Amboseli Hydrology (M.D. Mifflin, ca. 1993)

[The following report was scanned directly from an undated typescript by M.D. Mifflin circa 1993. We believe that the report had been commissioned by Dr. Richard Leakey when he was Director of Kenya Wildlife Service.]

Amboseli Hydrology - (report by M.D. Mifflin, ca. 1993)

Amboseli National Park totally lies within a closed drainage system. No major surface streams flow into the Basin. The park topography is flat while the soil composition is fine and forms a surface seal when wet and hence rainfall quickly collects in pools throughout the park. The major source of water for the park is from the springs which form Enkongu Narok, Longinye and 0l Tukai swamps Melting snow and rainfall on Kilimanjaro to the south infiltrates into porous lava terrain before reaching the lower foot hills. It then re-emerges in the Amboseli Basin as permanent springs and seeps along the .' foot of the piedmont to form the swamps. There are no perennial streams that reach the! '" lower piedmont of Kilimanjaro in the area to the south of the Amboseli area.,

Terms of Reference

This hydrologic review of the hydrologic and hydrogeologic conditions at the Amboseli National Park has been performed at the request of Dr. Richard, Director, Kenya Wildlife Service, under the world Bank Agreement. The review period was two weeks in duration; therefore, the findings, conclusions and recommendations are based on generally available information and should be tested, and perhaps adjusted or refined as more complete information is developed. Nevertheless, the database is reasonably complete to form the interpretations and conclusions established in this review. The many interested experts and observers interviewed during the review have materially aided the review and have increased my confidence in the conclusions.

The terms of reference for this review of the hydrology of the Amboseli National Park can be summarised as follows:

I Determine the cause of the increasing extent of marsh and rise of water table within the Park.

II Establish recommendations in the view of the findings and conclusions for incorporation/consideration in the 1992 Amboseli National Park Management Plan.

Within the above terms of reference, the following review approach was adopted:

I Identified the probable causative factors for the changing hydrology.

II Established the changes (quantitatively or semi- quantitatively ) over time.

III Develop a set of conclusions and recommendations based on the above and the general management objective for the Park.

Closed Basin Hydrology of Amboseli :

At the beginning of the review, possible and/or plausible causes of hydrologic changes of Amboseli National Park were considered recognising that:

a. The area is a hydrologically closed basin.

b. The water supply to the Park is seasonal surface-water run-off during two short rainy seasons and perennial ground-water discharged, in part from large spring areas.

c. The topography is very flat.

The areas of standing water and marsh will expand and contract markedly on a seasonal basis due to combination of direct surface-water runoff (Lake Amboseli) short and long- term changes in the flux of ground-water discharged, and seasonal evapotranspiration and direct evaporation. the best way to understand the changing areas of marsh and lake surface is to consider the following continuity equation (Mifflin and Wheat, 1979):

Annual input of moisture = annual output of moisture OR

Rp + Dgw = (Ale + Ame + Amp) (Et - Pt)

Where:

Rp is the total annual volume of runoff which reaches the Amboseli lowlands.

Dgw is the total annual volume of ground-water discharge in the lowlands.

Ale is the area of the ephemeral lakes (mostly Lake Amboseli).

Ame is the area of ephemeral marsh (mostly the northern extent of Longinye swamp and around Lake Conch and Sinet swamps.

Amp is the area of perennial marsh or swamps.

Pt is the total annual direct participation on the Amboseli lowlands.

Et is annual evapotranspiration of direct evaporation (estimated at 2200mm/year).

Unfortunately, the above continuity equation has several parameters that cannot be quantitatively evaluated with the present databases. However, an approximate minimum estimate of the order of magnitude of ground-water discharge (Dgw) can be estimated for several periods where the mapping of perennial marsh has been established by ignoring the ephemeral march areas (Ame) and ephemeral lake areas (Ale) as well as the ephemeral runoff (Rp). The approximating equation becomes:

Dgw = Amp (Et - Pt) i

This approach assumes the perennial marsh areas are totally supplied by a steady rate of ground-water discharge, and the ephemeral marshes are related to only seasonal rainfall and runoff.

The closed basin nature of the basin creates the expanding and contracting habitat (perennial marsh, ephemeral marsh, and short term areas inundated by seasonal runoff or direct rainfall that ponds. The extremely flat nature of the terrain in areas surrounding north extent of Longinye swamp and the margins of lake Amboseli (Conch and Sinet lakes) cause relatively small seasonal changes in water volumes to spread over large areas, which in turn creates a variety of seasonal habitat for grazing herbivores. It also creates management problems in terms of ~e practical design of park infrastructure such as access roads and tourist facilities. Expanded problems created by markedly increased changes in the total supply has prompted this hydrologic review.

FINDINGS AND CONCLUSIONS

The following findings and conclusions seem to be justified by the observations and databases.

I. The rapid increase in ground-water discharge in the Lake Amboseli region springs beginning in October 1957 is best explained as a result of a climatic changes causing more effect ground-water recharge on the flanks of Kilimanjaro. It seems likely that the lake levels for Lake Victoria as reported in Lamb (1966) are reasonable analogue for the timing of climatic changes in the Lake Amboseli region. Accordingly, there apparently occurred a 80 year period of reduced ground-water discharge which ended dramatically in 1957 when Dave Lovatt Smith and Tabs Taberer observed the ancient spring channel become wet near Observation Hill near the end of the dry season (Lovatt Smith, 1986). It is recognised that the catchment basin of Lake Victoria is within a differing climatic zone, yet the timing of the Lake Victoria lake level changes fit reasonably well with the sparse Amboseli region evidence and other, less complete climatic records. An attempt. was made analyse the precipitation records from the Amboseli region, but the database was not received in time to test this interpretation with the local records. Three other postulated causes of increased ground-water discharge at Amboseli have been suggested over the years, but no evidence has been recognised to : support these postulates. One suggests that vegetation change on the flanks of Kilimanjaro might have increased the recharge rate. Another suggests seismic events have changed the pattern of ground-water discharge, and a third calls upon melting glaciers to add more run off for recharge.

II. There is good evidence that the increased water supply to the springs has persisted to the present since the change in the late 50's, with probably two cycles of increased/decreased discharge superposed on the overall increased vigour of the system. . The overall ground-water discharge has remained at state of flow that is markedly greater f (1991 ground-water discharge about two times the 1950 rate of ground-water discharge) and there is no clear way to predict as to whether, or when, the climate may shift back to the more arid state experienced during the late 19th century to the mid 20th century.

III. The gradual rise in the water table in the ground-water discharge area of the Lake Amboseli region has been the result of increased recharge (an increase in effective moisture for ground-water recharge). The rise in the water table has occurred and been documented in Sinya Meerschaum Mine area as well as the Amboseli Park area (Lahi, 1967). This argues strongly against explaining the increase in spring discharge in Amboseli Park by a localized shift in position of discharge due to seismic events. The increase in ground-water discharge is widespread.

IV. The high water table in the areas of the Amboseli Park tourist facilities (0l Tukai) has resulted from three factors in the past 2 years.

A. The long term increase in recharge caused by the climate shift and a gradual regional rise in the water table in the discharge areas in response to the greater flux of the ground-water system.

B. The shift of the pattern of permanent marsh to the South near spring area to a pattern of marsh development well to the North of the former marsh area with the majority of perennial water and associated marsh development now extending to and circling around the Park facility area.

C. The heavy rainfall event (approximately 3 inches) on March 29, 1991.

V. The above factors have caused the water table to rise to within about one meter of the land surface in the 01 tukai facility area, with the capillary fringe extending to the land surface in some areas. The heavy rainfall event caused extensive and prolonged flooding because the vadose zone (unsaturated zone) was very thin (about 1m) and partly saturated by the capillary fringe in the fine-grained soil. This condition does not allow for sufficient porosity for infiltration of heavy rainfall events, and therefore the majority of the precipitation now ponds and remains on the surface in the facility area.

VI. Topographically unconfined perennial marsh in the extremely flat areas of Longinye swamp tend to shift over time due to several complex processes. The perennial' marshes are extremely productive in terms of aquatic plant growth, and the organic debris tends to raise the level of the water body through accumulation of a layer of organic muck. The large herbivores tend to channelize and mobilize both the organic debris and associated floating aquatic vegetation mats. in addition, rooted aquatic vegetation mats. In addition, rooted aquatic plants, such as sedge, various reeds and papyrus, may form effective barriers to water flow when compacted by the large herbivores. this process is possibly aided in part by varying (seasonal and cyclic) changes in spring discharge as well as the growth cycles.

VII. There is no evidence of significant changing land use or changes in vegetation patterns in the most important catchment areas of recharge for the ground water that discharges in the Amboseli Lake region. These areas are postulated by most observers (including this observer) to be on flanks of Kilimanjaro. The North flank slope of Kilimanjaro has escaped any important land use changes in terms of total area. There is no evidence of heavy grazing on the lower piedmont and there has been little change in the forest communities on the higher zones (believed to be the most important recharge areas) with possible uncertain impact of fires above the forested areas. There has been a general recession of the glaciers on the mountain in historic times, but the volumes of water related to net recession of ice are not of the magnitude required to equal the marked increase in ground-water discharge as measured by the net increase in perennial and seasonal marsh areas in Amboseli National Park. Thus, climate is the best explanation.

VIII. In 1950 there appears, on the basis of airphoto interpretation, to have been approximately 17.25 x 10 m3/year of ground-water discharges. In 1979 there appears to have been approximately 31.5 x 10 m3/year of ground-water discharge. These quantitative estimates do not include seasonal marsh that may also be fed in part by the spring discharge and therefore they are likely minimum estimates.

IX. There is good evidence that the extent of perennial and ephemeral marsh that exists in 1991 is within the range of ground-water conditions that have occurred in former "modem" climates. Aerial photographs clearly show several old out-flow channels leading to Lake Amboseli from the spring/marsh areas. These are defined by continuous and very faint linear depressions with associated organic soils. They are likely late Holocene to "modem" in age (they have not been infilled for the most part). However, the earliest observations by European explorers and the traditions of the Maasai do not include any accounts of such conditions in the late 19th century or early 20th century. If Lamb's Lake Victoria water-level record is related to the climatic pattern for the Lake Amboseli region, the last period of expanded perennial marsh and outflow channel occupation by perennial flows was likely about 1880, too long ago for the early local accounts to have documented the broader extent of marsh.

X. The fact that there are at least three ancient outflow channels from the Longinye swamp (or 01 Kenya swamp) area to the northeast extent of Lake Amboseli indicates that the shifting patterns of swamp development observed since 1957 were also operative in pre-history times. The extremely flat nature of this area probably caused the development' of several overflow channels as the ancient marshes shifted, just as they have been shifting with the documentation period.

XI. The Enkong Narok swamp has been confined to date in a natural overflow channel that is sufficiently incised into the ancient plain (and partly diverted to the Conch lake area by an artificial channel). This does have a potential overflow (ancient) channel course about midway that could be reoccupied as the organic debris continues to build and raise the general level of the permanent marsh. This ancient channel trends northeast .: then north towards the extreme northeast comer of Lake Amboseli (see Fig. 7 in Western, ! 1973).

XII. The hydrographically closed basin and extremely flat topography and the hydrologic role of the area (a sink for both surface and ground water) tend to concentrate salt carried to the area by both the surface water and ground water. Therefore, it is important to recognise the benefits of the washing action of remobilizing salt when the periodic expansion marsh or flooding occurs. Salts are concentrated in areas with both capillary rise and evapotranspiration of water, and direct evaporation of standing water. Therefore, all water management practices must take into consideration the need for periodic rinsing in areas of vegetation, whether by direct rainfall, or expanding and contracting flooding. the migration of ephemeral marsh and periodic flooding tends to redistribute the salts to lower areas over time, and this pattern of salt redistribution is probably key to healthy plant production.

XIII. The very high water table may weaken or kill the remaining 01 Tukai mature fever trees if it persists due to the partial drowning of established root systems. Younger fever trees may not be as heavily impacted because the root system will be better distributed above, the water table. This conclusion is tentative as the evidence is not clear due to the relatively short period of very high water table and the prior loss of fever trees due to elephant damage. I believe, based in part on the Western (1973) study in the late 1960's, that the rise in water table in the early 1960's from a level of about 10m to a level of about 5m below land surface killed the majority of trees. Similar die offs are common world wide when the combination of a strong water table rise and associated salt concentration tend to drown root systems of mature stands of trees, and move higher concentration of salts into a much reduced vadose zone (unsaturated zone). ~e elephant damage factor and regeneration of young fever trees tend to confuse the i8~ue, but experience from elsewhere and Western's database argue strongly for the rise in water- table cause.

XIV. The calcareous deposits called limestone ( sinet channel cut) and sporadic cemented caps to the north of Ol Tukai on the plains are likely formed by ancient former extent of the marshes supported by ground-water discharge and spring flow. They suggest a long history of marsh complexes and widespread shifting of marsh areas and associated shallow water-tables. Much of the calcareous material is likely related to shallow water- tables and spring discharge in semi-arid climatic conditions. Often a combination of shallow capillary fringe and direct precipitation of calcite in spring flow ponds create! such deposits when moderate to high Ca HCO3 waters discharge in the semi-arid or arid boson environments. Similar deposits in many parts of the world and typically have been interpreted (incorrectly) to be lake deposits. Careful study of the fauna associated with the deposit usually demonstrate wet meadow/marsh ecology.

RECOMMENDATIONS

The preceding findings and conclusions lead to the following recommendations. These attempt to incorporate the objectives of the Kenya wildlife Service (KWS) for the management of the Amboseli National Park, the realities of the hydrology, and the maintenance of the unique ecological systems that are dependent upon the seasonal and perennial marsh areas. the recommendations are professional opinion, and draw upon experience as well as the experience of several keen observers who freely offered their expertise, data, time and insight to aid materially in this investigation. it should be pointed out that general consensus seems to exist, and the recommendations that I offer herein are consonant with the judgement of the individuals that materially aided in the investigation. However, I take full responsibility for recommendations.

I. The Park should be managed on the basis of the assumption that the hydrologic conditions which began in the late 1950's will persist for the foreseeable future. There is no confident way to predict if and when the climate will shift back to the pre-1957 state.

II. Due to the importance and necessity of managing the increased amount of water supply and resulting expanded marshes and favourable wildlife habitats, a program of hydrological monitoring and associated water management should be developed over the next several years. The monitoring of ground-water levels, associated salinities, and surface-water monitoring, and associated salinities should be established. Good information on the depth to the water table will be key in reaching good management decisions for both water and habitat management. The same will be true for surface-water management. Dramatic ecological changes have already occurred, and future changes should be anticipated, many of which may be costly to infrastructure or detrimental to habitat without good information base to act upon. Surface-water management is also already necessary (desirable) in order to maintain desired habitat, and continue to protect infrastructure investments.

III. The ground water levels can only be managed to a limited degree. They can perhaps be lowered to about the 3-4m level in the Oltukai area if the Longinyie Swamp spring water can be trained to reoccupy former permanent marsh areas (a more northern trend in perennial marsh pattern). The attempt should be made to train part of the spring flow in ,a phase I effort to the former northern channels which head in the spring source area. I suggest that relatively coarse fill be trucked to the SW side of the source area about 300m! from the source head to construct a partial dike trending due north. The dike should be about 1 m. high above the existing marsh level, trend due north, and cross the channel which carried about 40 to 50% of the source area flow estimated (crudely measured in the field) at 18 cfs in Oct, 1991. The dike should be carefully extended until the flow shifts to the former channels trending due north and they accept the total flow. Care must be taken during the construction as the perennial marsh bed will be a thick accumulation of organic muck. The top of the dike should be wide enough to safely back trucks to the end of the dike to extend the dike into the permanent marsh. After diversion, the flow, reaching the Oltukai area may be markedly reduced..

IV. A phase II water training effort may be necessary to reduce the existing pattern of Longinye spring outflow concentrating in the Oltukai areas. If the phase I effort is only partly successful, a channel may be necessary to be constructed about 4Km. slightly West of North of the spring area to carry the Longinye spring waters further north toward Ndundus causeway. Several hundred metres of channel may be necessary.

V. Due to the extremely flat terrain in the north and east of the 01 Tukai lodge area, a GPS survey of terrain should be considered. Such GPS surveys are extremely accurate in terms of both location and elevation, and are rapid to perform. The GPS system necessary is that which has a few centimetres of evaluation accuracy. Until such a survey can be performed, or as an alternative to the GPS survey, careful observations and mapping of standing bodies of water should be made (documented photographically from the air as the bodies of water expand and retract) and also monitored by placing monuments along the slightly low areas to record the patterns of ephemeral flooding on the ground to supplement the photography. Such monitoring will greatly aid management decisions as to where raised roads should be constructed, where culverts should be placed and where raised, roads may be used to train the pattern of ephemeral water flow.

VI. Raised roadways (as currently being constructed) have the potential to serve a dual purpose of controlling the extent of flooding and providing access. This approach, where possible, will prove to be the most cost effect. the objective should be to keep the ephemeral flooding away from the infrastructure investments, and to ensure that habitat is protected from unacceptable salinity changes. A certain amount of "trial and error" experience will be required because of the extreme flatness of terrain and the presently limited database on water quality and rates of salinity changes. The water training program must be carefully integrated with water-quality monitoring. The water-quality monitoring must also be integrated with habitat monitoring or surveys to establish what levels of water salinity are acceptable or desirable for the diversity in the ecosystems that have developed. For example, the heavy water fowl and shore bird concentrations may be impacted by changing levels of salinity. It may become desirable to freshen Lake Leakey, and it may be necessary to design water control to accomplish this periodically.

VII. Under current hydrologic conditions, salinity of both ground water and surface water will generally increase from lower salinities in the south to higher salinities to the ' north due to the overall evapotranspiration/evaporation and flushing action of northward flow. This is why both water-table and salinity monitoring of both ground water and surface-water is desirable. With the development of shallow ground water levels, capillarity will place more stored salt into surface water bodies during periodic flooding, and the northern surface-water bodies may tend to become more saline due to the concentrating effect of evaporation and transpiration. The 01 Kenya swamp and northern extent are behaving as a closed hydrologic system, and therefore the salts are not flushed from the area. if the vigorous hydrologic state of this system continues to persist, it may.. prove desirable from a habitat maintenance perspective to establish an overflow channel! to the Amboseli Lake basin to the north. At least two natural overflow channels exist, and can be seen on airphotos and the 01 tukai 1 :50,000 topographic map. It may prove desirable, over a prolonged period at the present water supply pattern, to artificially re- establish periodic overflow to one of these channels by trenching. This may help freshen the system by providing a way for some of the salt to leave the system.

VIII. Regional water-level monitoring with boreholes is recommended. Such a program may provide a better early warning of changing ground-water supply than the precipitation records. Existing boreholes should be inventoried and those that can be measured for water level incorporated into a biannual survey (once near the end of the dry season, and once about one month after the long rains). The Amboseli Park ground-water monitoring will require the installation of generally shallow piezometers (4" PV C casing to 5m below the encountered water level). This should be accompanied by a comprehensive hydrogeologic study of the initially established data of water quality and spring discharge, as well as surface-water salinity to establish baseline conditions. Such a study should be accomplished over a three year period to establish patterns of water level, spring discharge, and seasonal salinity. The hydrogeologic study would make a good PhD thesis, if sufficient funds are available for the installation of the shallow piezometers and water quality analyses. Such a comprehensive study is complex and should be supervised by an experienced research hydrogeologist and research ecologist to get the most out of it. I believe the results would reduce much of the uncertainty in future water and habitat management decisions for the Park.

XI: If my conclusions with respect to the fever tree mortalities are correct A. xanthophloea is basically a moderately salt tolerant phreatophyte that grows best in areas with a water table at about 10 metres) it may be possible to re-establish groves where the water table is between 5 and 10m in depth. There is such a water table depth zone along the northern edge of the piedmont slope, between the woodlands of Acacia xanthophloea and the marsh areas. Generation experiments might be attempted in areas where the water table is between 5m and 10m of depth. This probably should occur after the ground-water monitoring network is established in the Park.