Water quality trends at an upland site in wales,UK, 1983–1993

Ten years of detailed upland stream and bulk deposition water quality data from Plynlimon, mid-Wales, are examined for trend. A robust statistical test (the seasonal Kendall test) is applied and data are presented graphically. Smoothing techniques are used to highlight the patterns of change which underlie high data scatter. The graphs show long-term cycles within the data which violate the assumptions of common statistical tests for trend. These cycles relate to fluctuations in the weather patterns at Plynlimon. Even though the seasonal Kendall test is significant for some determinands, the evidence from the graphs suggests that many of these ‘trends’ are unlikely to continue. For solutes in rainfall, there is no convincing long-term trend. There is a possible increase in ammonium concentrations, which may indicate an increasing atmospheric source generated by farming activities, but this will require a longer data series for confirmation. Several trace metal concentrations increased significantly part way through the study period, but later returned to the original levels. The bulk precipitation sea salt input has been uneven over the 10-year sampling period, with the highest inputs occurring during the wetter winters. For solutes in streamwaters, there are clear trends in dissolved organic carbon (DOC), iodine and bromide, which increase over time and may be attributed to an increase in organic decomposition in the catchment. Previous studies in Wales have shown similar behaviour for colour, which is related to DOC, but the corresponding changes for bromide and iodine are new. For most other streamwater determinands, any changes are masked by the effects of year to year variations in the quality and quantity of rainfall. For example, zinc and chromium variations parallel the corresponding rainfall quantity variations. The effect of rainfall quality variation is marked for marine-derived elements such as chloride. For sulphate, streamwater variations are inverted relative to chloride. This suggests that dry deposition may vary with weather conditions: high when the wind direction is from the land and low when weather systems are predominantly frontal and laden with sea salts. Alternatively, high sea salt rainfall may be affecting absorption/solubility reactions in the soils. There are four main conclusions. Firstly, there is no indication of changing acid deposition inputs or changing acidity within the runoff, despite a decline in UK sulphur dioxide emissions. Secondly, streamwater DOC has shown an increase over time, but there is no clear corresponding decrease in pH as might be expected from acidification theory. Thirdly, there are cyclical variations in bulk precipitation inputs and in streamwater quality, which mean that trends cannot be established even with 10 years of data. Long-term cycles are likely to exist in other environmental data and extreme care is required for the interpretation of trend, especially if data sets are short. This aspect strongly supports the continuation of long-term monitoring programmes over several decades. Finally, the graphical application strongly enhances data analysis and should be considered an essential component of trend investigation.