Record Details
Providing the best turbulent heat flux estimates from eddy correlation and bulk methods using DYNAMO data
ScholarsArchive at Oregon State University
| Field | Value |
|---|---|
| Title | Providing the best turbulent heat flux estimates from eddy correlation and bulk methods using DYNAMO data |
| Names |
Marion, June Raven
(creator) de Szoeke, Simon P. (advisor) |
| Date Issued | 2014-08-19 (iso8601) |
| Note | Graduation date: 2015 |
| Abstract | In this thesis, data collected during the Dynamics of the Madden Julian Oscillation (DYNAMO) field campaign, conducted in the Indian Ocean in Fall of 2011, is used to compute heat fluxes at the air-sea interface by evaluating eddy covariances and bulk aerodynamic formulae. Errors in daily average gridded fluxes computed with the COARE version 3.5 bulk aerodynamic formula are assessed with respect to five independent in situ time series from DYNAMO and the Tropical Ocean-Global Atmosphere Coupled Ocean-Atmosphere Response Experiment (TOGA-COARE) in the Western Pacific (Nov. 1992 - Feb. 1993). Oregon State University (OSU), the NOAA Physical Science Division (PSD), and University of Connecticut (UConn) deployed three nearly collocated covariance flux measurement systems on the R/V Revelle during DYNAMO. Covariance and bulk fluxes are compared among these systems, and the experimental setup and calculation methods used for the OSU system are described. OAFlux and TropFlux are two gridded flux products, both of which use global atmospheric reanalyses and in situ observations to produce estimates of surface heat and momentum flux. An array of 106 buoys deployed in the Pacific, Atlantic, and Indian Oceans provide valuable in situ observations of surface meteorological variables including air and ocean temperatures, wind speed and relative humidity. The buoy data is assimilated into the reanalyses and incorporated into the bias correction strategy employed by TropFlux, and the optimal interpolation weights of OAFlux. Locations not constrained by observations have higher root mean square difference between these two products than those near buoys. In fact, OAFlux and TropFlux can sometimes disagree by 100% of the mean flux when buoy data sources are not nearby. Estimating net air-sea surface heat flux to an accuracy of 10 W/m² requires resolution of diurnal solar warming of the ocean surface, known as the diurnal warm layer (Price, et al., 1986), and diffusive cooling of the viscous sub-layer known as the cool skin (Saunders, 1967). Warm layer and cool skin corrections to the bulk ocean temperature are modeled in the COARE bulk flux algorithm (Fairall et al., 1996). Rigorously calibrated and quality controlled DYNAMO data is used to assess the sensitivity of bulk flux calculations to warm layer and cool skin phenomena. Ignoring both corrections results in positive biases of 1.9 W/m² and 8.7 W/m² for sensible and latent heat respectively, mostly because of systematic overestimates of the SST due to the prominence of the cool skin. Two new techniques for including the effects of the warm layer and cool skin on daily fluxes are presented and tested using DYNAMO observations. Both techniques make use of a simple solar radiation model that distributes the daily average solar radiation in a half cosine over 12 hours of the day at hourly resolution. In the first technique the COARE algorithm computes warm layer and cool skin corrections hourly with the solar radiation model. This reduces the root mean square errors of sensible and latent heat fluxes to 0.7 and 2.9 W/m². This improvement requires more than 10 bulk aerodynamic computations per day, a considerable computational expense when evaluating fluxes for decades of global gridded data. In the second technique, the daily average flux is evaluated beforehand in COARE using the solar radiation model with a range of daily average radiation and wind speed values. The results are sorted by solar radiation and wind speed in a lookup table that specifies an adjustment to the fluxes due to the warm layer and cool skin corrections. The adjustment corrects the daily fluxes computed without the warm layer and cool skin corrections. Using the lookup table corrections reduces the root mean square errors to 0.96 and 3.5 W/m², nearly as well as the more computationally intensive diurnal solar model. The cool skin correction makes the biggest difference to the fluxes. Ignoring it can cause 10% overestimation of sensible and latent heat flux. Though they ignore the cool skin, both OAFlux and TropFlux estimate the R/V Revelle sensible and latent fluxes to within 0.1% of the mean flux. It is possible that fortuitous errors in the gridded products compensate the neglected effect of the cool skin. Diurnal warm layers intermittently warm the surface temperature during the convectively suppressed phase of the MJO. The warmest net (warm layer minus cool skin) daily mean difference between the bulk and interface temperature reaches 1° C, which is significant compared to the sea-air temperature difference on the order of 2° C. Observations of stronger diurnal warm layers in the convectively suppressed phase suggest that systematic intraseasonal modulation of the warm layer affects air-sea interaction in the MJO. |
| Genre | Thesis/Dissertation |
| Access Condition | http://creativecommons.org/licenses/by-nc/3.0/us/ |
| Topic | Flux |
| Identifier | http://hdl.handle.net/1957/52558 |
