The use of principal component analysis for interpreting ground water hydrographs

Principal component analysis was used to define patterns in water table hydrographs at four small, lake-watershed research sites in the United States. The analysis provided insights into (1) characteristics of ground water recharge in different parts of the watersheds; (2) the effect of seepage from lakes on water table fluctuations; and (3) the effect of differences in geologic properties on water table fluctuations. At two sites where all of the water table wells were completed in permeable deposits, glacial out-wash in Minnesota and dune sand in Nebraska, the patterns of water table fluctuation primarily reflected timing and magnitude of recharge. The water table had more frequent and wider ranges in fluctuations where it was shallow compared with where it was deep. At two sites where the water table wells were completed in sand or till, a glaciated mountain valley in New Hampshire and stagnation moraine in North Dakota, the patterns of water table fluctuations were strongly related to the type of geologic unit in which the wells are completed. Furthermore, at the New Hampshire site, the patterns of water table fluctuations were clearly different for wells completed in sand downgradient of a lake compared with those completed in sandy terraces on a mountainside. The study indicates that principal component analysis would be particularly useful for summarizing large data sets for the purpose of selecting index wells for long-term monitoring, which would greatly reduce the cost of monitoring programs.Principal component analysis was used to define patterns in water table hydrographs at four small, lake-watershed research sites in the United States. The analysis provided insights into (1) characteristics of ground water recharge in different parts of the watersheds; (2) the effect of seepage from lakes on water table fluctuations; and (3) the effect of differences in geologic properties on water table fluctuations. At two sites where all of the water table wells were completed in permeable deposits, glacial outwash in Minnesota and dune sand in Nebraska, the patterns of water table fluctuation primarily reflected timing and magnitude of recharge. The water table had more frequent and wider ranges in fluctuations where it was shallow compared with where it was deep. At two sites where the water table wells were completed in sand or till, a glaciated mountain valley in New Hampshire and stagnation-moraine in North Dakota, the patterns of water table fluctuations were strongly related to the type of geologic unit in which the wells are completed. Furthermore, at the New Hampshire site, the patterns of water table fluctuations were clearly different for wells completed in sand downgradient of a lake compared with those completed in sandy terraces on a mountainside. The study indicates that principal component analysis would be particularly useful for summarizing large data sets for the purpose of selecting index wells for long-term monitoring, which would greatly reduce the cost of monitoring programs.