Freshwater accounts for just 2.5% of the earth’s water resources, and its quality and availability are becoming an issue of global concern in the 21st century. Growing human population, over-exploitation of water sources and pressures of global warming mean that both water quantity and quality are affected. In order to effectively manage water quality there is a need for increased monitoring and predictive modelling of freshwater resources. To address these concerns in New Zealand inland waters, an approach which integrates biological and physical sciences is needed. Remote sensing has the potential to allow this integration and vastly increase the temporal and spatial resolution of current monitoring techniques, which typically involve collecting grab-samples. In a complementary way, lake modelling has the potential to enable more effective management of water resources by testing the effectiveness of a range of possible management scenarios prior to implementation. Together, the combination of remote sensing and modelling data allows for improved model initialisation, calibration and validation, which ultimately aid in understanding of complex lake ecosystem processes. This study investigated the use of remote sensing using empirical and semi-analytical algorithms for the retrieval of chlorophyll a (chl a), tripton, suspended minerals (SM), total suspended sediment (SS) and water surface temperature. It demonstrated the use of spatially resolved statistical techniques for comparing satellite estimated and 3-D simulated water quality and temperature. An automated procedure was developed for retrieval of chl a from Landsat Enhanced Thematic Mapper (ETM+) imagery, using 106 satellite images captured from 1999 to 2011. Radiative transfer-based atmospheric correction was applied to images using the Second Simulation of the Satellite in the Solar Spectrum model (6sv). For the estimation of chl a over a time series of images, the use of symbolic regression resulted in a significant improvement in the precision of chl a hindcasts compared with traditional regression equations. Results from this investigation suggest that remote sensing provides a valuable tool to assess temporal and spatial distributions of chl a. Bio-optical models were applied to quantify the physical processes responsible for the relationship between chl a concentrations and subsurface irradiance reflectance used in regression algorithms, allowing the identification of possible sources of error in chl a estimation. While the symbolic regression model was more accurate than traditional empirical models, it was still susceptible to errors in optically complex waters such as Lake Rotorua, due to the effect of variations of SS and CDOM on reflectance. Atmospheric correction of Landsat 7 ETM+ thermal data was carried out for the purpose of retrieval of lake water surface temperature in Rotorua lakes, and Lake Taupo, North Island, New Zealand. Atmospheric correction was repeated using four sources of atmospheric profile data as input to a radiative transfer model, MODerate resolution atmospheric TRANsmission (MODTRAN) v.3.7. The retrieved water temperatures from 14 images between 2007 and 2009 were validated using a high-frequency temperature sensor deployed from a mid-lake monitoring buoy at the water surface of Lake Rotorua. The most accurate temperature estimation for Lake Rotorua was with radiosonde data as an input into MODTRAN, followed by Moderate Resolution Imaging Spectroradiometer (MODIS) Level 2, Atmospheric Infrared Sounder (AIRS) Level 3, and NASA data. Retrieved surface water temperature was used for assessing spatial heterogeneity of surface water temperature simulated with a three-dimensional (3-D) hydrodynamic model (ELCOM) of Lake Rotoehu, located approximately 20 km east of Lake Rotorua. This comparison demonstrated that simulations reproduced the dominant horizontal variations in surface water temperature in the lake. The transport and mixing of a geothermal inflow and basin-scale circulation patterns were inferred from thermal distributions from satellite and model estimations of surface water temperature and a spatially resolved statistical evaluation was used to validate simulations. This study has demonstrated the potential of accurate satellite-based thermal monitoring to validate water surface temperature simulated by 3-D hydrodynamic models. Semi-analytical and empirical algorithms were derived to determine spatial and temporal variations in SS in Lake Ellesmere, South Island, New Zealand, using MODIS band 1. The semi-analytical model and empirical model had a similar level of precision in SS estimation, however, the semi-analytical model has the advantage of being applicable to different satellite sensors, spatial locations, and SS concentration ranges. The estimations of SS concentration (and estimated SM concentration) from the semi-analytical model were used for a spatially resolved validation of simulations of SM derived from ELCOM-CAEDYM. Visual comparisons were compared with spatially-resolved statistical techniques. The spatial statistics derived from the Map Comparison Kit allowed a non-subjective and quantitative method to rank simulation performance on different dates. The visual and statistical comparison between satellite estimated and model simulated SM showed that the model did not perform well in reproducing both basin-scale and fine-scale spatial variation in SM derived from MODIS satellite imagery. Application of the semi-analytical model to estimate SS over the lifetime of the MODIS sensor will greatly extend its spatial and temporal coverage for historical monitoring purposes, and provide a tool to validate SM simulated by 1-D and 3-D models on a daily basis. A bio-optical model was developed to derive chl a, SS concentrations, and coloured dissolved organic matter /detritus absorption at 443 nm, from MODIS Aqua subsurface remote sensing reflectance of Lake Taupo, a large, deep, oligotrophic lake in North Island, New Zealand. The model was optimised using in situ inherent optical properties (IOPs) from the literature. Images were atmospherically corrected using the radiative transfer model 6sv. Application of the bio-optical model using a single chl a-specific absorption spectrum (a*ϕ(λ)) resulted in low correlation between estimated and observed values. Therefore, two different absorption curves were used, based on the seasonal dominance of phytoplankton phyla with differing absorption properties. The application of this model resulted in reasonable agreement between modelled and in situ chl a concentrations. Highest concentrations were observed during winter when Bacillariophytes (diatoms) dominated the phytoplankton assemblage. On 4 and 5 March 2004 an unusually large turbidity current was observed originating from the Tongariro River inflow in the south-east of the lake. In order to resolve fine details of the plume, empirical relationships were developed between MODIS band 1 reflectance (250 m resolution) and SS estimated from MODIS bio-optical features (1 km resolution) were used estimate SS at 250 m resolution. Complex lake circulation patterns were observed including a large clockwise gyre. With the development of this bio-optical model MODIS can potentially be used to remotely sense water quality in near real time, and the relationship developed for B1 SS allows for resolution of fine-scale features such turbidity currents.