Soil erosion is a major global environmental problem that is increasing year by year in Turkey. Preventing soil erosion requires political, economic and technical actions / before these actions we must learn properties and behaviors of our soil resources. The aims of this study are to estimate annual soil loss rates of a watershed with integrated models within GIS framework and to map the soil erosion risk for a complex terrain. In this study, annual soil loss rates are estimated using the Universal Soil Loss Equation (USLE) that has been used for five decades all over the world. The main problem in estimating the soil loss rate is determining suitable slope length parameters of USLE for complex terrains in grid based approaches. Different algorithms are evaluated for calculating slope length parameters of the study area namely Kocadere Creek Watershed, which can be considered as a complex terrain. Hickey&amp / #8217 / s algorithm gives more reliable topographic factor values than Mitasova&amp / #8217 / s and Moore&amp / #8217 / s. Satellite image driven cover and management parameter (C) determination is performed by scaling NDVI values to approximate C values by using European Soil Bureau&amp / #8217 / s formula. After the estimation of annual soil loss rates, watershed is mapped into three different erosion risk classes (low, moderate, high) by using two different classification approaches: boolean and fuzzy classifications. Fuzzy classifications are based on (I) only topographic factor and, (II) both topographic and C factors of USLE. By comparing three different classified risk maps, it is found that! in the study area topography dominates erosion process on bare soils and areas having sparse vegetation.

Forest roads and stream crossings can be important sources of sediment in forested watersheds. The purpose of this research was to compare trapped sediment and forestry best management practice (BMP) effectiveness from haul road stream crossing approaches and ditches. The three studies in this dissertation provide a quantitative assessment of sediment production and potential sediment delivery from forest haul roads in the Virginia Piedmont and Ridge and Valley regions. Sediment production rates were measured and modeled to evaluate and compare road and ditch segments near stream crossings with various ranges of road attributes, BMPs, and management objectives. Sediment mass delivered to traps from 37 haul road stream crossing approaches ranged from <0.1 to 2.7 Mg for the one year collection. Collectively, five approaches accounted for 82% of the total sediment mass trapped. Approaches were categorized into Low, Standard, and High road quality rankings according to road attributes. Seventy-one percent (5 of 7) of Low ranked approaches delivered sediment to traps at rates greater than 11.2 Mg ha-1 yr-1. Nearly 90% of Standard or High road quality approaches generated less than 0.1 Mg of sediment over one year. Among approaches with less than 0.1 Mg of trapped sediment, road gradients ranged from 1% to 13%, bare soil ranged from 2% to 94%, and distances to nearest water control structures ranged from 8.2 to 427.0 m. Such a wide spectrum of road attributes with relatively low levels of trapped sediment indicate that contemporary BMPs can mitigate problematic road attributes and reduce erosion and sediment delivery. Three erosion models, USLE-forest, RUSLE2, and WEPP were compared to trapped sediment data from the 37 forest haul road stream crossing approaches in the first study. The second study assessed model performance from five variations of the three erosion models that have been used in previous forest operations research, USLE-roadway, USLE-soil survey, RUSLE2, WEPP-default, and WEPP-modified. The results suggest that these soil erosion models could estimate erosion and sediment delivery within 5 Mg ha-1 yr-1 for most approaches with erosion rates less than 11.2 Mg ha-1 yr-1, while model estimates varied widely for approaches that eroded above 11.2 Mg ha-1 yr-1. Based on the results from the 12 evaluations of model performance, the modified version of WEPP consistently performed better compared to all other model variations tested. However, results from the study suggest that additional field evaluations and improvement of soil erosion models are needed for stream crossings. The soil erosion models evaluated are not an adequate surrogate for informing policy decisions. The third study evaluated sediment control effectiveness of four commonly recommended ditch BMPs on forest haul road stream crossing approaches. Sixty ditch segments near stream crossings were reconstructed and four ditch BMP treatments were tested. Ditch treatments were bare (Bare), grass seed with lime fertilizer (Seed), grass seed with lime fertilizer and erosion control mat (Mat), rock check dams (Dam), and completely rocked (Rock). Mat treatments had significantly lower erosion rates than Bare and Dam, while Rock and Seed produced intermediate levels. Findings of this study suggest Mat, Seed, and Rock ditch BMPs were effective at reducing erosion, but Mat was most effective directly following construction because Mat provided immediate soil protection measures. Any BMPs that reduce bare soil can provide reduction in erosion and even natural site condition, including litterfall and invasive vegetation can provide erosion control. However, ditch BMPs cannot mitigate inadequate water control structures. Overall, forest roads and stream crossings have the potential to be major contributors of sediment in forested watersheds when roads are not designed well or when BMPs are not properly implemented. Forestry BMPs reduce stormwater runoff velocity and volume from forest roads, but can have varying levels of effectiveness due to site-specific conditions. Operational field studies provide valuable information regarding erosion and sediment delivery rates, which helps guide BMP recommendations and subsequently enhances water quality protection. / Ph. D.

In der vorliegenden Arbeit wurden Doppelringinfiltrationsexperimente an teilweise gefro­renen Böden durchgeführt. Diese Experimente wurden anschließend mit den zwei computer­ge­stützten Modellen, Erosion 3D / Winter und COUP, nachgestellt, um die Frage zu beantworten, ob es möglich ist, die Infiltration in teilweise gefrorene Böden vorherzusagen. Die Doppelringinfiltrationsexperimente wurden auf einem ackerbaulich genutzten Lehm­boden mit geringer Lagerungsdichte und Bodenfeuchten im Bereich der Feld­kapa­zität, an der nördlichen Grenze des hydrologischen Untersuchungsgebietes „Schäfertal“ durch­ge­führt. Drei Experimente erfolgten bei teilweise gefrorenen und ein Experiment bei unge­frorenem Boden. Bei diesen Experimenten wurde herausgefunden, dass die Endinfiltrationsrate des gefro­renen Bodens mit 7·10-5 m/s gleich der Endinfiltrationsrate des ungefrorenen Bodens war. Während bei dem Infiltrationsexperiment mit ungefrorenem Boden die Endinfiltrations­rate bereits nach 10 bis 20 min erreicht war, wurden bei den Experimenten mit gefrorenen Böden aufgrund der zusätzlichen Sättigung des kryoturbativen Sekundärporenvolumens mehr Zeit benötigt. Zu den im Boden ablaufenden Prozessen bei Zugabe von Infiltrationswasser (Tem­pe­ratur­veränderung, Gefrier- und Auftauprozesse, Veränderung der Porosität) besteht noch Klärungsbedarf. Der für die Modellierung wichtige Eingabeparameter der Anfangsbodenfeuchte konnte bei winterlichen Bedingungen nicht genau bestimmt werden. Gravimetrische Boden­feuchtebestimmungen liefern aufgrund des Eintrags von zusätzlichen Eis- und Schnee-Wasser zu hohe Werte. TDR- und Watermark-Messungen unterschätzen hingegen die Bodenfeuchten, weil sie nur den Anteil des flüssigen Wassers berücksichtigen. Mit Erosion 3D / Winter konnten die Ergebnisse der Infiltrationsexperimente, unter der Voraussetzung, dass die effektive gesättigte hydraulische Leitfähigkeit des ungefrorenen Bodens exakt bekannt war, sehr gut nachgestellt werden. Eine Modellierung der Infiltration in einen teilweise gefrorenen Boden ist damit, zumindest für den untersuchten Boden und die betrachteten meteorologischen Bedingungen, möglich. Das COUP - Modell lieferte dagegen völlig andere Ergebnisse, weil von einem Ein­frieren des infiltrierten Wassers bei negativen Temperaturen ausgegangen wird. Eine Verbesserung der Infiltrationsbeschreibungen könnte hier wahrscheinlich durch die Vorgabe einer größeren Anzahl von Eingabeparametern, die die natürliche Situation besser repräsentieren als die für die Modellierung verwendeten Daten, erfolgen.