Analysis of Groundwater Level Time Series and the Relation to Rainfall and Recharge

Analysis of Groundwater Level Time Series and the Relation to Rainfall and Recharge
Report No. 1323/1/05
March 2005

EXECUTIVE SUMMARY

Introduction

Groundwater levels in South Africa have been recorded routinely and on a national scale by the Department of Water Affairs and Forestry since the mid 1950s. Since that time this task has been expanded significantly and currently a few hundred boreholes are monitored on a regular basis. Knowledge of long-term rainfall variability and the associated response of the country’s groundwater resources are essential for efficient and sustainable groundwater resource and land management. Despite the relatively short period for which water level records are available, these records contain valuable information that can be used beneficially in the development of technologies to assist in the responsible management of our limited groundwater resources.

With this in mind, some South African groundwater level time series information was selected and studied using a new and novel approach based on a recently developed technique to monitor drought conditions. The results are presented in this report.

Aims and objectives

The main aims of the project were three fold:

To collate long-term groundwater level information and investigate new methods that can be used to correlate these long term groundwater level fluctuations with rainfall and climatic cycles;
To determine whether the apparent long-term declining trends observed in South African groundwater level data is indicative of a current and/or longer term groundwater drought cycle(s) or whether other factors are contributing to this trend; and
To document the analytical techniques available and used to analyse relevant geohydrological information to assist planners in the longer term planning of groundwater utilization as part of the envisaged national Integrated Water Resource Management plans.

A methodology to simultaneously analyse groundwater level and rainfall information of southern Africa and correlate this information to drought and other long-term climatic indicators or cycles is proposed in this report. It is envisaged that the further development of the concepts expressed in this report, may eventually lead to the development of indicators that can be used to improve the current groundwater recharge estimations and thereby the long-term management of the important national groundwater resources.

In view of the international research emphasis on climatic change, it was believed to be important to assess the status of knowledge on climatic variability in southern Africa over recent geological times and specifically the last few thousands years. This review highlighted the large variations in climatic conditions that have occurred over time and allows one to consider the climatic changes that have taken place over the last 50 to 100 years, and the changes that are predicted for the next centuries, in perspective.

Brief description of some results

Based on the documented record and numerous studies on the palaeo-climatic conditions in southern Africa, climatic cycles of different length and intensity appear to have been present during the last approximately 200 000 years and more specifically the last few centuries. These can be summarised as:

an approximately 23 000 year cycle of major wet and dry periods during the last approximate 200 000 years;
a possible 400 – 500 year “cold/warm” or “wet/dry” cycle;
an ~120 year cycle;
an ~18 year cycle; and
an ~11 year cycle.

In the documented South African groundwater level information, a number of different water level fluctuation cycles or trends are present. These are classified as

Short term (~1 year) fluctuations in the groundwater level data;
Medium term (~10 years+) cycle in the groundwater level data; and
Longer term trend which may be of a cyclic nature, but due to the short time series, it is not clear whether it may perhaps be part of a longer cyclic period.

Of particular interest are the declining trends, often extending for 10 or more years that are observed in some of the data sets. These do not appear to be associated with a particular geohydrological, geological or climatological region, but are present throughout the country. Another important observation is the rapid response of groundwater levels to periods of excessive rainfall, often by several metres over a period of a few months.

Closely linked to periodicity in water levels, is the concept of “groundwater drought”, which may have important management implications, but has not yet received sufficient attention. Researcher from the United States used the Palmer Drought Index (PDI) and the more recent Standardized Precipitation Index (SPI) to formulate a “Drought Index” that provides information on drought duration, frequency and time scales. Selected groundwater level and monthly rainfall time series from South Africa were analysed by correlating the groundwater level records with the Standarised Precipitation Index (SPI) values derived for South African rainfall data. An example illustrating the correlation between a water level record of more than 40 years from a borehole near De Aar and the associated SPI is given below Figure 1. The correlation between the two datasets is remarkable over the entire record period and both the larger and smaller anomalies (amplitude and duration) correlate well, especially for the first section of the record. Where groundwater levels follow even the small variations in SPI values, indicates that in selected cases, at least some groundwater recharge occurs during almost every year.

In the second example sudden groundwater level changes of around 25m (rapid recharge) correlate well with periods of high SPI values (consistently wet years), while extended periods of declining water level again correlate with extended periods of low SPI values (consistently dry years). However, not all examples analysed during this research project showed similar high correlations. The reasons for this are not apparent.

Figure 1: Groundwater level response at recording station D6N500 (borehole 3024CA00328) on the farm Caroluspoort, De Aar district with the associated SPI using the rainfall record of Rainfall District 54.

Figure 2: Groundwater level response at recording station C3N511 (borehole 2624DC00032) on the farm Commonage, Vryburg district with the associated SPI using the rainfall record of Rainfall District 90.

No evidence was found that a consistent and constantly declining trend in South African groundwater level information is present, at least over the approximately 50 years for which information is available. There are, however, in many of the long-term records periods of extended declining groundwater level. These were found to continue in some instances for up to 20 years. These long declining trends are also often terminated by sometimes abrupt rises or return of the groundwater level to the original reference level, or sometimes even higher. Depending on the intensity and duration of the rainfall event or the period over which above average rainfall is recorded, this return of the groundwater level occurs within months after the onset of the rainfall event. This can be explained by a combination of factors, such as the intensity and duration of the rainfall period, the geohydrological conditions in the vicinity of the recharge area for a particular aquifer of the monitoring borehole, the physical parameters of the unsaturated zone, and others. Further, no indication could be found that the current cycle of climatic change experienced throughout the world and that is predicted to intensify during the next decades, is reflected in any way in the South African groundwater level records. It must also be stressed that the predicted changes in climate for the southern part of Africa, are relatively minor compared to those that are documented in the geological record of the last few million years.

It is proposed that the observed correlations between SPI and groundwater level fluctuations can be used to calculate a generic “recharge index” for a region. Based on the way the SPI values were calculated and reported in this project, the “recharge index” could be representative of a rainfall district, and not necessarily a region of similar geohydrological conditions. When developed further, this approach may result in improved quantification of groundwater recharge on a regional basis in Southern Africa.

From an analysis of the SPI and long-term groundwater level graphs and its possible use in recharge calculations, the following observations were made:

the SPI value has to be at least positive in order to cause recharge. In the context of this report, recharge is considered to occur when water levels are increasing. In many cases it would appear that the SPI value has to be in excess of +2 to have a meaningful impact on groundwater recharge.
During a period of positive SPI values, water levels have to show an increasing trend for a period generally at least equivalent in duration to that of the SPI anomaly, otherwise limited or no recharge will take place.
During periods where SPI values are close to zero or negative, no recharge takes place.
For recharge to occur, the annual rainfall needs to be above the long term MAP.

A methodology has been proposed to calculate the “Recharge Index”. A number of factors are calculated with each being given a weight. The total weight is then a relative indication of the rate of recharge. The method proposed is still very subjective and a great deal of additional work needs to be done to develop this into a reliable way to turn the “Recharge Index” into a realistic representation of effective recharge. The example in Figure 1 above was used to calculate the proposed recharge index for that specific rainfall district. Based on the outcome of the calculations it is suggested that the period between 1973 and 1978 was one of high groundwater recharge, that of 1978 to 1986 one of no or very low recharge, while the period between 1986 and 1992 was again one of high recharge. No attempt was made to express the “recharge index” value into a percentage of MAP or any other more quantitative way of expressing recharge.

Conclusions and recommendations

Some of the main conclusions and recommendations emerging from this research project can be summarised as follows:

A number of different climatic cycles (cold/warm or wet/dry cycles) of different duration and intensity have been present or are postulated to have been present during the last 200 000 years. These range between a cycle of approximately 23 000 years, to one of approximately 11 years cycle of major wet and dry periods during the last approximate 200 000 years. For the purpose of this study, the 18 and 11 year cycles are perhaps those of most significance in assisting the groundwater resource planners.
Although the present investigation has not investigated this issue in any great detail, it is concluded that when viewed from a geological perspective, the short documented record of the climatic conditions, including the groundwater level records, no conclusive statement can be made regarding changing geohydrological conditions as a result of changing hydrological or climatic conditions over say the last 50-100 years. The South African groundwater level records cover a too short time period to make any conclusive statements whether a steady decline in our groundwater levels is present or not.
The Standardized Precipitation Index (SPI) is a relatively new development in drought prediction technology. It has been proven in the United States and elsewhere in the world to be a reliable tool for this purpose. The application of this technique in the study of the impact of drought on groundwater resources as was done in this project, is as far as could be ascertained from the literature, unique. As such there are no well-proven methodologies developed yet according to which geohydrological information should be handled to obtain the most benefit from such a correlation between the two datasets.
This limited study has shown that correlations between SPI values and water level fluctuations are present in South African data sets. The quality or degree of correlation however, varies. Examples are presented where excellent correlations exist, but on the other hand, some examples of poor correlations have also been recorded. The reasons for the presence of correlations, whether these are good or poor, are not well understood and warrant further investigation. It was shown that only water level data that are not impacted upon by external forces should be used when correlating SPI values and water level fluctuations.
It is generally accepted that there is a relation between water level fluctuations and recharge. The numerous techniques developed in the past in this field are a proof of this. The results obtained during this project, have confirmed that the SPI methodology can also make a valuable contribution to this still challenging field of aquifer recharge estimation.
A first attempt in developing a “Recharge Index” derived from SPI data is presented in this report. Although the concepts that have been formulated and included in the methodology may seem to be simplistic, it is believed that the development of such an index has merit and should be investigated further. The method on its own may not yet provide the required quantitative assessment of groundwater recharge, but together with other available methods to calculate recharge, may add the required information to provide more quantitative answers in future.

Some recommendations emerging from this research include:

A more in depth analysis of the nature of groundwater level time series should be done to be able to explain the wide variety of response patterns to climatic conditions observed in the available South African groundwater level records.
The fundamental theory behind the SPI analysis should be investigated and its possible application to geohydrological datasets should be given a theoretical basis.
The development of the “Recharge Index” should be expanded and placed on a sound theoretical base.
Once a sound calculation methodology for the Recharge Index has been developed, it should be calibrated in order to produce meaningful figures representing the actual recharge that contributes to our groundwater resources.
The process of calculating a Recharge Index should be automated and the derived values should be given a reliability rating.
Once a reliable methodology has been developed to calculate recharge, a groundwater recharge map of the country should be compiled based on for example the rainfall districts as defined by the Weather Bureau.

It is further recommended that the Department of Water Affairs and Forestry should consider the establishment of rainfall recording equipment close to the water level monitoring recorders. This would enhance the reliability of the correlation between rainfall and water level fluctuations. It is further also recommended that the Department reconsider any recommendations made to close down any of the existing monitoring programmes. Groundwater level records contain important information on the long-term behaviour of aquifers that have not been studied in any great detail, but could provide valuable information in future that will be beneficial to the managers of the national water resources.