Accurate weather forecast information has the potential to improve water resources management, energy, and agriculture. This study evaluates the accuracy of recently developed short-range (0- 18 hour) precipitation forecasts from the High-Resolution Rapid Refresh (HRRR) for selected extreme events over the US using NOAA’s Gauge-corrected Multi-Radar/Multi-Sensor (MRMS- GC) radar-gauge merged rainfall observations. This study also evaluates the accuracy of medium-range (1-15 day) precipitation forecasts from the Global Forecast System (GFS) over transboundary river basins in Africa using NASA’s Integrated Multi-satellitE Retrievals (IMERG) “Final Run” satellite-gauge merged rainfall observations. The assessment of HRRR on an hourly basis shows that there was a good agreement between the forecasted and observed precipitation in terms of temporal variability despite the forecasts tend to overestimate rainfall for hurricanes. Spatially, the forecasts were able to capture the general spatial pattern of hurricane driven events but failed to reproduce the characteristics of frontal storms. With regard to the effect of lead times, the 1-h lead forecasts have often lower accuracy than the other lead-time forecasts, while there was not much systematic difference in accuracy among the 2-h to 18-h lead-time forecasts. The evaluation of GFS reveals that the accuracy of forecasts varies a lot among different regions. GFS forecasts tend to overestimate precipitation in the wet climatic regimes but produce almost unbiased precipitation in dry regions. The GFS forecast accuracy decreases as the lead time increases, but the rate of decrement depends on the region. Aggregating the forecasts at temporal scales (1-day to 15-day) may increase or decrease the performance of GFS forecasts, depending on the region. We recommend exploring methods to increase the performance of short-range and medium-range forecasts, including post-processing techniques products before their application in water resources management.

The role of climatic and anthropogenic drivers in groundwater storage depletion and recovery in California’s Central Valley has been investigated. Specifically, the dissertation addresses three research questions: (1) How does climate change impact the groundwater storage? (2) Can managed aquifer recharge (MAR) mitigate groundwater overdraft? (3) How much of the drought-caused groundwater overdraft in Central Valley has recovered during the post-drought years and what is the role of climate and water management in the fast versus slow post-drought overdraft recovery? In the first part of the dissertation, integrated hydrologic models have been simulated to predict future groundwater storage changes under multiple climate change scenarios and to evaluate the relative contribution of crop water use and surface water inflow to Central Valley regions. It is shown that climate change will accelerate groundwater depletion in the future and an increase in future crop water use will be the dominant cause of future groundwater decline without mitigation measure. In the second part, the impact of large-scale MAR implementation on groundwater overdraft recovery, flood peak, and low flow have been investigated via numerical experiments. It is shown that MAR has limited capacity to recovery historical groundwater overdraft due to lack of surface water availability in the southern Central Valley (i.e., San Joaquin and Tulare regions). Delivering excess surface water from the delta to the Tulare and San Joaquin region can significantly solve the groundwater overdraft problem. Moreover, MAR can reduce flood peaks, and increase dry season flow. Finally, an ensemble of groundwater storage change estimates has been made using multiple methods and numerical experiments conducted to understand the role of climate and water management to recover drought-caused groundwater overdraft during post-drought years. The result shows that the Central Valley aquifer is not resilient to drought under existing conditions, and it is very challenging to recover drought-caused groundwater overdraft. However, water management measures that restrict groundwater extraction can significantly reduce the groundwater overdraft recovery time.

The accurate estimation of saturated hydraulic conductivity (Ks) is of high relevance to correctly reproduce water movement in hydrological simulations. Yet, estimation of Ks is challenging particularly in semi-arid regions with particular soil surface characteristics like crusting and sealing. This study presents results of a field campaign in the semi-arid Walnut Gulch Experimental Watershed, Arizona (US), where surface and subsurface Ks measurements were undertaken across the watershed. Results reveal that the following commonly-used assumptions used in estimation of Ks are not plausible in such regions: (i) Ks decreases with increasing soil depth, (ii) soils with coarse-grained texture (sandy loam) have higher surface Ks values compared to relatively fine-grained texture (fine sandy loam), and (iii) pedo-transfer functions are not reliable methods of estimating Ks. Our results also reveal that remote sensing data can provide useful information for estimation of surface and subsurface Ks values.

Midlatitude mountain snowpacks are a critical freshwater resource. However, climate warming is endangering its natural water tower function by decreasing the fraction of winter precipitation falling as snow in an intensifying precipitation regime. In a warmer future, snowpacks may therefore shift from seasonal reservoirs to posing more flood hazards, stressing current ecosystems and water resource management practices. Successful adaptation and hazard preparation to snowpack shifts is predicated on a first-principles understanding of its driving processes. However, most long-term observations and model-projections are incapable of resolving process-level snowpack behaviors, leaving an incomplete toolkit for assessing and interpreting environmental threat levels in snow-fed watersheds. This dissertation addresses this problem by investigating process-level snowpack behaviors during warm storms in California’s Sierra Nevada through sub-daily snow, soil and hydrometeorological measurements, and with supporting observations from atmospheric reanalyses and satellite remote sensing in three studies. The first study delineates an hourly relationship between air temperature, atmospheric moisture, and changes to snow water equivalent (SWE) spanning the Sierra during water years 2010-2019. Conditions balancing precipitable water (humidity) and snowfall requirements (temperature) favor SWE accumulation. SWE increases are observed above 1.0˚C when moisture supplies are modest or snowpacks sufficiently deep, and are otherwise immediately followed by SWE loss. The second study focuses on this SWE “oscillation” in the Northern Sierra during rain-on-snow (ROS) events in January and February 2017. Snowmelt was a weaker driver in ROS flooding than previously documented. Snowpacks may rather “passively” route rainwater through snow, where the saturation and liquid drainage of snow causes observed SWE to rise and immediately fall. The final study elaborates on this case study to other ROS events in water years 2017-2019. Results describe both the limitations common to standard daily observations, and the importance of storm sequencing on augmenting ROS impact. Sierra Nevada watersheds remember large prior storm inputs to soils and streams that imminent storms stand on top of. These observations present a nuanced and integrated perspective to understanding how snowpacks respond to warm storms, which may be beneficial to future water resource and flood risk forecasting interpretations and discussions.

The role of satellite-based remote sensing in improving hydrologic and water resources studies in data-scarce regions is investigated. Specifically, the dissertation focuses on the development of a 1) validation framework for remotely sensed precipitation and evapotranspiration without the use of ground-based observations, 2) methodological framework for calibration of large scale hydrologic models with multiple fluxes, and 3) a seasonal hydropower planning framework for data-scarce regions. In the first part of the dissertation, a root mean square error (RMSE)-based error metric capable of translating individual biasies in precipitation and evapotranspiration onto the Budyko space is developed. It is shown that the framework succeeds in arriving at the same conclusions as a traditional validation method. In the second part, the value of incorporating multiple hydrologic variables such as evapotranspiration, soil moisture and streamflow into model calibration is investigated. It is shown that parameters which are insensitive to individual model responses can influence the trade-off relationship between them. Finally, the potential of using remotely sensed precipitation and evapotranspiration datasets in generating reliable seasonal reservoir inflow forecasts for hydropower planning is investigated. Results highlight the importance of accounting for input and parameter uncertainty in hydropower planning.

In recent years, quantitative precipitation forecast became viable for the emerging needs around the world. The quality of the precipitation forecast would be crucial for further application. The purpose of this study is to understand the accuracy of the seasonal precipitation forecast and explore the potential of improving the forecast, as well as understanding the precipitation forecast performance from lead time 1 day to 10 years.To understand the accuracy of the seasonal forecast around the world, the forecast of North American Multi-Model Ensemble (NMME) in the Blue Nile basin, Western United States, Tigris-Euphrates River basin and Brazil is evaluated, and compared with gridded precipitation observation products. Results show that NMME overall can explain from 0% to up to 55% of the wet season precipitation variance, varying different models and regions. Also, in all selected basins, atmospheric climate indices can be used as a reliable predictor to forecast the wet season precipitation in linear regression. By combining the climate indices and NMME, the newly formed statistical model can have better performance, compared to NMME model results. This opens the possibility of utilizing machine learning techniques to reliably improve the seasonal precipitation forecast. The analysis of precipitation forecast across different lead time shows that the performance from lead time 1 day to 10 years is not dropping as lead time increases. When two products’ lead time overlaps, using the longer-range forecast product can have better forecast results, except decadal forecast.