Semidistributed hydrologic modeling using remotely sensed data and GIS.

摘要

A semi-distributed, physically based hydrologic model (called DPHM-RS) is designed to take advantage of distributed hydrologic information retrieved from various space platform and topographic information processed from Digital Terrain Elevation Data (DTED). DPHM-RS was applied to the Paddle River Basin of central Alberta which was characterized by 5 sub-basins with each sub-basin having its own land cover types and terrain features. Input data to the model included meteorological data collected from 2 meteorological towers set up at the study site, field soil moisture data, topographic information derived from DTED, and distributed hydrologic information retrieved from NOAA-AVHRR, Landsat-TM, and Radarsat SAR data.; DPHM-RS was calibrated with the data of summer, 1996 and validated with data of summer, 1997 and 1998. Excellent agreements between simulated and observed runoff at the basin outlet, energy fluxes and surface temperature demonstrated that DPHM-RS is capable of modeling basin-scale hydrologic processes. This is further confirmed by logical differences in the actual evapotranspiration (ET) simulated for different land covers and by sensible temporal variations of soil moisture simulated for each sub-basin.; Given that in many aspects the performance of DPHM-RS is creditable, the ET component is used, the two-source model, to assess two popular ET models, the Penman-Menteith equation and the modified Penman equation of Granger and Gray (1989) for non-saturated surface. Based on the ET simulated for several land use classes, it seems that the closed canopy assumption of Penman-Menteith is applicable to coniferous forest and agricultural lands but not to mixed forest and pasturelands of the Canadian Prairies. The modified Penman model is generally applicable under dry environment but could estimate ET that is biased under cloudy, rainy days and wet environment.; From 6 scenes of Radarsat SAR images acquired for the Paddle River Basin, and 1350 soil moisture samples collected in the same days from 9 selected sites of the Basin, we demonstrated the feasibility of retrieving near-surface soil moisture from Radarsat SAR images using a linear regression and the theoretical integration equation model (IEM) of Fung et al. (1992). From these data, we also found that for a single land use, the relationship between the cross-correlation of soil moisture and inter-site distance breaks down at a distance of about 250 m.