Call number: LD2668 .T4 1961 R65

Résumé : Le satellite Soil Moisture and Ocean Salinity (SMOS), lancé en novembre 2009, est le premier satellite en mode passif opérant en bande-L. Cette fréquence est considérée comme optimale pour estimer l’humidité du sol. SMOS est destiné à cartographier l’humidité de la couche 0-5 cm du sol à l’échelle globale, avec une précision attendue inférieure à 0,04 m3/m3, une répétitivité temporelle inférieure à 3 jours et une résolution spatiale d’environ 40 km. L’objectif de cette thèse est de valider l’humidité du sol de SMOS sur des sites agricoles et forestiers situés au Canada, et de contribuer au développement de méthodes de désagrégation de l’humidité du sol estimée par SMOS dans le but d’exploiter ces données dans les études à l’échelle locale telle qu’en agriculture. Les données de la campagne de terrain CanEx-SM10, effectuée sur un site agricole (Kenaston) et un site forestier (BERMS) situés à Saskatchewan, et celles de la campagne SMAPVEX12, effectuée sur un site majoritairement agricole (Winnipeg) situé au Manitoba, sont utilisées. Les données d’humidité du sol de SMOS ont montré une amélioration de la version v.309 à la version v.551. La version 551 des données d’humidité du sol de SMOS se compare mieux aux mesures in situ que les autres versions, aussi bien sur les sites agricoles que sur le site forestier. Sur les sites agricoles, l’humidité du sol de SMOS a montré une bonne corrélation avec les mesures au sol, particulièrement avec la version 551 (R ≥ 0,58, en modes ascendant et descendant), ainsi qu’une certaine sensibilité à la pluviométrie. Néanmoins, SMOS sous-estime l’humidité du sol en général. Cette sous-estimation est moins marquée sur le site de Kenaston en mode descendant (|biais| ≈ 0,03 m3/m3, avec la version v.551). Sur le site forestier, en raison de la densité de la végétation, les algorithmes d’estimation de l’humidité du sol à partir des mesures SMOS ne sont pas encore efficaces, malgré les améliorations apportées dans la version v.551. Par ailleurs, sur le site agricole de Kenaston et le site forestier de BERMS, les données d’humidité du sol de SMOS ont montré, généralement, de meilleures performances par rapport aux produits d’humidité du sol d’AMSR-E/NSIDC, AMSR-E/VUA et ASCAT/SSM. DISaggregation based on Physical And Theoretical scale Change (DISPATCH), un algorithme de désagrégation à base physique, est utilisé pour désagréger à 1 km de résolution spatiale l’humidité du sol de SMOS (40 km de résolution) sur les deux sites agricoles situés à Kenaston et à Winnipeg. DISPATCH est basé sur l’efficacité d’évaporation du sol (SEE) estimée à partir des données optique/ thermique de MODIS, et un modèle linéaire/non-linéaire liant l’efficacité d’évaporation et l’humidité du sol à l’échelle locale. Sur un site présentant une bonne dynamique spatiale et temporelle de l’humidité du sol (le site de Winnipeg au cours de la campagne de terrain SMAPVEX12), les résultats de DISPATCH obtenus avec le modèle linéaire sont légèrement meilleurs (R = 0,81 ; RMSE = 0.05 m3/m3 et pente = 0,52, par rapport aux mesures in situ) comparés aux résultats obtenus avec le modèle non-linéaire (R = 0,72 ; RMSE = 0.06 m3/m3 et pente = 0,61, par rapport aux mesures in situ). La précision de l’humidité du sol dérivée de DISPATCH, en se basant sur les deux modèles linéaire et non linéaire, décroit quand l’humidité du sol à grande échelle croît. Cette étude a montré, également, que DISPATCH peut être généralisé sur des sites particulièrement humides (le site de Kenaston au cours de la campagne de terrain CanEx-SM10). Cependant, en conditions humides, les résultats dérivés avec le modèle non-linéaire (R > 0,70, RMSE = 0,04 m3/m3 et pente ≈ 0,80, par rapport aux valeurs d’humidité du sol dérivées des mesures aéroportées de la température de brillance en bande L) ont montré de meilleures performances comparées à ceux dérivés avec le modèle linéaire (R > 0,73, RMSE = 0,08 m3/m3 et pente > 1.5, par rapport aux valeurs d’humidité du sol dérivées des mesures aéroportées de la température de brillance en bande L). Ceci est dû à une sous-estimation systématique de la limite sèche Tsmax. Par ailleurs, l’humidité du sol désagrégée présente une forte sensibilité à〖 Ts〗_max, particulièrement avec le modèle linéaire. Une approche simple a été proposée pour améliorer l’estimation de〖 Ts〗_max, dans des zones particulièrement humides. Elle a permis de réduire l’impact de l’incertitude sur〖 Ts〗_max dans le processus de désagrégation. Avec 〖 Ts〗_max améliorée, le modèle linaire aboutit à de meilleurs résultats (R > 0,72, RMSE = 0,04 m3/m3 et pente ≈ 0,80, par rapport aux valeurs d’humidité du sol estimées à partir des mesures aéroportées de la température de brillance en bande-L) que le modèle non-linéaire (R > 0,64, RMSE = 0,05 m3/m3 et pente ≈ 0,3, par rapport aux valeurs d’humidité du sol estimées à partir des mesures aéroportées de la température de brillance en bande-L). Basé sur des données optiques/ thermiques de MODIS, DISPATCH n’est pas applicable pour les journées nuageuses. Pour surmonter cette limitation, une nouvelle méthode a été proposée. Elle consiste à combiner DISPATCH avec le schéma de surface Canadian Land Surface Scheme (CLASS). Les données d’humidité du sol à 1 km de résolution dérivées de DISPATCH pour les journées non nuageuses sont utilisées pour calibrer les simulations de CLASS disponibles continuellement aux heures de passage de SMOS. Une approche de calibration basée sur la correction de la pente entre les valeurs d’humidité du sol dérivées de CLASS et les valeurs d’humidité du sol dérivées de DISPATCH (données de référence) a été mise au point. Les résultats montrent que les données d’humidité du sol à 1 km de résolution dérivées de cette nouvelle approche pour les journées nuageuses se comparent bien aux mesures in situ (R = 0,80 ; biais = -0,01 m3/m3 et pente = 0,74). Pour les journées non nuageuses, les valeurs d’humidité du sol dérivées de DISPATCH seul se comparent mieux aux mesures in situ que les valeurs dérivées en combinant DISPATCH à CLASS. / Abstract : The Soil Moisture and Ocean Salinity (SMOS), launched in November 2009, is the first passive microwave satellite operating in L band which is considered as optimal for soil moisture estimation. It is designed to provide global soil moisture maps at 0 – 5 cm layer from soil surface with a targeted accuracy of 0.04 m3 / m3, revisit time of less than 3 days anda spatial resolution of about 40 km. The objective of this thesis is to validate SMOS soil moisture data over agricultural and forested sites located in Canada, and to contribute to the development of SMOS downscaling methods in order to exploit these data in local scale studies such as agriculture. The data used are collected during the CanEX-SM10 field campaign, conducted over an agricultural site (Kenaston) and a forested site (BERMS) located in Saskatchewan, and during SMAPVEX12 field campaign conducted over a mostly agricultural area (Winnipeg) located in Manitoba. SMOS soil moisture data showed an improvement from the processor versions 309 to 551. Version 551 was found to be closer and more correlated to ground measurements over both agricultural and forested sites. For the agricultural sites, SMOS soil moisture showed high correlation coefficient with ground data especially with version 551(R ≥ 0.58, for ascending and descending overpasses), as well as a certain sensitivity to rainfall events. However, the SMOS soil moisture values were underestimated compared with ground measurements. This underestimation is less pronounced for the descending overpass over the Kenaston site (|bias| viii ≈ 0.03 m3/m3, for version v.551). For the forested site, due to the vegetation density, the SMOS soil moisture estimation algorithms were not very efficient despite the improvements brought to version 551. Moreover, over the agricultural site of Kenaston and the forested site of BERMS, SMOS soil moisture data showed, in general, good performances compared to AMSR-E/NSIDC, AMSR-E/VUA and ASCAT/SSM soil moisture products. DISaggregation based on Physical And Theoretical scale Change (DISPATCH), a physically-based downscaling algorithm, was used to downscale at 1-km spatial resolution the SMOS soil moisture estimates (40-km resolution) over the agricultural sites located in Kenaston and Winnipeg. DISPATCH is based on the Soil Evaporative Efficiency (SEE) derived from optical/thermal MODIS data, and a linear/non-linear model linking the Soil Evaporative Efficiency to the near-surface soil moisture at local scale. Over a site with a good spatial and temporal dynamics of soil moisture (such as Winnipeg’s site during the SMAPVEX12 field campaign), slightly better results were obtained with DISPATCH based on the linear model (R = 0.81, RMSE = 0.05 m3 /m3 and slope = 0.52, with respect to ground data) compared to results obtained from the non-linear model (R = 0.72, RMSE = 0.06 m3 /m3 and slope = 0.61, with respect to ground data). The accuracy of the DISPATCH-derived soil moisture, using both linear and non-linear models, decreases when the large-scale soil moisture increases. This study also showed, also, that DISPATCH can be generalized for very wet soil conditions (Kenaston’s site during the CanEX-SM10 field campaign). However, under wet soil conditions, better results were obtained with DISTACH based on the nonlinear (R > 0.70, RMSE = 0.04 m3/m3 and slope ≈ 0.80, with respect to the estimated soil moisture form L-band airborne brightness temperature) compared to results obtained with ix DISPATCH based on the linear model (R > 0.73, RMSE = 0.08 m3/m3 and slope > 1.5, with respect to the estimated soil moisture form L-band airborne brightness temperature). This is due to a systematic underestimation of the dry edge Tsmax. Furthermore, the downscaling results were found to be very sensitive to 𝑇𝑠𝑚𝑎𝑥, particularly with the linear model. A simple approach was proposed to improve the estimation of Tsmax under very wet soil conditions. It allowed reducing the impact of 𝑇𝑠𝑚𝑎𝑥 uncertainty in the disaggregation process. Using the improved Tsmax value, better results were obtained with the linear model (R > 0.72, RMSE = 0.04 m3/m3 and slope ≈ 0.80, with respect to the estimated soil moisture form L-band airborne brightness temperature) compared to the non-linear model (R > 0.64, RMSE = 0.05m3/m3 and slope ≈ 0.3, with respect to the estimated soil moisture form L-band airborne brightness temperature). Based on optical/thermal MODIS data, DISPATCH is not applicable for cloudy days. To overcome this limitation, a new method was proposed. It involves the combination of DISPATCH with the Canadian Land Surface Scheme (CLASS). DISPATCH-derived soil moisture data for cloud-free days are used to calibrate CLASS soil moisture simulations which are continually available at SMOS overpasses times. A calibration approach based on slope correction between the CLASS-derived and DISPATCH-derived (reference data) soil moisture datasets is considered. Results showed that soil moisture values derived from this newly developed method during cloudy days compare well with in situ data (R = 0.80, RMSE = 0.07 m3/m3 and slope = 0.73). For no-cloudy days, DISTATCH-derived soil moisture data are closer to in situ data than those derived when combining DISPATCH with CLASS.

Master of Arts / Department of Geography / Kendra K. McLauchlan / Fossil pollen assemblages are widely used in paleoenvironmental reconstruction of vegetation regimes and climate conditions. The modern analog technique (MAT) is a popular method used for analysis of these fossil pollen assemblages, but a large modern pollen dataset, such as the North American Pollen Database (NAPD), is needed to provide modern comparisons for interpretation of analog/no-analog situations. While many climate types are well represented within the NAPD, the climates of the southern and central Great Plains of North America are poorly represented. In this study, I collected 31 sediment samples containing pollen from these underrepresented climate types across the Great Plains in the U.S.A. Analysis of these 31 pollen assemblages, along with 504 samples classified as “prairie” from the NAPD and 24 pollen samples from the Flint Hills of Kansas, U.S.A. was conducted to determine if the three major prairie types (short grass, mixed grass, and tallgrass prairies) could be delineated from pollen records alone. Two different MAT dissimilarity metrics (Squared Chord Distance and Canberra Distance Metric) were assessed for their ability to delineate among prairie types and Squared Chord Distance (SCD) was found to a be the better prairie type classifier than Canberra Distance Metric (CDM). Receiver Operator Characteristic (ROC) curve analysis was used to assess the ability of each metric to identify similar pollen assemblages. It has been show in previous studies that two genera found in this region – Ambrosia and Artemisia –respond to temperature and moisture availability in different ways. Using the ratio of the proportions of Ambrosia and Artemisia pollen grains in a pollen assemblage it was found that tallgrass prairies are significantly different from the other two prairie types. The Ambrosia/Artemisia ratio is also useful in determining climatic conditions. This ratio provides paleoenvironmental researchers with a simple quantitative tool to quickly assess general climatic conditions and prairie type.

Wetland soils form an integral part of the agricultural hummocky landscape in the Canadian Prairies. These soils sequester carbon and can serve as sources of greenhouse gases. Three distinctly different but contiguous soils Humic Luvic Gleysols (HLG), Eluviated Dark Brown Chernozems (EDBC) and Calcareous Dark Brown Chernozems (CDBC) located in the St. Denis National Wildlife Area (SDNWA) in four wetlands were selected for study with the aim of comparing the carbon (C) and nitrogen (N) mineralization parameters and determining soil-related factors that influence C and N mineralization in these soils. A short-term aerobic incubation study (16 d) was conducted to determine C mineralization. Nitrogen mineralization was examined using two soil N availability indices: nutrient supply rate (NSR) in a short-term incubation study (14 d) and aerobic leaching-incubation in a long-term study (16 wk). A first order model using non-linear least squares regression was fitted to cumulative C and N curves to determine C and N mineralization parameters (C mineralization potential, Co and C mineralization rate constant, kC; N mineralization potential, No and N mineralization rate constant, kN) for each soil type. Mean cumulative C mineralization, Co, mean cumulative N mineralization and No were highest in the surface horizons and decreased with depth in all the soils. The mean cumulative CO2 production values for the surface horizons were > 150 mg CO2-C kg1 soil while the lower horizon values were < 80 mg CO2-C kg1 soil. Surface mean cumulative N mineralization values were between 5 mg N kg1 soil and 10 mg N kg1 soil with the lower horizons being < 5 mg N kg1 soil. The pattern was similar for Co and No in the surface horizons with values ranging from 200 mg CO2-C kg1 soil to > 300 mg CO2-C kg1 soil and from 8 mg N kg1 soil to 28 mg N kg1 soil, respectively. Nutrient supply rate also showed a similar pattern. The clay fraction showed a stronger negative correlation with the C mineralization parameters in the CDBC than in the other two soils. Organic C and N showed a highly significant positive correlation with almost all the mineralization parameters in all the soils. Overall, notwithstanding the differences in pedogenetic characteristics of the three soils, few significant differences were observed when their C and N mineralization assays were compared. The similarity in the biochemical characteristics of the soils suggests that the observed pedogenic differences do not reflect significantly in the C and N mineralization. Although the pedogenic differences are large, the effects of these differences on soil management are not agronomically significant and the soils can be managed together.

Wetland soils form an integral part of the agricultural hummocky landscape in the Canadian Prairies. These soils sequester carbon and can serve as sources of greenhouse gases. Three distinctly different but contiguous soils Humic Luvic Gleysols (HLG), Eluviated Dark Brown Chernozems (EDBC) and Calcareous Dark Brown Chernozems (CDBC) located in the St. Denis National Wildlife Area (SDNWA) in four wetlands were selected for study with the aim of comparing the carbon (C) and nitrogen (N) mineralization parameters and determining soil-related factors that influence C and N mineralization in these soils. A short-term aerobic incubation study (16 d) was conducted to determine C mineralization. Nitrogen mineralization was examined using two soil N availability indices: nutrient supply rate (NSR) in a short-term incubation study (14 d) and aerobic leaching-incubation in a long-term study (16 wk). A first order model using non-linear least squares regression was fitted to cumulative C and N curves to determine C and N mineralization parameters (C mineralization potential, Co and C mineralization rate constant, kC; N mineralization potential, No and N mineralization rate constant, kN) for each soil type. Mean cumulative C mineralization, Co, mean cumulative N mineralization and No were highest in the surface horizons and decreased with depth in all the soils. The mean cumulative CO2 production values for the surface horizons were > 150 mg CO2-C kg1 soil while the lower horizon values were < 80 mg CO2-C kg1 soil. Surface mean cumulative N mineralization values were between 5 mg N kg1 soil and 10 mg N kg1 soil with the lower horizons being < 5 mg N kg1 soil. The pattern was similar for Co and No in the surface horizons with values ranging from 200 mg CO2-C kg1 soil to > 300 mg CO2-C kg1 soil and from 8 mg N kg1 soil to 28 mg N kg1 soil, respectively. Nutrient supply rate also showed a similar pattern. The clay fraction showed a stronger negative correlation with the C mineralization parameters in the CDBC than in the other two soils. Organic C and N showed a highly significant positive correlation with almost all the mineralization parameters in all the soils. Overall, notwithstanding the differences in pedogenetic characteristics of the three soils, few significant differences were observed when their C and N mineralization assays were compared. The similarity in the biochemical characteristics of the soils suggests that the observed pedogenic differences do not reflect significantly in the C and N mineralization. Although the pedogenic differences are large, the effects of these differences on soil management are not agronomically significant and the soils can be managed together.

Reproduction de : Thèse de doctorat : Chimie organique, minérale et industrielle : Poitiers : 2008. / Titre provenant de l'écran-titre. Bibliogr. 239 réf.

Xeric limestone prairies (XLPs) are open, nonforested communities dominatedby native, C4 perennial grasses. In eastern United States, they occur on shallow, rockycalcareous soils in various physiographic provinces from Missouri and Pennsylvaniasouth to Arkansas and Georgia. Floristic, vegetation, and physical environmental datawere collected from 18 XLPs in Kentucky and used in conjunction with data collected inother studies to provide a synthesis for XLP vegetation in eastern United States and acomparison of this vegetation type with limestone cedar glades and deep-soil barrens.XLPs occur on 33 soil series in five orders (Alfisols, Mollisols, Ultisols, Inceptisols, andVertisols) and on limestone, dolomite, and calcareous shale of Cambrian through Tertiarysystems. In Kentucky, XLPs are restricted to the Interior Low Plateaus, where they aremost frequent on the Upper Mississippian Salem Limestone in the KnobstoneEscarpment and Knobs. Three hundred and thirty-five taxa were recorded in the 18 sitessampled in Kentucky, of which 20 (6.0%) are nonnative and 24 (7.2%) state-listed asrare. The majority of the flora is intraneous C3 hemicryptophytes. Thirteen taxa areendemic to XLPs of eastern United States, but none to those in Kentucky. The native, C4perennial grass Schizachyrium scoparium was dominant in 10 of 12 community types inKentucky identified at a scale of 100-m2 and in 21 of 23 studies of XLPs in easternUnited States. The C4 annual grass Sporobolus vaginiflorus had high frequency values inthe majority of sites in Kentucky, where it often was dominant in the most shallow-soilzones. Variability among XLPs in Kentucky and among the floras of XLPs in differentregions is largely attributable to differences in forb species composition. Soil depth is theprimary abiotic source of variability in XLP community types in Kentucky and in manyother regions of eastern United States. Dendrochronological and aerial photographicstudies support the conclusion that the vast majority of XLPs are not primarycommunities. XLPs on the Cambrian Ketona Dolomite in the Ridge and Valley in BibbCounty, Alabama, are perhaps the only sites which represent an edaphic climax.

Thesis (Ph. D.)--University of Wisconsin--Madison, 1955. / Typescript. Vita. eContent provider-neutral record in process. Description based on print version record. Includes bibliographical references (leaves 58-64).

Thesis (M.A.)--University of Wisconsin--Madison, 1948. / Typescript. eContent provider-neutral record in process. Description based on print version record. Includes bibliographical references (leaves [27]).

Thesis (M.S.)--University of Wisconsin, 1951. / eContent provider-neutral record in process. Description based on print version record. Includes bibliographical references.