2014-05-15_REVISION - M1987064 - Laserfiche WebLink

Home Browse Search

292 D.C. Goodrich et al. /Agricultural and Forest Meteorology 105 (2000) 281 -309 which to compute the water balance. DOY 101 -191 was selected for the water balance calculation. Dur- ing this period of time there was very little rainfall ( <12.5 mm) and no apparent runoff in response to rainfall in this or upstream portions of the San Pe- dro basin. C/W and mesquite vegetation were thus as- sumed to be transpiring only groundwater from the alluvial or regional aquifer. The isotope analysis pre- sented in Snyder and Williams (2000) is also consis- tent with this assumption. Small changes in surface soil moisture were accounted for in the water balance but they were minor compared to C/W and mesquite ET fluxes (roughly 8% of the fluxes). By utilizing this time period for the water balance, the difficulty of plant water source partitioning between ground and near -term surface water sources is therefore largely avoided. A discussion on the data sources and procedures used to estimate each of the water balance compo- nents and an approximate estimate of their uncertainty for the DOY 101 -191 period follows. Uncertainty estimates were determined by computing the standard error of the function for each water balance compo- nent (Brinker, 1969; Wolf, 1980). For a general func- tion z = f (a, b, c, ... , p), where a, b, c, ... , p are independently observed quantities, and the standard error of the function QZ was computed as az = l �aaa�2 + [�bab]2 + [��c]2 ++ r��P12 L (J7) where Oa, ab, ... , up are the standard errors of each component. Measurement errors for direct measure- ments such as stream stage or vegetation areas were estimated based on knowledge of the measurement devices or methods employed. Standard errors from regression estimates such as stage-discharge ratings or model estimates such as the P -M model for C/W transpiration were employed in the standard error function. For those components requiring spatial scal- ing, the areas of various classes from Lewis Springs to Charleston are contained in Column 2 of Table 1. The inflow and outflow volumes Win and Qout) were obtained from continuous stage measurements and stage-discharge rating curves at Lewis Springs and Charleston. The discharge values in [L3T -1] were then integrated from DOY 101 at 00:00 h to DOY 191 at 00:00 h. The uncertainty for these volumes was esti- mated by assuming that the standard error of estimate in the rating curves at Charleston was equal to that at Lewis Springs (0.005 m3 s -1 or 0.192 ft3 s -1) and the error in stage was 3 mm. The volume of water evaporated from the stream water surface (E,,,S) was computed as follows. For each day of the water balance, the Penman potential evaporation was computed using meteorological data from the Lewis Springs mesquite site (Scott et al., 2000) in millimeters per day. This procedure assumes that the meteorological conditions observed at this location are representative of the entire reach. Ad- ditional meteorological observations at the Escapule Wash sap flow site, approximately 7 km north of the Lewis Springs, were used to evaluate this assumption. Common periods of data collection at the two sites occurred in 1997 for DOY 161.5 - 163.7, 191.7- 193.4, and 219.5- 221.7. For these periods, the difference in the mean air temperature between the two sites was 0.50 °C, with a correlation coefficient of 0.98. Va- por pressure (kPa) was also very similar at the two sites. The mean difference between the two sites was 0.10 kPa, with a correlation coefficient of 0.97. Thus, the mesquite tower measurements at Lewis Springs are fairly representative over the distance between these two sites. Using measured radiation from the open mesquite tower results in a much higher estimate of Penman potential evaporation then in the shaded stream areas. To account for shading of the open water surface by streamside vegetation a factor of 0.6 was applied to the Penman potential evaporation estimates. This quantity was multiplied by the water area in Table 1. It should be noted this area was obtained from remotely sensed data acquired in May of 1996. The surface area of the stream exposed to the atmosphere obviously fluctuates with changes in stream stage. However, the estimate based on the area determined from May 1996 is the best available. Given the lack of measurements of the stream surface area over time, a relatively large uncer- tainty of 40% was assigned to stream surface area and a 20% uncertainty was estimated in the calculation of the Penman potential evaporation. The volume of water added to the control volume in the form of precipitation on the stream water surface