1977±1978 (ref. 23). A re-evaluation of systematic errors between GEOSECS and I8NR data is made by comparing deep-water values to determine the appropriate corrections before making the direct comparison for detecting the anthropogenic carbon signal.To determine the mean systematic difference between GEOSECS and I8NR, we compared samples from deeper than 2,000 m. First, the I8NR stations are organized into 6 groups corresponding to 6 stations occupied during the GEOSECS. Each group covers a range of 58 latitudes, with the centre location representing the re-occupation of GEOSECS stations. The mean concentration is computed for samples taken at 8 isopycnal surfaces and j v ranges from 27:75 6 0:01 to 27:82 6 0:01, which cover all samples from deeper than 2,000 m. The concentration difference along the same isopycnal surface at the same location between the two cruises is then computed. Results of the comparison are given in Table 1.We corrected the natural variations of DIC before making the intercomparison. The total carbon released into the water by the respiration of organic matter is estimated using the Red®eld ratio 12 C:O 2 of 117 6 14:170 6 10. The AOU-corrected DIC (DIC aou ) can be given as DIC aou DIC j 2 0:69 3 AOU, where DIC j is the observed DIC on an isopycnal surface. The choice of this Red®eld ratio is based on newer published results and has an insigni®cant effect on the magnitude of the detected CO 2 signal. If a ratio of 140/172, as derived from Indian Ocean 24 , is used, the signal will change by ,1 mmol kg -1 because the anthropogenic DIC increase is derived by difference of two data sets in which the same ratio is applied.The effect of changes in carbonate dissolution at two different times can be corrected by using Alk data. The DIC corrected for Alk (DIC alk ) is given by DIC alk DIC aou 2 0:5 3 Alk 2 Alk 0 , where Alk is the observed alkalinity for the isopycnal surface, and Alk 0 is the preformed salinity-normalized Alk, which can be calculated by using the empirical equation 25 : Alk 0 2;291 2 2:52t 2 20 0:06t 2 20 2 , where t is the potential temperature at the isopycnal horizon.To eliminate changes caused by the variations of salt, the corrected DIC is normalized to a salinity of 35.0: DIC n DIC alk =S 3 35:0, where DIC n is the salinity-normalized ®nal DIC value used for comparison between two different cruises.For estimating DIC increase from atmospheric CO 2 increase, the following needs to be considered. The potential temperature for waters at isopycnal surfaces between 26.6 and 27.2 ranges from 8 to 12 8C in the equatorial region and 6 to 13 8C in the temperature zone. Taking a mean potential temperature of 10 8C for this upper thermocline water, and a mean surface water alkalinity of 2,290 mmol kg -1 , the DIC at chemical equilibrium with the atmospheric CO 2 can be calculated. The rate of atmospheric CO 2 increase varies from 0.8 p.p.m. yr -1 in the early 1960s to ,1.8 p.p.m. yr -1 in recent years 26,27 . The DIC increase is computed using the rate of atmospheric CO 2 increase...

No abstract

Abstract. Volcanic CO2 emission rate data are sparse despite their potential importance for constraining the role of magma degassing in the biogeochemical cycle of carbon and for assessing volcanic hazards. We used a LI-COR CO2 analyzer to determine volcanic CO2 emission rates by airborne measurements in volcanic plumes at Popocatdpetl volcano on June 7 and 10, 1995. LI-COR sample paths of ---72 m, compared with ---1 km for the analyzer customarily used, together with fast Fourier transforms to remove instrument noise from raw data greatly improve resolution of volcanic CO2 anomalies. Parametric models fit to background CO2 provide a statistical tool for distinguishing volcanic from ambient CO2. Global Positioning System referenced flight traverses provide vastly improved data on the shape, coherence, and spatial distribution of volcanic CO2 in plume cross sections and contrast markedly with previous results based on traverse stacking. The continuous escape of CO2 and SO2 from Popocatdpetl was fundamentally noneruptive and represented quiescent magma degassing from the top of a magma chamber ---5 km deep.The average CO2 emission rate for January-June 1995 is estimated to be at least 6400 t d -l, one of the highest determined for a quiescently degassing volcano, although correction for downwind dispersion effects on volcanic CO2 indicates a higher rate of ---9000 t d -1. Analysis of random errors indicates emission rates have 95% confidence intervals of ---_+20%, with uncertainty contributed mostly by wind speed variance, although the variance of plume cross-sectional areas during traversing is poorly constrained and possibly significant. This report presents several new developments in the study of CO2 in volcanic plumes. These include Global Positioning System (GPS) constrained CO2 data to provide better crosssectional plume geometry, fast Fourier transforms (FFT) to remove high-frequency instrument noise and enhance detection of the volcanic CO2 "signal" despite high atmospheric background CO2, parametric statistical models to fit background CO2 and distinguish volcanic CO2 in the plume from ambient atmospheric CO2, concentration contour maps that display the spatial variability of CO2 in ambient air as well as in the volcanic plume, models of the effects of water vapor on CO2 concentrations in volcanic plumes, and correction of volcanic CO2 emission rates for downward dispersion effects. Procedures and Methods TheoryThe CO2 emission rate of a volcano, E, determined from a volcanic plume (Figure 2), is the product of the plume crosssectional area A normal to the wind direction, the wind speed S, and the density of volcanic CO2 in the plume, p.• Equations (6) and (7) neglect covariance terms, which unfortunately cannot be evaluated from available data. The covariance terms probably include negatNe values of significant magnitude owing to the strong negative correlations expected be•een A and S, as well as be•een A and P•. Equation ( The standard error for A is problematic. We assumed the plume cross-section ar...

No abstract

During late December 2000, the giant volcano Popocatépetl near Mexico City exhibited violent explosions. About 20,000 people evacuated their homes for a week to 10 days before returning. On January 22, 2001, an even larger explosion occurred. This article explores the short‐term hazards that are probable at this volcano.