Predicting daily soil temperature and available soil water capacity through topographical analysis at various scales using climatic and satellite data.

摘要

This dissertation research is divided into 2 chapters. The first uses daily air temperature and precipitation data obtained from the National Climatic Data Center to predict daily soil temperature at 10-cm depth for 7 model development sites cross the U.S.A. The model was tested temporally and spatially. Frequency analyses for 17 of 19 data sets showed that the number of days which were within a {dollar}pm{dollar}3.5{dollar}spcirc{dollar}C range centered on the measured soil data varied from 77-96%. The values of R{dollar}sp2{dollar} between observed and final predicted soil temperatures ranged from 0.85 to 0.96 for all 19 simulations. The results suggest that most sites within the same climatic region could use the same regional equation, although it may not be appropriate to use a single regression equation to predict daily soil temperature from daily air temperature over a continent. The maximum relative error of the estimated annual soil respiration for the 7 model development sites using the predicted daily soil temperature was {dollar}-{dollar}6% assuming different Q{dollar}sb{lcub}10{rcub}{dollar} values. Soil temperature under vegetation cover was also simulated.; The second chapter develops a method to predict available soil water capacity (ASWC) through topographic analysis at local (Seeley-Swan valley, Montana, with pixel resolutions of 100 m and 1 km respectively) and regional (the state of Montana, U.S.A., 1 km) scales. A linear relationship was found between the observed ASWC values and topographic index--{dollar}ell{dollar}n({dollar}alpha{dollar}/tan{dollar}beta{dollar}). Results have shown that frequency distribution patterns of {dollar}ell{dollar}n({dollar}alpha{dollar}/tan{dollar}beta{dollar}) values are scale-independent. Topographical data, with various pixel resolutions available from satellites, may be used as an alternative to soil type and soil depth for estimating ASWC at local and regional scales for hydrological study and ecosystem modeling if a normal distribution of mean ASWC data for a specific area is assumed. Spatial distributions of the predicted ASWC show better match for the change of topography than the observed ASWC. Overall performance accuracies between the predicted and observed images with the classification of 5 cm interval ranged from 52 to 61%. Some of the mismatches may indicate an improvement of existing ASWC data because the topography-based model can reflect a high spatial variability due to a high-resolution input.