Estimating one-time-of-day meteorological data from standard daily data as inputs to thermal remote sensing based energy balance models

Air temperature, atmospheric water vapor (e.g. relative humidity or vapor pressure), solar radiation and wind speed are required inputs when remotely sensed thermal observations are used with resistance energy balance models (REBM) to estimate specific time-of-day components of the energy balance at the time of remotely sensed data acquisition. Times of interest include the afternoon, when Advanced Very High Resolution Radiometer (AVHRR) data are acquired, and midmorning, when LANDSAT Thematic Mapper (TM) data are acquired. Using data from sites in Australia and China, it has been shown how: (1) air temperature (T(a)) at these times can be estimated effectively using the model of Parton and Logan [Parton, W.J., Logan, J.A., 1981. Agric. For. Meteorol. 23, 205-216]. This model only use daily maximum (T(x)) and minimum (T(n)) air temperatures as inputs; (2) to calculate relative humidity (h), dew point has often been equated to T(n). An improvement due to Kimball et al.[ Kimball, J.S., Running, S.W., Nemani, R.R., 1997. Agric. For. Meteorol. 85, 87-98] reduces the bias in the consequent estimates of h, but the root mean squared deviation (RMSD) about the mean does not reduce. Here, the estimate is improved by converting h to vapor pressure (e(a)) and linearly interpolating the e(a) between the times of sunrise for consecutive days; (3) shortwave solar radiation (R(s)) can be effectively estimated by modifying the model of Hungerford et al. [Hungerford, R.D., Nemani, R,R., Running, S.W., Coughlin, J.C., 1989. INT-414, USDA Forest Service] and locally calibrating the Bristow and Campbell [Bristow. K.L., Campbell. G.S, 1984. Agric. For. Meteorol. 31, 159-166] model which estimates the daily total atmospheric transmittance as a function of a rain modulated difference between T(x) and T(n). On clear, rain free days, when good quality remotely sensed data are most likely be acquired, the estimates of R(s) using both models were within 13% of observations; (4) for wind speed it was found that the sensitivity to using default assumptions such as locally set average was greater for potential evapotranspiration (λE(p)) than for actual evapotranspiration (λE(a)), which affects the resulting moisture availability (the ratio of λE(a) to λE(p)). The variations in λE(a) and net radiation (R(n)) estimates, introduced by applying these interpolation methods to provide specific time-of-day meteorological inputs to the REBM, were examined for a site in China. When estimating λE(a) at AVHRR times on days when no rain fell, the RMSD/Mean of measurement was 27% when all meteorological values were estimated and remained 27% when all values were known. Estimating R(n) for the same times and conditions the RMSD/Mean of measurement was 15% when all meteorological values were estimated. This value reduced to 9% when all values were known. The results of this investigation provide support for the approach of estimating meteorological variables from daily air temperature extremes and daily rainfall at the times when clear remotely sensed data are acquired.