Comparisons between buoy-observed, satellite-derived, and modeled surface shortwave flux over the subtropical North Atlantic during the Subduction Experiment

Two years of surface shortwave flux data, from five buoys in the subtropica l North Atlantic Ocean during the Subduction Experiment, were used to exami ne shortwave absorption in the atmosphere, and its partitioning between the clear and cloudy sky. Robust methods were used to isolate the clear-sky sh ortwave observations so that they could be directly compared to values deri ved using a single-column version of the National Center for Atmospheric Re search Community Climate Model radiation code. The model-derived values agr eed with the observations to within 0.5% mean relative error. Additional an alysis showed that the model-data clear-sky surface shortwave differences s howed no systematic relationship with respect to column water vapor amount. These results indicate that clear-sky absorption of shortwave radiation ap pears to be well modeled by current theory. Model-derived clear-sky surface shortwave values were combined with the observed (all-sky) values to deter mine the surface shortwave cloud forcing. The mean of these series were com bined with 5-year mean Earth Radiation Budget Experiment derived top of the atmosphere (TOA) cloud forcing values to estimate the surface to TOA cloud forcing ratio. The resulting values range between 1.25 and 1.59. These val ues, along with the agreement between modeled and observed clear-sky surfac e shortwave, support the suggestion that our current theoretical radiative transfer models do not properly account for the amount of shortwave energy absorbed by the cloudy atmosphere. Mean values from the 2-year shortwave fl ux time series were compared to mean values from two climatologies derived from bulk parameterizations that utilize ship-based cloud reports. These co mparisons show that the Oberhuber climatology underestimates the surface sh ortwave flux by similar to 20% (similar to 40 W m(-2)), while the Esbensen and Kushnir climatology underestimates the flux by similar to 4% (similar t o 8 W m(-2)). The observed mean values were also compared to five satellite -derived climatologies. These comparisons showed much better and more consi stent agreement, with relative bias errors ranging from about -1 to 69%. Co mparisons to contemporaneous, daily-average satellite derived values show r elatively good agreement as well, with relative biases of the order of 2% ( similar to 3-9 W m(-2)) and root-mean-square differences of similar to 10% (25-30 W m(-2)). Aspects of the role aerosols play in the above results are discussed along with the implications of the above results on the integrit y of open-ocean buoy measurements of surface shortwave flux and the possibi lity of using the techniques developed in this study to remotely monitor th e operating condition of buoy-based shortwave radiometers.