Abstract. GOES satellite-retrieved surface net radiation (gn) and in situ estimates of surface sensible-latent heat fluxes from eddy correlation measurements are used to test for energy balance closure at the top of the boreal forest. The tests are carried out at five tower flux (TF) sites within the Boreal Ecosystem-Atmosphere Study (BOREAS) experimental area for June-September 1996.The main result is that measured eddy correlation fluxes appear biased low. Surface radiation budget algorithms for directional shortwave/longwave fluxes are used with GOES 8 imager measurements to obtain Rn. The Rn retrievals are made for different types of sky conditions and validated against in situ measurements obtained at the 10 BOREAS mesonet slations. The analysis indicates an overall bias uncertainty in the satellite-retrieved R n of approximately +7.5%, a value large enough that it could affect our interpretation of the energy balance closure analysis. However, in considering the sign of the all-sky bias, our conclusion that the eddy correlation fluxes are biased low may be understated. Results of the closure analysis are in near agreement to results from a set of seven independent energy balance closure studies conducted at individual

No abstract

Abstract. A published physical algorithm designed for estimating total solar and photosynthetically active radiation (PAR) fluxes at the Earth's surface from GOES visible imagery has been modified for the Boreal Ecosystem-Atmosphere Study (BOREAS) applications at high space and time resolutions (1 km/half-hourly). Substantive changes to the algorithm are described, along with descriptions of various additional features needed to apply the algorithm over the boreal forest. Because of the propensity of forest fire smoke to impact the BOREAS study area during the summer period, particular attention has been given to the treatment of aerosol effects. To validate the algorithm, instantaneous estimates of downwelling total solar and PAR fluxes at half-hourly time steps obtained from a GOES 7 data set are compared with 15-min averaged in situ radiometer measurements obtained during the second summer intensive field campaign of 1994 from the array of BOREAS automatic meteorological stations. The validation results have been stratified according to sky conditions to help understand the detailed nature of algorithm performance and to identify weaknesses in the current algorithm design. A variety of sensitivity tests have also been conducted to help evaluate the algorithm's strengths and weaknesses. In addition, a large-scale analysis of the retrievals over the five 1994 field campaigns has been carried out to provide background information for modelers on the nature of the solar component of the surface radiation budget across the boreal forest zone. The overall accuracies of the algorithm are 1.6% and 6.5% for total solar and PAR fluxes, with relative precisions of -20% considering all days, including those with extensive cloud cover and/or high concentrations of forest fire smoke. Such precisions are consistent with current published expectations at an hourly timescale. Better precisions of around 7% are found for clear, relatively aerosol-free days, which represents noteworthy algorithm performance in terms of expectations at these timescales. IntroductionThe Boreal Ecosystem-Atmosphere Study (BOREAS), by design, is to understand the exchanges of radiative energy, sensible heat, water, CO2, and additional trace gases between the boreal forest and the atmosphere. Of paramount importance to understanding these exchanges are quantifying the flux of total solar and photosynthetically active radiation (PAR) that is assimilated by the surface [Hall et al., 1992]. This is because the former controls the total energy exchange between the atmosphere and the surface, while the latter regulates the rate of photosynthesis.Earlier studies of the surface radiation budget (SRB) for the BOREAS region may be found in the work of Moats et al. [1994]. Their results (covering the period from 1985 to 1989) contain only the total solar component, obtained from Earth Radiation Budget Experiment (ERBE) satellite measurements and International Satellite Cloud Climatology Project (ISCCP) meteorological satellite retrievals. These results were e...

RESUMEOn analyse la possibilite de deduire le flux net de rayonnement diurne de surface d grande longueur d'onde (L*) d partir d'information sur les flux solaires de surface et les temperatures de surface determines apartir des mesures du satellite GOES. Un algorithme statistique base sur un ensemble de donnees d'entrainement developpe apartir de mesures in situ acquises en 1995, au cours de la phase de terrain de BOREAS, est developpe pour l' ensemble de la region d'etude de BOREAS et valide pour le reseau de stations meteorologiques automatisees AMS (Automated Meteorological Station) du Conseil de recherches de ill Saskatchewan. Apartir des flux solaires de surface deduits des mesures realisees au cours de la seconde phase intensive de mesures de 1994, dans le canal visible du satellite GOES 7 et des temperatures de surface approximees d l'aide des mesures de temperature in situ du reseau AMS, nous avons deduit des flux nets diurnes de surface agrande longueur d'onde, a ill frequence d'une demi-heure et a une echelle spatiale de 8 x 8 km. Une discussion permet d'identifier des sources possibles d'erreur dans l'application de l'algorithme. Lorsque comparees avec des mesures in situ de L *, les erreurs quadratiques moyennes dans les flux deduits peuvent etre aussifaibles que 19 Wm-2 avec des correlations aussi elevees que 0,85 pour chacun des sites AMS. Quoique les erreurs de validation soient significativement plus elevees pour certains sites, nous montrons que ces resultats peuvent provenir d'une mauvaise mise a niveau des radiometres AMS. SUMMARY An investigation ofretrieving daytime surface net longwave radiation flux (L*) from information on surface solar flux and surface temperature determined from GOES satellite measurements is explored. A statistical algorithm based on a training dataset developed from in situ measurements taken during the 1995 BOREAS field phase is developed for the large scale BOREAS study area, and validated at the Saskatchewan Research Council'sAutomated Meteorological Station (AMS) networkfor summer conditions. Thus, at this juncture, the algorithm is only applicable for a snow-free boreal forest ecotone. Using surface solar fluxes retrieved from GOES 7 visible channel measurements obtained during the second intensive field phase of1994, and surface .. temperatures approximated with in situ AMS temperature measurements, we have retrieved daytime surface net longwave fluxes on a halfhourly time scale and an 8 x 8 km spatial scale. A discussion is included identifying possible sources of errors in applying the algorithm. When compared with in situ L * measurements, the rms errors in the retrieved fluxes are as low as 19 W m? with correlations as high as 0.85 at the individual AMS sites. Although validation errors are significantly greater at various of the sites, we show that these results may have resulted from mis-leveling of the AMS radiometers.

variability and intensity of surface and subsurface fluxes of heat and moisture and showed that they exerted a significant influence on mesoscale dynamics over the relatively small (15 x 15 km) FIFE study area.The problem of properly assimilating small-scale fluxes into large-scale models has also received considerable attention. In regional models, a single horizontal grid element can cover between 103 and 10 4 km 2, while in global climate models (GCMs), the grid size can be l0 s km 2, or larger. At these scales the spatial variation in land surface properties can be considerable. Since the physical properties of the land surface and soils exert a significant control over the spatial variability of surface fluxes and since the interaction between the land surface and the fluxes is nonlinear, linking the sub-grid-scale fluxes to larger scales is a difficult problem. Li and Avissat [1994] examined the nonlinear relationship between the spatial variability of land surface characteristics, vegetation physiology, and the resulting area-averaged heat fluxes and showed that considerable errors arise in trying to estimate grid fluxes using plant and surface properties averaged from detailed subgrid-scale inputs. 29,235