The last four months of 2006 and the first two months of 2007 had above average rainfall and appeared to herald the end of the 2003-6 drought. Although the extra 100 mm of rain was modest in absolute terms, winter rain has a disproportionately large impact on water resources as evaporation is low and the bulk of the water can replenish soils and aquifers. Weather conditions changed rapidly, however; March had slightly below average rainfall and then April 2007 had negligible rainfall, followed by an extremely wet May. This Report summarises the changes over the first five months of 2007, and shows that although dramatic at the time, the impact of the dry April was muted given the high amount of recharge over the previous seven months and the subsequent May rainfall.
Soil water contents in 2006 began recovering from September onwards and approached the highest values for the time of year since measurements began in 2003. The increase was particularly marked on the Palaeogene deposits where the soil profile storage had been trending downwards since measurements began in early 2003. The Chalk soils were also approaching their highest observed year-end levels. Very importantly for long-term water resources the groundwater levels which by summer 2006 were at their lowest since records began in 2003, started to recover. This meant that there was replenishment taking place of water stored in the Chalk aquifer which in turn will sustain the local rivers, streams and springs.
This report describes hydrological conditions at sites in the catchments of the rivers Pang (171 km2) and the Lambourn (234 km2) - approximately the area between Swindon and Reading in southern England (see Figure 1). Both catchments are predominantly rural and overlie the Chalk aquifer - the country's most important groundwater supply. The Chalk is generally at or very close to the ground surface, except in the south of the Pang catchment where it is often covered by clays and sands (Palaeogene deposits) that can be up to 40 m thick. The flows in both rivers are primarily sustained by water emerging from the aquifer with the result that the river network has a seasonal fluctuation - expanding during the winter and contracting during the summer.
The Centre for Ecology and Hydrology (CEH) installed instruments in the area in late 2002/early 2003, as part of its core monitoring program, with the result that there is a particularly comprehensive set of measurements of the hydrological cycle in these catchments. In the context of the recent drought in southern England, the W Berkshire area was one of the hardest hit, and so the detailed measurements here provide a good guide as to how the lack of rainfall impacted on the water resources.
Weather records, from 1972 to the present, at Wallingford (10 kilometres to the NE of the Pang catchment) show that the rainfall in the four calendar years of 2003-6 was 90% of the long-term annual average (LTA) of 598 mm. This apparently unexceptional figure masks a great deal of annual and seasonal variation. Individual years' rain varied from +5.8% above LTA in 2004 to -24% in 2005. The critical period for water resources is the winter season (often defined as November to April) when soil water and groundwater reserves are normally replenished. The winters of 2002/3 and 2003/4 were wet (+25% and +20% above the seasonal average respectively) but those of both 2004/5 and 2005/6 were very dry (-37% and -25%) whilst 2006/07 had close to the average (+4%), (see Table 1).
At first sight the winter rainfall of 2006/07 does not appear to be particularly wet, but the values are critically dependent upon the definition of the winter period duration; if instead the six months September 06 to February are used then last winter has 133% of the long term average, an extra 100 mm of rainfall. The wetter conditions in 2006/07 compared with earlier years can be seen in the plot of monthly rainfall (Figure 2). The almost complete absence of rainfall in April (1mm) appeared to mark the end of the recharge season, but this was then followed by the wettest May in the period of record (126mm).
As well as the overall quantity of rain, its timing through the year is also important because of the strong seasonal cycle of evaporation. In this part of Southern England monthly evaporation typically increases from only 0-5mm in December and January, to about 50mm in April, rising to over 80 mm in June and July before falling back to under 50 mm by September. The benefits of summer rains are therefore quickly lost to evaporation, with little or no water passing through the soil to the underlying water table. For meaningful soil profile rewetting and aquifer recharge to occur, rainfall must comfortably exceed these evaporation values - hence the critical importance of the winter rainfall.
Conventionally, water resource and drought studies have compared rainfall with estimates of the potential evaporation. The latter is estimated from two factors: the supply of energy for latent heat (net radiation) and the drying power of the air (moisture content and wind speed). Such meteorological estimates of water loss will be prone to overestimate the actual evaporation during dry summers since soil drying severely limits the amount of water available, and so can exaggerate the further depletion of soil moisture and hence groundwater resources.
To overcome this limitation of conventional drought studies detailed micrometeorological estimates of the actual evaporation were made from direct measurements taken at Sheepdrove Farm from April 2003 onwards. These measurements provide a much truer picture of the processes operating, particularly as soil conditions become progressively drier. Figure 3 shows the changing balance of 'net' rainfall, between measured rain and the actual evaporation at Sheepdrove. This graph clearly identifies the year to year changes in the magnitude and duration of the crucial effective winter recharge: 2003/04 (415mm), 2004/05 (260mm), 2005/06 (230mm) and 2006/07 (510mm). It reveals in a way that was not apparent from rainfall values alone that the effective recharge of 2006/07 was double that of the previous two years, whereas from the 'gross' rainfall totals alone it appeared to be only about 25-35% higher. This significant replenishment may mark the ending of the drought; the large amount of water held in the unsaturated zone will percolate downwards to the water table over the coming months and should ensure higher summer groundwater levels than in the previous years. It should also be noted that a return to average water levels is not necessary to ensure adequate water resources for the coming summer.
Soil moisture has been measured down to 3.8 m at two sites (Figure 1): Sheepdrove Farm has thin soils under grass over Chalk, and Grimsbury Wood has mixed deciduous woodland over Palaeogene deposits (mainly clays with some sands). A seasonal trend is clearly visible at both sites (Figure 4) with the soils drying out during the spring and summer and wetting up during the late autumn and winter. The greater range in values (i.e. magnitude of depletion from winter to summer) at Grimsbury is thought to be primarily due to the higher evaporation from the trees.
Changes in the water content at Sheepdrove Farm were mainly confined to the upper 2 m, below which there was little change as water moved down through the Chalk from the upper layers. At Grimsbury Wood the water content changes extended over the whole of the monitored profile with a strong downward trend below 2 m over most of the period. This greater depth of water extraction reflects both the larger rooting depth of the trees, versus grass, and the nature of the soils. Thus, by September 2006 recharge to groundwater could begin to take place under grass - and had probably already begun under cereals due to the period of bare soil and hence reduced evaporation. By the end of 2006 soil water deficits persisted at depth at Grimsbury Wood and is was only after further months of high rainfall that the profile began to wet to greater depths and total profile water contents rose above previously maximum values.
Water levels are monitored at three boreholes in the Chalk (see map, Figure 1). At Beche Park the Chalk comes to near the surface, at Pikes Row it underlies about 13 m of dominantly sandy material, whilst at Grimsbury Wood the Chalk is below around 33 m of material that is sandy in the lower third but dominantly clay in the upper portion. The regional groundwater gradient at Beche Park is eastward towards the river Thames. At the other two boreholes it is towards the southeast: at Pikes Row, towards the river Pang, and at Grimsbury Wood to the river Kennet. All the wells show a general trend of declining water levels until the end of 2006, and then a strong recovery (see Figure 5).
At the start of the records in January 2003, the water levels throughout the aquifer were probably close to the highest likely to occur, as evidenced in nearby boreholes with longer records, e.g. that at Stonor Park, in the Chilterns near Henley. This was due to the exceptionally heavy rainfall in both the winters of 2001/2 and 2002/3. However, below-average rainfall in the spring of 2003 led to an early end to recharge and the seasonal decline in water levels began earlier than usual. The unusually dry and sunny conditions during summer 2003 meant that the recharge in the following winter recommenced later than normal. The rainfall in winter 2003/4 was above average (see Table 1) and water levels rose from their low values, and the recharge was sufficient to balance the subsequent losses during the summer of 2004. The winter of 2004/5 began with above average rainfall with the result that the outlook for recharge was positive. However, the subsequent winter months had rainfall that was significantly below average with the result that the amount of recharge was well below average and, by the spring of 2005, water levels had not returned to anything like the levels recorded at the same time in the previous year. The winter of 2005/6 proved to be drier than average with a resulting further reduction in recharge and a subsequent limited rise in the water levels. It is interesting and instructive to compare these patterns of groundwater levels with the effective recharge from 2003/04 at Sheepdrove (Figure 3). The centre of mass of the recharge occurs about December, and the groundwater rises to a peak around April or May, reflecting the characteristic slow response times of water movement through the Chalk. The sequence of decreasing winter recharges is mirrored in the declining groundwater peaks and minima, broken by the upsurge following the above average quantity of recharge in 2006/07. The groundwater levels began to rise first at Grimsbury Wood and Pikes Row following a period of rainfall that began in mid-November and levels at Beche Park began to rise later, at the beginning of December probably due to the thick unsaturated zone at this site, and the possibility of localised concentrated recharge from the Paleogene at the other sites.
At the end of 2006 river flows were significantly below average and continued to decline as a consequence of the lack of recharge to groundwater (Figure 6). The logarithmic scale used in the graphs tends to hide just how low the flows became. The small responses to rainfall evident in the Pang record are almost certainly due to limited runoff from the Palaeogene areas in the southern part of the catchment. Flows then increased at all three gauges in response to the winter recharge.
The catchment map is based on work of the Ordnance Survey. The groundwater borehole data were supplied by the Natural Environment Research Council, river flows were provided by the Environment Agency. All other data were produced by the Centre for Ecology and Hydrology (CEH) at Wallingford (www.ceh.ac.uk).
Seasonal average
is the average measured
conditions for the period of the year referred to
(e.g. average rainfall in each of the Februarys
in the period 1972- 2001).
Potential evaporation
is the theoretical
maximum evaporation based on the drying power of
the air (depends on humidity, wind speed and
sunshine). The actual evaporation may be less if
the ground is dry, and requires specialist
measurements that are available at only a few
sites such as used here.