was to be assigned a composite level 76 (flood depth between 4 m and 1.5 m). In the calculation of flooded areas, it is always possible to “decompose” the composite levels by allocating a share of the area to each of the component levels. Thus for a mosaic at level 76 such as BP/VB, we can by convention allocate half of the area of the unit to BP, therefore at level 7, and the other half to VB, at level 6. If the mosaic spans over a stronger gradient, for example a mosaic at level 53, a third of the area is attributed to level 5, another third to level 4 and the last third to level 3, since an area bearing a mosaic at level 53 cannot physically pass from level 3 to level 5 without having a share of its area in intermediate level 4. However, while it is possible to decompose the mosaic composition by following these set rules in a somewhat arbitrary but plausible way, it is very difficult or even impossible to assign a precise spatial location to each of the components (see the method described in detail in Table 1A and summarized in Table 1, which can be downloaded) Table 1: Flooded areas by levels after mosaic decomposition The total area of the study is 2,229,950 ha The preceding table establishes a first "model" of potentially floodable areas in relation to different levels of flood depth. We must first define the expression "potentially floodable areas" we have chosen to use instead of the simpler "flooded areas". This comes from the fact that, in the model, when each of the conditions of the flood depth – namely flood levels other than level 1 – is successively reached, all the component vegetation associations characterised by that “average” flooding level are assumed to be flooded. However, we can only refer to potential flooding because the model treats each vegetation unit as an independent entity. As a result, the effects of topographic thresholds which would prevent a basin from being flooded, even if the flood levels corresponding to the vegetation associations it contains are reached, are not and cannot be taken into account. One can only think that the presence of these specific vegetation associations at this precise location indicates that the spot is usually flooded under the conditions described by the model, though without any certainty. Besides, the relationship between vegetation associations and flood height is based on a single flood gauge: that of Mopti, which serves as a reference. This assumes that the so-called reference flood is also valid for the other gauges in the Delta: Ke Macina at the entry of the Niger River into the Delta, Beneny Kegni or Sofara on the Bani River, and Akka at the exit of the Debo lake. To define the corresponding reference floods for these three gauges, we relied on the work of J.P. Lamagat "Analyse de la vitesse de propagation des crues, application à la prévision des crues et des étiages" , Orstom, 1983. This work makes it possible to define reference floods for each of those three gauges corresponding to different flood heights as measured in Mopti, but, as we will see later, "real" floods rarely correspond to reference floods. This also leads us to reflect on the meaning of the “zero” reference, which marks the limit between flooded and non-flooded vegetation areas. It is defined as the maximum flood height most regularly reached and its correspondence was established with a flood height of 660 cm at the Mopti gauge. The relationship between the “zero” reference and vegetation associations – under the conditions of validation of multivariate floristic profile/state of ecological variables analyses – therefore applies, regardless of which location is being considered in the Delta. However, this zero flood, other than at Mopti – where it corresponds to an altitude of 267.20 m – as well as at the three other reference stations (Ke Macina, Sofara, Akka), cannot be attached to a precise altitude everywhere else in the Delta. To move from a relative model, calibrated with respect to this “zero”, to a topographic model, it would first be necessary to know the relation uniting “zero” and altitude at every single point of the Delta. As a first approximation, we can assume that “zero” represents the trace in space of the surface generated by the maximum reference flood wave. This surface is probably complex corresponding to the period of slack between the end of the flood and the beginning of the recession, when the slope of the flow is at its lowest. We will later see how to try to solve this problem. According to table n°1, the potenrially floodable area corresponding to the reference flood as defined previously (660cm at the Mopti gauge) covers 1,820,289 ha, including the Farimaké area and the areas initially flooded first by run-off then by the flood. The second lesson to be derived from this table is show sensitive the Delta proves to be to small variations in the water heights. Between 660 cm and 600 cm, potentially flooded areas decrease by 7% to 9% for every loss of 10cm in water height. Below 600cm at Mopti, however, there seems to be a sharp shift in the regression pattern, with a less than 3% loss in flooded areas for each loss of 10 cm in water height. This suggests a very theoretical profile for the inner Delta, considered as a single entity – which it obviously is not. Its higher part appears to have a weak cross-slope, which makes it very sensitive to fairly small variations in water heights. Beyond level 4, however, deep basins with steeper cross-slopes are therefore less sensitive to such variations. 2- Flood mapping: the CRUE3 layer CRUE3 layer is derived from VEG4 by copying and creating specific items: H_0, H_10, H_30 ...Among the ecological variables, two were selectively chosen: soil texture, which has been briefly dealt with in the preceding part, and flood heights or depths. •SIGLE (fr:Sigle): is directly derived from VEG4 and matches each vegetation association with a geographical unit. •LEVEL (fr:niveau): concerns the flood level of the vegetation association. A number between 1 and 7 is ascribed to each vegetation association, 8 being reserved for open water. (see table n°3 page 39 : the relationship between vegetal associations, water heights and the Mopti gauge measurements) The mosaics are represented by a two-digit number in reference to their component associations. Thus BP/VB, respectively belonging to levels 7 and 6 will be coded 76, while O/VOR will be coded 55 since its components both belong to level 5. For the sake of simplicity, single vegetation associations are coded from 11 to 77, with 80 reserved for MB and 90 for open water. •HIGH (fr:Profond): translates the LEVEL item into water depth. The detail of those heights is discussed page XXX. Let us just say that it corresponds to the bottom level of the corresponding water range for single vegetation formations, and the average one for the mosaics. Thus B will be given a depth of -2.80m in keeping with level 66 and B/VOR will be given a depth of -2.15 cm, in keeping with level 65. Non-flooded vegetation formations (TA to TS) are conventionally given a 0 depth, so that P/TA, at level 21, will be given a depth of 0.05m (average between -0.1m and 0m) As we are going to see, such conventions required by cartography tend to maximise flooded areas by comparison to table 1, which was reached by de-composing the mosaics •H_0 to H_280: these fields derive directly from the HIGH item. They are of the yes/no type and contain the following numerical values: •0 when the area is not flooded under the conditions of the field H_0, H_10, etc. •1 when the area is flooded under the conditions of field H_0, or H_10 etc. •2 when the area is first flooded by run-off (PAN, PAR, PAS, PAM type formations) under field conditions H_0, H_10, etc. • The same reasoning applies to items H_30, H_60, H_150, H_280. • A convention is set for mosaics including associations PAN, PAR or PAS, PAM. When one of the two associations is non- flooded as for PAN/TA for example, the mosaic is considered first flooded by run-off, therefore coded 2. When the other association within the mosaic is a flooded type – as for PAN/ZB for example – the river flood prevails on the run-off flood and the mosaic is coded 1. This scenario only concerns a very small number of polygons. • For a flood reaching 660 cm, the floodable surfaces occupy almost the entire Delta. The Farimaké in the northwest is largely flooded by run-off first, with the flood coming in late (November-December. Inside the basin, the spaces that remain exposed are mainly located in the following areas -first, along a double string of "toggere" forming an alignment running parallel to the main course of the Niger river; from Koubaye to the south (at the latitude of Mopti), this turns into a large tree-shape area around Dialloubé, south of Lake Débo. -second, east of Djenné, the erg of Femaye, along the Bani and the highlands of southern Sébéra. -third, near Diafarabé, south of Niger and west of the Diaka, between the defluent and the western margin. • For a flood reaching 630 cm, the western margin of the flood recedes and approaches Ténenkou. The " Togge " occupy a larger area and in the southern part, the water table in Djenné is already visibly beginning to separate into a northern basin and a southern basin. • For a flood reaching 600 cm, the highlands in the center-east of the Delta form a continuous area from Kouakourou north of Dialloubé. The fragmentation of the water table which completely covered the bowl at 660 cm is now well marked. To the west, the flood continues to stretch massively from Ténenkou to Lake Walado; to the east, it is still continuous from the Bani-Niger mesopotamia, continues along the right bank of the Niger from Mopti to Konna before joining Lake Débo in the north. In the southern part of the Delta, the separation of the basins to the right of Djenné is almost complete and the highlands of Diafarabé are out of water, except for a string of pools south of the river. • For a flood reaching 510 cm, the majority of the Delta basin is no longer flooded and the highly fragmented water surfaces occupy only the heart of the deep basins which constitute the resistant core of the inner Delta Table 2: Areas potentially flooded by levels of flood height As a conclusion, the model we have sketched allows us to calculate and map out potentially flooded areas for each class of water-heights. The presence of intergraded mosaics makes it necessary to define conventions by which the latter are allocated to specific level, so that numerical results will differ between tables 1 and 2, with the cartographic method overestimating the areas concerned. Nevertheless, the cartography allows us to catch a glimpse of the way in which the inner Delta is structured, with deep basins (see move form 600 cm to 510 cm) whose precise contours, contents and boundaries cannot be identified. This matches the conclusions derived from the analysis of the map of vegetation formations which shows how very subtle combinations allow the Delta’s structure to display several distinct vegetation landscapes. We shall attempt to further establish the Delta’s structure by moving from a discrete model to a continuous one relying on matrix data, allowing us to move to a 3D model of the potentially flooded areas. We shall also endeavour to derive a Digital Elevation Model of the Delta from it, after setting the relevant altitudes for the reference flood.

* Level 8 corresponds to water (Niger, Bani, Lakes….) **Level 2 corresponds to a 10 cm layer (0 - 10 cm), level 3 to two slices (10 - 30 cm), level 4 to three layers (30 - 60 cm), level 5 to nine layers (60 - 150 cm), level 6 to thirteen layers(150 - 280 cm) and level 7 to twelve layers (280 - 400 cm). The height of 380 cm has never been observed in Mopti as the maximum height of an annual flood, the lowest recorded value was 440 cm in 1984 .

Download