Diurnal Modeling Of Surface And Soil Temperature Dynamics Using Temporally Discrete Thermal Observations

Earth temperature that includes surface and soil temperatures is an important parameter to climate change studies. The modeling of diurnal land surface and soil temperature dynamics is crucial to investigate the exchange of matter and energy between atmosphere, ground surface and soil beneath the surface. Although great improvements on the modeling of diurnal temperature have been made for decades, improper descriptions of heat variation at surface and its propagation into soil are remained in some prevailing models, resulting in limited expansibility and poor accuracy. To deal with those problems, this paper attempts to develop a generic strategy to model diurnal variations of land surface and soil temperature using temporally discrete observations, theoretically based on the surface energy balance and the heat conduction equations.First, this paper proposed a generic framework (GEM) to model the diurnal land surface temperature (LST) dynamics in clear-sky days. In a single diurnal temperature cycle, we derived seven cases of the GEM, i.e., from the GEM-Ⅰ to -Ⅶ, in which two to twelve controlling parameters are required. These controlling parameters are taken as unknowns in backward inversion using temporally discrete surface temperatures. Validations were performed using in-situ brightness temperatures, and Moderate Resolution Imaging Spectroradiometer (MODIS) and Spinning Enhanced Visible and Infrared Imager (SEVIRI) LSTs. The results show that the accuracy generally increases from the GEM-Ⅰ to -Ⅶ, with the mean absolute error (MAE) decreasing from 1.71 to 0.33℃. Particularly, GEM-Ⅱ and -Ⅲ can be used to normalize the four MODIS LSTs in a diurnal cycle without additional information required; and GEM-Ⅵ and -Ⅶ are useful once high accuracy is required. Further modeling also indicates that GEM is capable of interpolating LSTs under nonstandard cases with arbitrary starting time, duration length, and local latitude. We consider this study could provide practitioners more options within particular applications for modeling the temperature dynamics under diverse requirements of modeling accuracy and controlling parameter number.Second, this paper improved the traditional Fourier heat conduction (T-FHC) method to reconstruct diurnal soil temperature field using two-depth measurements of soil temperature. The revised FHC method employed a’direct approach’ as well as an ’indirect approach’to model the diurnal soil temperature field, during which the annual temperature cycle and the diurnal temperature cycle are combined. Two validation experiments (i.e. Test-1 and Test-2) were performed with soil temperature measurements at five stations chosen from the Soil Climate Analysis Network (SCAN). The Root mean square errors (RMSE) of the T-FHC method range from 1.0 to 2.1℃ (2.4 to 4.8℃) in Test-1 (Test-2); while the errors of the’indirect approach’ are reduced to less than 1.0 ℃ in both Test-1 and Test-2; the errors by the’direct approach’are further reduced to less than 0.7℃ in both Test-1 and Test-2. The results show that the revised FHC method improves the accuracy of modeling diurnal soil temperature field in comparison to its traditional form.In conclusion, theoretical and technical improvements have been achieved by this study towards modeling diurnal dynamics of earth temperature using temporally discrete thermal observations.