Accurate estimates of photosynthetically active radiation (PAR) are critical for the development of realistic models of plant productivity. However, in many areas such as the vast Amazon region of South America, there have been few empirical studies of PAR. Here, we analyzed the relationship between PAR and broadband solar irradiance (R s ) and formulated models to estimate PAR in two experimental sites (pasture and forest) in the Brazilian Amazon. Three different models of increasing complexity were developed based on information from R s (model 1), R s and clearness index (k t ; model 2), and R s , k t , and water vapor pressure (model 3). Estimates of PAR were generated for each season and for the entire year. All models had very high determination coefficients and indices of agreement for both pasture and forest sites. This strongly supports the use of R s and k t to produce robust estimates of PAR. The results obtained by annual models were close than that found by seasonal models, demonstrating that a single annual model is able to estimate PAR, albeit with lower accuracy.

A radiação de onda longa proveniente da atmosfera (Lin) é a componente do balanço de radiação mais difícil de ser medida. Na Amazônia praticamente não existem medidas regulares dessa componente, mesmo sendo uma importante variável no cálculo do balanço de radiação à superfície e muito usada para alimentar modelos climáticos. Tendo em vista a necessidade desses dados, o objetivo do presente trabalho é avaliar o desempenho de sete equações na estimativa da Lin para dias de céu claro em áreas de floresta (Reserva Biológica do Jaru, 10º4'48''S; 61º55'48''W) e de pastagem (Fazenda Nossa Senhora, 10º45'S; 62º21'W) no sudoeste da Amazônia. Medidas de radiação de onda longa atmosférica realizadas no período de junho de 2005 a maio de 2006 foram comparadas com as estimativas. As equações testadas tiveram desempenho satisfatório apenas durante a estação seca. As condições de alta nebulosidade, dominantes na estação chuvosa, restringiram a quantidade de dados utilizados na avaliação das equações. As equações que utilizam informações de temperatura do ar e pressão de vapor d'água para a estimativa da Lin tiveram melhor desempenho em relação às que utilizam apenas a temperatura do ar. As equações de Brutsaert (1975), Idso (1981) e Prata (1996) foram as que apresentaram melhor desempenho, apresentando os maiores índices de concordância, e sendo, portanto, as equações mais indicadas para a estimativa da radiação de onda longa atmosférica no sudoeste da Amazônia.

RESUMO ABSTRACT: NUMBER OF RAINFALL DAyS FORECASTING FOR THE SOUTHERN HALF OF RIO GRANDE DO SUL USING THE SEA SURFACE TEMPERATURE (SST)This study presents a climate forecasting model of the Number of Rainfall Days (NRD) for some meteorological stations in Rio Grande do Sul using the Sea Surface Temperatures (SST) as the predictable variables. Two sets of data were used in this research: the monthly data of NRD, which were obtained from 5 meteorological stations located in the southern half of Rio Grande do Sul, in the period of 1982 until 2005; and the SST data, measured in the same period. This series was divided in two periods: the dependent period is from 1982 to 2002, and it was used to determine the predictable equations and the regression coefficients; and the independent period, which is from 2003 to 2005, and was used to validate the model. The SST data were employed to establish the relations between the variables through the regression analysis. Good results were obtained in the prediction of the NRD for the regions and all the months analyzed. The predictable and observed data had a very similar distribution of the variables. Although there was some predictable values that differed from the observed ones, but these differences were not significant. The higher differences between the foreseen and the observed values occurred in the independent period.

No abstract

No abstract