Characterization and Modelization of Surface Net Radiation through Neural Networks

Background

The energy exchanges between the land surface and the atmosphere can be described by the surface energy balance equation given by the algebraic sum of fluxes over the surfaceQ* + Q_H + Q_λE + Q_G = 0 (1) where Q* is net radiation at surface, Q_H is the sensible heat flux, Q_λE is the latent heat flux, a product of the evaporative rate E and the latent heat per unit quantity of water evaporated, λ, and Q_G is the soil heat flux, the rate at which heat is transferred from the surface downward into the soil profile, all in units of W m^-2. In Eq. (1), the fluxes are considered as positive if directed toward the surface and negative in the opposite case (Hillel, 2004) (See Figure 1).

Figure 1.

Typical variation of terms of the surface energy balance for: (a) daytime over land and (b) night-time over land. The arrow size indicates the relative magnitude

In Eq. (1), Q* is also the algebraic sum of net components of shortwave (K*) and longwave (L*) radiation, which can be written asQ* = K* + L*(2) where the symbol * represents net flux. The K* and L* in Eq. (2) can be written asK* = K↓ - K↑(3) andL* = L↓ - L↑(4) finally resultingQ* = K↓ - K↑ + L↓ - L↑(5) where the downward arrows (↓) and upward arrows (↑) indicate incoming and outgoing radiation components respectively, and the energy moving toward the surface is also considered positive and the energy moving away from the surface is considered negative.