In 1995, a study to measure and model ET in the Everglades was begun as part of the US Geological Survey South Florida Ecosystem Program. The principle objective of the study is to develop an understanding of ET within the Everglades drainage unit, excluding agricultural and brackish environments. To achieve this, a network of eight ET-measurement sites was established, representing the various types of hydrologic and vegetative environments. Continuous measurement of ET at these sites for at least a 2-year period (October 1995 through September 1997) is being used to develop regional models of ET that can be used to estimate ET at other times throughout the Everglades.
The Bowen ratio can be approximated as a function of vertical differences of temperature and vapor pressure in the air. At vegetated sites, air temperature and vapor-pressure measurements are made simultaneously every 30 seconds at two points that are several feet above the land surface and separated vertically by 3 to 5 feet (ft). Because the temperature and vapor-pressure differentials generally are small in comparison to sensor calibration bias, the upper and lower sensors are reversed in position every 15 minutes. This reversal of position makes it possible to eliminate the effect of sensor bias by averaging the differences in the mean measured air temperature and vapor-pressure differentials during two successive 15-minute intervals, and by using the resultant average differentials to compute latent heat transport at 1/2-hour intervals.
At open-water sites with little or no emergent vegetation, the air temperature and vapor-pressure differentials are determined from measurements of water temperature at the water surface and air temperature and vapor pressure at a point 3 to 4 ft above the water surface. The water-surface temperature is measured by using a float-mounted thermocouple and is assumed to represent the air temperature at the water-air interface. The vapor pressure at that point is assumed to be equivalent to 100 percent relative humidity. Because the water-surface to air differences are much greater than differences in the air over similar distances, the effect of air and vapor pressure sensor bias is negligible. Therefore, the sensor exchange mechanism is not required and only one vapor pressure sensor is needed at such sites.
A modified Priestley-Taylor model of ET was calibrated for each site. These individual site models were then combined into two regional models: one applicable to vegetated wet-prairie and sawgrass-marsh sites, and the other applicable to freshwater sloughs and other open areas with little or no emergent vegetation.
The Priestley-Taylor model of evaporation is a semi-empirical model that expresses ET as a function of aerodynamic resistance (a function of wind speed, canopy characteristics, and atmospheric stability) and canopy resistance (a measure of stomatal resistance to vapor transport from plants). In the Priestley-Taylor model, the atmosphere is assumed to be saturated and an empirical term, the Priestley-Taylor coefficient (P-T), is added to account for the fact that the atmosphere does not generally attain saturation.
Priestley-Taylor models were developed for the nine ET sites, in which the P-T coefficient was expressed as a function of incoming solar energy and water level. Only data for 1996-97 that passed screening tests for accuracy were used to develop the site models using equation 5. The screening tests were based on range limits, visual inspection of plotted net radiation, temperature and humidity readings to eliminate periods when sensors were obviously malfunctioning, and on the criteria that flux calculations are inappropriate if the calculated latent heat flux is not in the opposite direction from the observed vapor-pressure gradient. Such a situation would indicate an error in determination of either the energy budget or the vapor-pressure or temperature gradient. Resolution limits for this study are 0.013 degree Celsius for vertical temperature differences and 0.003 kPa for vapor-pressure differences. These screening criteria eliminated about one-half of the available data from model development, mostly because of sensor failure and resolution limits. Most of the data rejected because of resolution limits or flux directions were for night-time hours, when energy inputs, air-temperature gradients, and vapor-pressure gradients are all relatively low. Regression statistics and values for the coefficients were calculated for all nine sites.
Both water level and incoming solar radiation were significant at the 95-percent level in explaining variation in the P-T coefficient at all sites. The sign of the first regression coefficient is positive for all sites, indicating that the P-T coefficient increases as water depth increases. At vegetated sites, the P-T coefficient decreases as incoming solar radiation increases. At open-water sites the P-T coefficient increases as incoming solar radiation increases.
The effect of water depth on the P-T coefficient when the water surface is above the land surface might be related to the presence of dead plant debris on the land surface. The dead plant material that is above the water surface intercepts some of the incoming solar energy, thereby preventing it from heating the water surface and enhancing evaporation. Instead, the dead plant debris is heated, which enhances convective heat transport. During periods of high water when some dead plant debris is submerged, lesser amounts of the debris are exposed to solar heating, and the water surface receives a greater portion of the solar energy than during periods of lower water. As a result, the portion of solar energy that is transformed into latent heat could be directly proportional to the water level, as is indicated by the positive value of the water-level coefficient at all sites. When water level is below land surface, as occurred occasionally at two sites, the P-T coefficient ?is still related directly to water level. This might be because moisture availability at the land surface decreases as the water level declines.
The inverse relation of the P-T coefficient to incoming solar energy at vegetated sites also could be an effect of the non-transpiring dead plant debris. Solar heating of this dead plant debris would be proportional to the quantity of incoming solar energy. This solar heating could result in an increased portion of the available energy being converted into sensible heat, at the expense of latent heat. Incoming solar energy and latent heat transport ?are directly related at two open-water sites.
The presence of some common attributes among the individual Priestley-Taylor models indicates that a generalized form of the model could provide a reasonable estimate of ET at all sites. This indicates that a generalized (regional) model would be appropriate for evaluating ET at other areas in the Everglades with similar hydrologic and vegetation characteristics to the sites modeled in this study.
The relation of the P-T coefficient to water level for solar intensities of 200 and 800 watts per square meter (watts/m 2 ) was plotted for all sites. The plots indicate that, at 200 watts/m 2 , the relations of the P-T coefficient ?to water level are similar but not identical among the sites. At higher solar-energy levels (800 watts/m 2 ), the plots of the P-T coefficient?as a function of water level define two obvious groups: open-water sites and vegetated sites. The large separation between the two site types (open water and vegetated) at the higher energy level indicates that a significant portion of the incoming solar energy at vegetated sites is used in heating plants and plant debris, with a resultant relative increase in sensible heat transport compared to latent heat transport.
A generalized relation of the Priestley ñTaylor coefficient to water level and incoming solar energy was developed for vegetated and open-water sites by using least-squares regression to fit a data set of the coefficients generated by the individual site models. The values of the coefficients were generated over a range of water level from -1 to 2 ft in 0.1-ft intervals and a range of incoming solar radiation from 0 to 1200 watts/m 2 in 100-watts/m 2 intervals.
See reference Regional Evaluation of Evapotranspiration in the Everglades for mathematical formulas and more detailed results.