| dc.creator | Serrano Amaya, Doris Helena | |
| dc.date | 2021-05-03T21:25:39Z | |
| dc.date | 2021-05-03T21:25:39Z | |
| dc.date | 2020 | |
| dc.date.accessioned | 2023-08-28T13:34:01Z | |
| dc.date.available | 2023-08-28T13:34:01Z | |
| dc.identifier | https://repository.ut.edu.co/handle/001/3263 | |
| dc.identifier.uri | https://repositorioslatinoamericanos.uchile.cl/handle/2250/8440266 | |
| dc.description | 206 p. Recurso Electr?nico | |
| dc.description | Esta investigaci?n tuvo por objetivo, estimar y modelar la humedad superficial del suelo (HS) por medici?n indirecta, e incorporar tecnolog?a satelital con informaci?n geo-ambiental. Para ello, se plante? un m?todo indirecto, un m?todo geom?trico y un m?todo por asimilaci?n de datos. El estudio se llev? a cabo en 10 unidades de muestreo localizadas en la cuenca hidrogr?fica del r?o Quind?o (Colombia). Se realizaron 2211 mediciones gravim?tricas en la capa del suelo (0-5 cm), durante siete per?odos de evaluaci?n, entre 2017 y 2018.
En las unidades de muestreo, se evaluaron las siguientes variables edafol?gicas: densidad aparente por el m?todo del cilindro, distribuci?n de tama?o de part?culas por el m?todo del hidr?metro, porosidad del suelo por el m?todo de la mesa de tensi?n, curvas de retenci?n de humedad del suelo, en seis puntos de presi?n utilizando las ollas de Richards y se midi? la humedad relativa y la temperatura del suelo, mediante 47 sensores iButton DS1923. El an?lisis estad?stico se realiz? a trav?s de un enfoque descriptivo e inferencial usando m?todos multivariantes:
Para el m?todo indirecto, se agregaron las mediciones gravim?tricas usando el promedio heroniano, con el cual se obtuvo igual dimensi?n de datos por las dos metodolog?as. Se realiz? un ajuste mediante la t?cnica de splines c?bicos. El efecto del tiempo se analiz? con el an?lisis de varianza no param?trico longitudinal. Se realiz? un an?lisis de conglomerados, conform?ndose cinco estratos, que fueron evaluados con el estad?stico Q. Finalmente, se determin? un modelo no lineal (curva S), entre las mediciones y las estimaciones, con un ajuste R2 de 68%.
Para el m?todo geom?trico, se conformaron los contornos convexos y se determin? la HS de im?genes satelitales SMAP de cuatro resoluciones espaciales. Las humedades por los dos m?todos, fueron ponderadas usando por analog?a un l?mite trascendental, en el que se consideraron seis par?metros geom?tricos. Se realiz? la clasificaci?n de las humedades con una red neuronal probabil?stica Bayesiana y se encontr? que la tasa de cambio de ?rea, present? mayor cantidad de casos clasificados correctamente, con un R2 de 74%, para 1 km de resoluci?n espacial. Se estim? una unidad espacial de cambio (downscaling), de 0,485 km, correspondiente a una reducci?n del 53%.
Para el m?todo por asimilaci?n de datos, se dise?? un modelo de regresi?n lineal para predecir la HS con la incorporaci?n de factores geo-ambientales y humedades del sat?lite SMAP. Se determinaron coeficientes R2 de 62%, 69%, 63% y 70%, para los modelos de predicci?n de 1,3, 9 y 36 km respectivamente. Se evalu? la exactitud de los promedios heronianos con las estimaciones, encontr?ndose un RMSE menor de 0.0614, para 9 km de resoluci?n espacial.
Los resultados de esta investigaci?n constituyen informaci?n in?dita en el pa?s, encaminada al monitoreo de la humedad superficial del suelo, mediante el uso de las nuevas tecnolog?as geoespaciales, para mejorar, entre otros aspectos, el monitoreo y gesti?n del riesgo de desastres como las sequ?as e inundaciones, la producci?n agr?cola, y en general la planificaci?n y uso de los recursos en las cuencas hidrogr?ficas.
Palabras claves: Humedad del suelo, potencial matricial, im?genes satelitales, downscaling geom?trico, asimilaci?n de datos. | |
| dc.description | The objective of this research was to estimate and model the surface moisture of the soil (SM) by indirect measurement, and to incorporate satellite technology with geo-environmental information. For this achievement, an indirect method, a geometric method and a method by data assimilation were proposed. The study was carried out in 10 sampling units located in the Quind?o River watershed (Colombia). A total of 2211 gravimetric measurements were made in the soil layer (0-5 cm), during seven evaluation periods, between 2017 and 2018.
In the sampling units, the following soil variables were evaluated: bulk density by the cylinder method, particle size distribution by the hydrometer method, soil porosity by the stress table method, soil moisture retention curves, at six pressure points using Richards' pans, and relative moisture and soil temperature were measured, using 47 iButton DS1923 sensors. The statistical analysis was performed through a descriptive and inferential approach using multivariate methods:
For the indirect method, gravimetric measurements were added using the heronian average, with which the same dimension of data was obtained by the two methodologies. An adjustment was made using the cubic spline technique. The effect of time was analyzed with the analysis of longitudinal nonparametric variance. A cluster analysis was performed, forming five strata, which were evaluated with the Q statistic. Finally, a non-linear model (S curve) was determined, between measurements and estimates, with an R2 adjustment of 68%.
For the geometric method, convex contours were formed and the SM of SMAP satellite images of four spatial resolutions was determined. The moisture content for both methods was weighted using a transcendental limit by analogy, in which six geometric parameters were considered. The moistures were classified using a Bayesian probabilistic neural network and it was found that the rate of change of area, presented a greater number of correctly classified cases, with an R2 of 74%, for 1 km of spatial resolution. A spatial unit of change (downscaling) of 0.485 km was estimated, corresponding to a reduction of 53%.
For the data assimilation method, a linear regression model was designed to predict SM with the incorporation of geo-environmental factors and SMAP satellite moisture. R2 coefficients of 62%, 69%, 63% and 70% were determined for the prediction models of 1.3, 9 and 36 km respectively. The accuracy of the heronian averages was evaluated with the estimates, finding an RMSE of less than 0.0614, for 9 km of spatial resolution.
The results of this research constitute unprecedented information in the country, aimed at monitoring soil surface moisture, through the use of new geospatial technologies, to improve, among other aspects, the monitoring and management of disaster risks such as droughts and floods, agricultural production, and in general the planning and use of resources in watersheds.
Keywords: Soil moisture, matrix potential, satellite images, geometric downscaling, data assimilation. | |
| dc.format | application/pdf | |
| dc.format | application/pdf | |
| dc.language | spa | |
| dc.publisher | Ibagu? : Universidad del Tolima, 2020 | |
| dc.publisher | (CO COL 170) | |
| dc.rights | Atribuci?n-NoComercial-SinDerivadas 4.0 Internacional (CC BY-NC-ND 4.0) | |
| dc.rights | info:eu-repo/semantics/openAccess | |
| dc.rights | https://creativecommons.org/licenses/by-nc-nd/4.0/legalcode | |
| dc.subject | humedad | |
| dc.subject | suelos | |
| dc.subject | sat?lites | |
| dc.title | Estimaci?n de la humedad superficial del suelo por medici?n indirecta y enfoque geom?trico y estad?stico con im?genes satelitales en una cuenca hidrogr?fica andina tropical colombiana | |
| dc.type | Trabajo de grado - Doctorado | |
| dc.type | Text | |
| dc.type | info:eu-repo/semantics/doctoralThesis | |
| dc.type | info:eu-repo/semantics/submittedVersion | |
| dc.coverage | (Ibagu? - Tolima - Colombia) | |
| dc.coverage | (Mundo, Suram?rica, Colombia, Tolima) [1023837] | |
