Hydrological modelling and evapotranspiration estimation based on remote sensing data. A case study in the doñana region and surrounding irrigated areas

Resumen

Water scarcity is rapidly increasing in many regions, and is expected to worsen in the coming decades. In semiarid regions, which extend over a 15% of the global land area, water is an important limitation for vegetation growth and a cause of concern for the long-term sustainability of its associated ecosystems, particularly where natural ecosystems and crops compete for the same water sources. Despite the importance of water scarcity in these regions, data availability is often limited and hydrological studies must still be improved. In order to inform water management decisions, it is crucial to assess the water balance components and particularly the evapotranspiration (ET), a crucial component in both the Earth water and energy balances. In the present thesis, hydrological modelling and remote sensing (RS)-based approaches are assessed and compared to investigate the potentiality to improve the estimation of the components of the water balance, particularly the ET. The study is accomplished in Spain, specifically in a semiarid region subject to the application of water management actions: the World Heritage protected UNESCO Doñana region and surrounding irrigated areas, including the BXII Irrigation District (BXII ID) situated where the old Guadalquivir marshes were located. The specific objectives were: (1) to develop a hydrological model, WATEN, aiming to assess the water balance components, (2) to deepen on the ET estimation through the reformulation of a RS-based ET model, the PT-JPL-thermal, (3) to compare the ET derived from WATEN, the PT-JPL-thermal model and the ET derived from the globally available RS-based ET product MOD16, and (4) to identify hot-spots of drying trends and pluri-annual periods of increasing or decreasing water use based on the spatio-temporal ET dynamics. In the first study of this work, the WATEN model was developed and implemented in the BXII ID, allowing the quantification of the water balance components from 2003 to 2012, in particular ET, drainage and soil moisture content. The model uses an alternative calibration method based on energy consumption data at the drainage pumping stations, which is an interesting approach for areas where access to streamflow information is not available. ET is considered linearly reduced when the water content falls below the readily available moisture in the soil and resulting in estimates about 18% below potential values, reaching up to a 40% reduction in August. The second study has focused on assessing ET patterns in the region by means of three methods: i) the locally calibrated hydrological model WATEN; ii) the RS-based PT-JPL-thermal model modified in this work and iii) the globally available RS-based ET product MOD16. In the mixed-irrigation BXII, the PT-JPL-thermal showed a reasonable agreement with the WATEN ET estimates, both in amount and dynamics. PT-JPL-thermal estimates were a 7% higher than those from WATEN and the correlation coefficient values were ρ = 0.78 and ρ1month-lag = 0.9. On the other hand, MOD16 underestimated ET by 40% and the correlation coefficient with WATEN was only ρ = 0.48. The PT-JPL-thermal and the MOD16 ET estimates were extended to the whole Doñana region and surrounding irrigated areas including rice fields and natural ecosystems, such as wetlands, shrublands and forests. For all the land cover classes, the PT-JPL-thermal ET estimations approached the documented ET rates in the region (both natural ecosystems and irrigated areas), whereas MOD16 underestimated ET in all the cases although it was able to identify ET spatial patterns and interannual variability. This study has shown that the re-formulation of the outgoing longwave radiation, together with the use of the thermal inertia concept as a proxy to soil moisture content is a more appropriate approach for assessing ET in semiarid environments than the use of vapor pressure deficit. This variable (used by MOD16) may reflect the surrounding arid conditions but fail to consider the real values of soil water availability. In addition, the PT-JPL-thermal only requires air temperature and incoming solar radiation, apart from standard satellites-products freely available, which makes it a model that can be easily applied in areas with little availability of information. In the third study, the PT-JPL-thermal model was used to assess the pressures on natural ecosystems and agriculture in the Doñana region, where natural ecosystems and crops compete for the same water sources, over the last 14 years, a competition that has increased as the region dries. The ET was found to be correlated with precipitation in all the ecosystems, especially in the wetlands and to a lesser extent in the shrublands, forests and croplands, suggesting a more pronounced susceptibility of the wetlands to overall drying. The analysis of the ET trends allowed to distinguish hot-spots, that is, areas where the water used by natural ecosystems was declining or increasing throughout the study period. Moreover, our results showed the significant increase of ET in some areas of intensive agriculture, which points to a greater diversion of water to the crops when the water supply has been dwindling. These results are relevant in terms of providing useful information for water management, helping in decision-making for the balance between regional demands for agriculture and environmental sustainability, and providing relevant information in the context of climatic changes.