Commodity prices have been rising at an unprecedented pace over the last years making commodity derivatives more and more popular in many sectors like energy, metals and agricultural products. The quick development of commodity market as well as commodity derivative market results in a continuously uprising demand of accuracy and consistency in commodity derivative modeling and pricing. The specification of commodity modeling is often reduced to an appropriate representation of convenience yield, intrinsic seasonality and mean reversion of commodity price. As a matter of fact, convenience yield can be extracted from forward strip curve and then be added as a drift term into pricing models such as Black Scholes model, local volatility model and stochastic volatility model. Besides those common models, some specific commodity models specially emphasize on the importance of convenience yield, seasonality or mean reversion feature. By giving the stochasticity to convenience yield, Gibson Schwartz model interprets the term structure of convenience yield directly in its model parameters, which makes the model extremely popular amongst researchers and market practitioners in commodity pricing. Gabillon model, in the other hand, focuses on the feature of seasonality and mean reversion, adding a stochastic long term price to correlate spot price. In this thesis, we prove that there is mathematical equivalence relation between Gibson Schwartz model and Gabillon model. Moreover, inspired by the idea of Gyöngy, we show that Gibson Schwartz model and Gabillon model can reduce to one-factor model with explicitly calculated marginal distribution under certain conditions, which contributes to find the analytic formulas for forward and vanilla options. Some of these formulas are new to our knowledge and other formulas confirm with the earlier results of other researchers. Indeed convenience yield, seasonality and mean reversion play a very important role, but for accurate pricing, hedging and risk management, it is also critical to have a good modeling of the dynamics of volatility in commodity markets as this market has very fluctuating volatility dynamics. While the formers (seasonality, mean reversion and convenience yield) have been highly emphasized in the literature on commodity derivatives pricing, the latter (the dynamics of the volatility) has often been forgotten. The family of stochastic volatility model is introduced to strengthen the dynamics of the volatility, capturing the dynamic smile of volatility surface thanks to a stochastic process on volatility itself. It is a very important characteristic for pricing derivatives of long maturity. Stochastic volatility model also corrects the problem of opposite underlying-volatility correlation against market data in many other models by introducing correlation parameter explicitly. The most popular stochastic volatility models include Heston model, Piterbarg model, SABR model, etc. As pointed out by Piterbarg, the need of time-dependent parameters in stochastic volatility models is real and serious. It is because in one hand stochastic volatility models with constant parameters are generally incapable of fitting market prices across option expiries, and in the other hand exotics do not only depend on the distribution of the underlying at the expiry, but on its dynamics through all time. This contradiction implies the necessity of time-dependent parameters. In this thesis, we extend Piterbarg's idea to the whole family of stochastic volatility model, making all the stochastic volatility models having time-dependent parameters and show various formulas for vanilla option price by employing various techniques such as characteristic function, Fourier transform, small error perturbation, parameter averaging, etc.