Use of remote sensing and GIS techniques in estimating the subsurface water use of lowland irrigated area in the dry season | |
| Author | Suphan Saykawlard |
| Call Number | AIT Diss no.SR-04-01 |
| Subject(s) | Remote sensing Geographic information systems Water use |
| Note | A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Technical Sciences, School of Advanced Teclmologies |
| Publisher | Asian Institute of Technology |
| Series Statement | Dissertation ; no. SR-04-01 |
| Abstract | This study emphasizes on the exploring feas ibility of vegetate index derived from satellite data in simulation the spatial variation in subsurface water use by crop and its level change. The methods of remote sensing technology and Geographic Information System (GIS) were integrated. Total six digital images from Landsat 5 Thematic Mapper of an irrigation project, Thailand, were used. Normalized Difference Vegetation Index (NDVI), the extracted index from satellite images, had been assessed to explore its relationship to ground data in the 1999 dry season. Two relationship models were established first: 1) subsurface water use and age of rice, and 2) age of rice and NDVI. The relationship models are: 1) Subsurface water use (in mm/day) = -0.533Age of rice (in day) + 7.5561 with R2 = 0.72, and 2) NDVI = 0.0026Age of rice - 0.0427 with R2 = 0.72. Then the relationships: 1) and 2) were combined and the relationship model of subsurface water use and NDVI was established. The model is: Subsurface water use (mm/day) = (0.0174 - 0.0533NDVI)/0.0026. NDVI had relationship with subsurface water levels. The relationship model is: Subsurface water level change (m) = 2. 1715NDVI - 0.3 with R2 = 0.77. The spatial models of subsurface water use and level change were developed from the spatial models derived from rice classification based on training areas from NDVI of tlu·ee dates covering the 1999 dry period. Firstly, the three classes of rice development (young, medium, and old) were classified by Maximum Likelihood Method on bands 3-4-5 on 20 March 1999 image, used NDVI from 45 rice fields as training areas. Then the percentage areas of tlu·ee rice classes from classification were used to assign training areas by dividing NDVI histogram into three ranges of rice development. The ranges were: young (-0.2039 to 0.0411 ), medium (0.0411 to 0. 11 62), and old (0.1162 to 0.3545). The other two images on 30 December 1998 and 16 February 1999 were classified using NDVI ranges selected on histograms and adjusted the ranges by crop records to be training areas. The new ranges were: 1) on 30 December 1998 image, -0.5625-0.2574 = young, 0.2574-0.4236 = medium, and 0.4236-0.7023 = old, and 2) on 16 February 1999 image, -0.28 1-0. 11 97 = young, 0. 11 97- 0.2408 = medium, and 0.2408-0.5386 =old. The Maximum Likelihood Classification method was applied on the new training areas using 7 bands. The results were spatial models of three rice development classes in the 1999 dry season. The two relationship models of NDVI and subsurface water use and subsurface water level change were applied to the spatial models of rice development from classification. Then spatial models of subsurface water use and level change were developed. For accurate result in simulation of the subsurface water level change, the . subsurface hydrological characteristic that is sand thickness of the water storage body was integrated with the spatial models of three rice classes. The spatial models resulted from integration in the 1999 dry season were tested by correlation to subsurface water level change. The spatial model on 30 December 1998 had the highest relationship to subsurface water level with R2 = 0.53. Thus, the relationship of satellite data with subsurface water levels was reliable at some period, for example at the beginning of the dry season planting in December. The Maximum Likelihood Classification method using NDVI ranges on histograms as training areas was used in simulating the potential areas of subsurface water change in the 1998 dry season without ground information on age of rice. Then the spatial model resulted from classification were integrated with sand thickness and the spatial models of subsurface water level change were developed. |
| Year | 2004 |
| Corresponding Series Added Entry | Asian Institute of Technology. Dissertation ; no. SR-04-01 |
| Type | Dissertation |
| School | School of Advanced Technologies (SAT) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Space Technology Application and Research (SR) |
| Chairperson(s) | Honda, Kiyoshi; |
| Examination Committee(s) | Gupta, Ashim Das; Apisit Eiumnoh; Chen, Xiaoyong;Herath, Srikantha ; |
| Scholarship Donor(s) | Royal Thai Government (RTG)- partial; |
| Degree | Thesis (Ph.D.) - Asian Institute of Technology, 2004 |