H21A-0802
Rainfall-Runoff Simulations with Regional Circulation Rainfall Model and Hydrological Watershed Model
The Taiwan Typhoon and Flood Research Institute (TTFRI) is a newly founded national laboratory in Taiwan. The core mission of TTFRI is to conduct research to improve our understanding of natural hazards, including typhoon and flood. One of the major tasks of TTFRI is to develop locally coupled meteorological rainfall model and hydrological watershed model in an earlier warning system to abate the lost of lives and properties. This paper will examine some simulations in our watershed scale modeling activities using precipitation from mesoscale numerical weather prediction systems [for examples, Weather Research Forecasting (WRF) model, Purdue Regional Circulation Model (PRCM), etc] as input. The long-term goal of TTFRI is to deliver fully coupled weather-hydrology interaction models. However, only off-line simulations are conducted in this paper because the development of a fully integrated or coupled rainfall-runoff model is yet to complete. The observed gauge rainfall data and the predicted rainfall data from WRF are used as our meteorological forcing on watershed modeling. A physically based distributed hydrological model, WASH123D, is applied to simulate floods during the typhoon landfall periods. The predicted rainfall hydrographs using WRF are compared with rain gauge observations to explore their discrepancies. Various starting points from cyclone generation processes are used to investigate the subsequent effects on floods. The simulated flood hydrographs are compared with observed flood discharges in terms of the magnitude and time lag of flood peaks. Sensitivities of both terrain and spatial precipitation distributions on areal inundations are explored. Finally, uncertainty assessments on rainfall-runoff modeling are discussed.
H21A-0803
Change-Point and Trend Analysis on Annual Maximum Discharge in Continental United States
Annual maximum discharge records from 36 stations representing different hydro-climatic regimes in the continental United States with at least 100 years of records are used to investigate the presence of temporal trends and abrupt changes in mean and variance. Change point analysis is performed by means of two non- parametric (Pettitt and CUSUM), one semi-parametric (Guan), and two parametric (Rodionov and Bayesian Change Point) tests. Two non-parametric (Mann-Kendall and Spearman) and one parametric (Pearson) tests are applied to detect the presence of temporal trends. Generalized Additive Model for Location Scale and Shape (GAMLSS) models are also used to parametrically model the streamflow data exploiting their flexibility to account for changes and temporal trends in the parameters of distribution functions. Additionally, serial correlation is assessed in advance by computing the autocorrelation function (ACF), and the Hurst parameter is estimated using two estimators (aggregated variance and differenced variance methods) to investigate the presence of long range dependence. The results of this study indicate lack of long range dependence in the maximum streamflow series. At some stations the authors found a statistically significant change point in the mean and/or variance, while in general they detected no statistically significant temporal trends.
H21A-0804
Trends in the Timing and Volume of Peak Flow Events in the Missouri River Basin
Long-term changes in streamflow may respond to climate change which will impact management practices for water resources and wetland ecosystems. Previous studies of mean annual streamflow in the Missouri River Basin indicate statistically significant decreases in the western part of the basin and increases in the eastern part over the last 50 years. Trends were further explored from 1950-2007 at three streams in the basin that are part of the USGS Hydro-Climatic Data Network (HCDN): Cheyenne River at Edgemont SD, Yellowstone River at Billings MT, and James River at Scotland SD. Patterns were examined for annual peak flow and flows exceeding the 2-year recurrence interval (RI). Changes in the nature and timing of peak flows are of particular geomorphic importance in that they sculpt and control channel form. The Cheyenne River exhibited a marked decrease in the volume of water conveyed by floods since 1980. From 1950 to 1980, 20 peak flow events exceeded the 2-year RI and conveyed a total of 763,000 acre-feet (AF). From 1980 to 2007, there were only 9 such events and they conveyed 134,000 AF. From 1950-70, the annual peak discharge occurred between May and August. Since 1970, timing of annual peak discharge is more variable, occurring from March to October. The Yellowstone River had 18 peak flows exceeding the 2-year RI prior to 1980 and conveyed a total of 16,300,000 AF, and 11 such events since 1980 conveyed 9,160,000 AF. By contrast, the James River showed increasing volumes conveyed by floods since 1980. From 1950 to 1980, the 11 peak flows exceeding the 2-year RI conveyed a total of 4,210,000 AF, and 15 events after 1980 conveyed 12,200,000 AF. Peak flow exceeded the 2-year RI in 8 out of 9 years from 1993 to 2001. The change in frequency of channel shaping flows and volume of runoff will over time change the geomorphic nature of these streams and rivers.
H21A-0805
Secular Streamflow Trends in Watersheds Receiving Mixed Rain and Snow, Pacific Coast and Cascades Ranges
Much existing research has focused on detecting climate change effects on snowmelt-dominated watersheds, but in the Pacific Coast and Cascades ranges, precipitation falls as either rain or snow, depending on latitude, elevation, and season. Watersheds often straddle the snow line, with some areas dominantly receiving rain and higher elevations accumulating seasonal snowpacks. These snowpacks are near the 0 ° C threshold, making them sensitive to the effects of climate warming. Climate sensitivity of seasonal and event hydrographs from watersheds with mixed rain and snow has not been fully explored. This project investigates detectable climate change signals in long-term streamflow records in the Washington, Oregon, and northern California Coast and Cascades Ranges. Watersheds with mean elevations above the seasonal snow line show significant increases in streamflow during January through March and decreases in the percent of annual flow during April through June, the historical snowmelt period. These changes were not detectable in watersheds with mean elevations below the seasonal snow line. There were no consistent trends in peakflow dates or volumes. The multiple drivers of peakflow occurrence make it unlikely that any changes in peakflow timing will be detectable for several decades. Results suggest that Coast Range hydrology has been minimally impacted by historical climate warming, but that Cascades Range watersheds are already experiencing altered hydrologic regimes.
H21A-0806
Impact of Climate Change on Extreme Rainfall in France Throughout Trend Detections in Average Climatic Characteristics
The great interest on climate change during these last years has led to a quasi unanimous conclusion for scientists: the Earth climate changes. An increase of precipitation in the middle and high latitude area of north hemisphere was detected. In order to prevent hydrological risks, it's interesting to know if these global changes lead to an increase of extreme events. Indeed in the hydrologic context, the estimation of return period is primary for hydraulic building dimensioning. When considering problem of fire, urbanisation, hydrological installations, the study of water runoff alone can be misleading. So we must do a preliminary work on rainfall. However climate models have difficulties to represent efficiently extreme events. Moreover classical statistical methods seem to be limited because of the lack of very long series which can lead to a bad estimation of distribution tails. An original approach is applied to estimate impacts of climate change on extreme events in using an hourly rainfall stochastic generator which can be coupled with a rainfall-runoff model. The climate evolution is detected by the generator parameters. These parameters are estimated by an average, and not by extreme values, of climatic characteristics. From daily information of 139 rain gauge stations, evolution of generator parameters has been studied in the metropolitan France between 1960 and 2003. We applied the Poisson-Pareto-Peak-Over-Threshold model and linear trend has been tested by the Maximum Likelihood Ratio test. Parameter evolution has been evaluated from a regional approach in clustering several stations. Changes on average rainfall characteristics are amplified on extremes. The observed trends occur mainly between December and May. Taking into account the climate change doesn't lead to a big change in the quantiles estimation compared to their estimation under a hypothesis of climate stationary. But if observed trends are confirmed in the future, extreme event occurrence probabilities will increase notably in the North- West, East and mountain landscape of France in the next years. Finally, we compared trends observed on data to scenarios proposed by the climatic model used for the IMFREX project. Indeed it proposes evolution scenarios of some average variables which are well correlated with the rainfall generator parameters.
H21A-0807
An Analysis on Spatiotemporal Variations of Soil and Vegetation Moisture from a 29 year Satellite Derived Dataset over Mainland Australia
The spatiotemporal behavior of soil and vegetation moisture over mainland Australia were analyzed with a 29 year satellite derived dataset. Such a recently available dataset was derived from passive microwave observations by four satellites going back to late 1978. Differences in measurement specifications prevented merging the data directly. A continuous product was developed for Australia by scaling percentiles of the cumulative moisture distribution within each grid cell to the same percentiles of a reference sensor. The R2 and root mean squared error (RMSE) between rescaled values and the reference and time series for three locations were derived. Using the resulting product, we confirmed a strong El Niño–Southern Oscillation signal in near-surface hydrology across Australia. Spatial patterns of trends in annual averages showed that western and north-western Australia experienced an increase in vegetation moisture, while the east and south-east experienced a decrease. Soil moisture showed a similar spatial pattern, but with larger regions experiencing a decrease, resulting from the decreasing rainfall and increasing potential evapotranspiration during the extended winter (May–October). Development of a merged long-term global data set would enable better estimation of hydrological variables and better understand the impacts of ocean circulation on soil and vegetation moisture dynamics.
H21A-0808
Some Examples Of Water Resources Variability In The Context Of Climatic Fluctuations
The determination of the impact of climate change on hydrological systems and their water resource constitutes a major stake of the 21st century to which the scientists must answer. First of all, it is necessary to understand how climate are expressed in the hydrosystems. For several years, the M2C laboratory of the University of Rouen has tried to answer this question by working within the framework of many regional, national and international programs as well as PhD works. Those studies involve analyses of hydrological systems located: (1) in various climatic and geomorphological contexts on both sides of the Atlantic Ocean, (2) in various hydrological compartments (surface and ground water), (3) at various spatial scales (watersheds smaller than 1000 km2 and large rivers). The approach consists in studying the long-term changes, oscillations and fluctuations of hydrologic variables by the analysis of time series (precipitation, discharge, piezometry), in particular by means of signal analysis and processing methods. The studied hydrosystems are small watersheds and aquifer in Haute- Normandie, the Seine river (NW France), north-african watersheds (W Morocco and N Algeria), small watersheds and aquifer in Texas, the Colorado river (Texas) and the Mississippi river. Although the identification of structured variations might be uneasy – sometimes just impossible – in raw data, wavelet analysis, for instance, makes it possible to detect localized energetic structures and possible periodicities in all the studied hydrosystems and to quantify them. In many surface hydrosystems we note an intensification of the annual energy band which corresponds to the hydrological cycle. In the NW of France and North Africa, we observe 2-3-year and 5-7-year modes which could be linearly related to fluctuations in the NAO using wavelet coherence. In the USA, we notice similar 2- 3-year and 5-7-year modes that might be possibly related to the characteristic 2-4-year and 4-8-year of SOI. In any case, two major temporal discontinuities were systematically recovered around the 1970's and the 1990's characterized respectively by the occurrence of the 5-7-year and 2-3-year peaks. The above- mentioned intensification of the hydrological cycle is observed from 1990 up to now. These results would describe a global pattern in hydrological processes as a response to climate fluctuations.
H21A-0809
Study on the spatiotemporal variability and affecting factors in soil moisture at a humid area
The spatiotemporal variability of soil moisture and its affecting factors in a humid area were examined based on the field measuring date in the Tai lake drainage basin, China. 24 sensors near the soil surface and 12 sensors in 2 profiles (6 in each) were set up for collecting hourly soil moisture data with the Frequency Domain Reflectometry (FDR) sensors in 2006. Coefficient of variation (CV) and semi-variogram were calculated to evaluate the temporal variability in different locations and the spatial variability in different periods. The surface soil moisture appears middle or weak variability, and most of the CV values are in the range of 0.13-0.26. Soil characteristics, topography, vegetation, meteorological factors and human activities influenced the soil moisture spatiotemporal variability significantly. The factors appear having different affecting abilities on the spatiotemporal variability, and the domain factors are different in four seasons. Soil characteristics mainly influence the temporal variability in the scale of hill slope. Coarser texture on the upper part of the slope results in a larger variability. Topography and micro-topography affects the spatial variability in all 3 dimensions. The variability is larger at upper locations and chine of the slope. The effect of vegetation on the soil moisture variability is stronger in spring, summer, and autumn than in winter, according to the different growth activities and water demand. The trees on the slope influence the CV values along the slope. Meteorological factors are the forcing factors of the soil water variation. Higher rainfall and evaporation variations produce higher variability in soil moisture while the rainfall has more influence in the summer and the evaporation has more in the fall. The results provide better understanding of soil moisture variation and base for further study on how the soil moisture variation could affect the rainfall runoff partitioning.
H21A-0810
Investigating the potential of using tree-ring records and river-dammed lake sediments to reconstruct paleohydrology in the Virgin River watershed, southern Utah: Initial observations and results
The Virgin River drains an area of over 13,000 km2 in southern Utah, northern Arizona and southern Nevada before entering into Lake Mead as a major tributary to the Colorado River. Peak flood discharges occur during the spring snowmelt runoff and floods related to monsoon-driven mid-summer convective storms (USGS gage data). Years with above average winter snowpack and high spring meltwater discharges have been linked to El Nino conditions, annual droughts correlate to La Nina years (e.g. Brown and Comrie, 2004). Weather station and river gage records used to reconstruct precipitation and flood frequency are limited to the last century. However, slackwater flood deposits (Enzel et al., 1994) and tree-ring (e.g. Hereford et al. 2006) and lake sediment records may be able to extend this record over the last several thousand years. Research has begun to investigate the use of tree-ring records from the middle to upper Virgin River watershed in southern Utah for paleohydrologic reconstruction. Initial pinyon pine tree-ring measurements show a good relationship to precipitation, peak spring snowmelt runoff and annual discharge in the North Fork of the Virgin River. One year lags in tree-ring thickness are common within the record due to energy storage from a preceding wetter year. Additional tree cores from a number of species and locations within the watershed are needed to produce an accurate tree-ring index and paleohydrologic reconstruction. In addition to tree-ring records, sediment archives within the Virgin River drainage are being examined for their potential for paleoflood reconstruction. A large landslide within Zion Canyon in Zion National Park dammed the North Fork of the Virgin 8000 years ago and created a 60 m deep lake that existed until just after 3600 years ago (Biek et al., 2000). Lake sediments consist of alternating coarse (silty-sand and fine sand) and fine-grained layers (clay, clayey-silt, silt). Initial descriptions suggest that these coarse-fine couplets may represent annual deposition. Moreover, variations in couplet thickness may be related to changes in annual/seasonal discharge and flood frequency. Initial results and observation from these sediments are presented along with relationships to other paleoflood studies in the region.
H21A-0811
Interannual Variation of Summertime Precipitation around the Northeast Part of the Tibetan Plateau in China
It is crucial to quantitatively estimate the orographical precipitation over the upper reaches of the rivers which
flow into arid/semi-arid regions for hydrological managements. Besides, to understand the mechanism of
interannual variation of the precipitation on such terrain is essential for predicting, monitoring and managing
the hydrological resources as well as interpreting paleo-climate proxy data such as ice cores, tree rings and
lake sediments.
‚ve analyzed a new rain-gauge based precipitation dataset which reproduce orographic effect by adopting
Parameter-elevation Regressions on Independent Slopes Model (PRISM) with a meteorological reanalysis
dataset.
To better understand the time-space structure of the interannual variation of precipitation in the summer
time, hereafter meaning June to August (JJA), in the Qilian Mountain region, an empirical
orthogonal function (EOF) analysis is applied to the summertime precipitation times for 41 years (1961-2001)
over there. The atmospheric circulation field, the moisture transport and the precipitation
patterns associated with the dominant mode is shown by using European Centre for Medium-range
Weather Forecast (ECMWF) reanalysis (ERA.40, 1.125 degree grid) dataset and a new precipitation
dataset which expresses an orographic enhancement of precipitation.
The first mode of the analysis, which explained 24.8% of the total precipitation variance, had a
single mode centered along the Hexi Corridor region of this mountainous region with a weak
increasing trend of precipitation. The second mode, which explained 15.8% of the total variance, had
a dipole structure between the Hexi Corridor (north of 38N) and the region south of this corridor.
This mode was correlated with the Indian Summer Monsoon Rainfall (IMR), and the northern part of
the Qilian Mountains had a negative correlation to IMR. The third mode had a dipole structure
between the Tibetan Plateau area and the northern desert area. The westerly circulation anomaly
was the dominant factor to define the wet (much precipitation) or dry (less precipitation) years around
the Qilian Mountains. The second mode, which had a significant negative correlation with the IMR,
was also related to the westerly circulation anomaly. In the southern part of the Qilian Mountains,
a pressure anomaly around the western plateau area and the southwesterly moisture flux were also
related to the interannual variation of the precipitation anomaly in this region.
http://www.chikyu.ac.jp/precip
H21A-0812
Distribution of Warm Season Precipitating Storms in the Southern Great Plains
We use a multisensor gridded precipitation product to examine the interannual and intraseasonal variability of precipitating storms during the months of April – September of 1996 -2006 in the Arkansas-Red River Basin. We are able to divide the precipitation field into individual storms by designating regions of contiguous precipitation (in space and time). We thus have a record of the size and lifespan of storms responsible for individual precipitation events. Our data set has a total of 519,562 storms whose numbers vary by year and month. Storms are most numerous in the eastern and western extremes but the central part has the most precipitation. We find that the average storm is small with a maximum size under 500 km2 and a lifetime between 1 and 1.5 hours. Storms in August are smaller but longer lasting than those in April. The linear correlation between mean storm size and number of storms per year is -0.77. Assuming all storms to be convective, we can divide the storms into single ordinary thunderstorms, multiple thunderstorms (includes supercells), and Mesoscale Convective Systems (MCS). The MCS are between 1 and 1.5% of all storms but account for about 85% of the precipitation. Nevertheless, the linear correlation between the number of MCS in each six month period and the total amount of precipitation in the period is only 0.42 We looked at the characteristics of storms which occurred in years with modest as well as abundant precipitation. We did not find any specific characteristics common to either years with less precipitation or years with more. A combination of factors controls interannual precipitation variability. In future work we plan to examine the special density of these storms in more detail as well as the initiation and termination times of the storms. We are also interested in storm movement and how it changes over the storm lifetime.
H21A-0813
Multiscale Riverflow Variability in the British Isles and Climate teleconnections
We have investigated riverflow variability in the British Isles by examining the reconstructed monthly
discharge time series at fifteen catchments in England and Wales for the period 1865-2002. The riverflow
fluctuations exhibit a strong annual cycle. The flow in the annual cycle is found to be intermittent, with the
degree of intermittency varying from one catchment to another. An intermittent flow is characterized by bursts
of high discharge separated by intervals with low or no discharge. By applying a continuous wavelet
transform to the time series and computing the wavelet power spectrum, we have identified the occurrence of
intermittency in the annual cycle. The riverflow activity is also found to exhibit variations at interannual and
interdecadal timescales. These variations may be linked to large-scale climatic processes such as the North
Atlantic Oscillation (NAO). The interannual fluctuations also exhibit temporal intermittency which is detected
from the wavelet power spectrum. The NAO is known to be intermittent at inetrannual timescales. We have
used the kurtosis of the probability density functions of the various time series as a measure of the degree of
intermittency. An intermittent flow is characterized by a peaked (super-Gaussian) probability density function
with kurtosis in excess of 3. A higher value of kurtosis signifies a higher degree of intermittency. We have
also performed a multiresolution analysis of the discharge time series using a discrete wavelet transform. By
separating the discharge signal into different frequency bands, we have delineated the influence of NAO on
riverflow variability at the interannual and interdecadal timescales. In addition, we have applied the method of
Empirical Orthogonal Functions (EOFs) on the fifteen discharge time series and performed a wavelet
analysis of the derived principal components. It is shown that the dominant principal component exhibits
fluctuations at the interannual and interdecadal timescales which are linked to NAO.
http://www.math.iupui.edu/~sen
H21A-0814
Long Lead-time Streamflow Forecasting Incorporating Climate Variability
An evaluation of the influence of oceanic-atmospheric climate variability on streamflow in the upper North Platte River basin (Wyoming, USA) is presented. Through the application of Singular Value Decomposition (SVD) statistical methods, sea surface temperatures (SSTs), 500 mbar geopotential height (Z500) values and North Platte streamflow were evaluated over a historical period from 1948 to 2006. This resulted in the identification of new regions of highly correlated SSTs and Z500 that may not be represented by existing index regions (Niño 3.4 —defined El Nino Southern Oscillation region, PDO--Pacific Decadal Oscillation, and AMO—Atlantic Multidecadal Oscillation). A long lead-time approach was utilized such that a three month lead- time (seasonal average of monthly SSTs or Z500 for October, November and December) as well as a six month lead-time (seasonal average of monthly SSTs or Z500 for July, August and September) of previous year variability were used as predictors for the following year spring streamflow (seasonal monthly average of April, May, June and July). Temporal expansion series from SVD were utilized as predictors in a non- parametric model to develop continuous exceedance probability forecasts. The results displayed good skill using SSTs for the six month lead-time forecast and excellent skill using Z500 values for the three month lead-time forecast. The improved skill found over basic climatology forecasts will be useful to water managers when trying to predict and manage expected streamflow volumes several months in advance.
H21A-0815
Trends in the First Two Moments of Peak Flow Data in California
The U.S. Geological Survey (USGS) is currently updating California's flood frequency statistics, for the first time in 30 years. This much overdue project is necessary to not only help protect lives and property, but for the effective planning management, and use of the State's land and water resources, both of which are coming under unprecedented demand in the 21st Century. This new study has the benefit of using an additional 30 years of peak discharge data, as well as new methodologies for incorporating historical floods to more accurately predict characteristics of flood frequency distributions. Yet with global climate change likely to affect long-term streamflow characteristics, a fundamental assumption of flood frequency analysis- stationarity(no systematic change over time) of the annual flood data--is being questioned. To test whether the first two moments (mean and standard deviation) required to fit probability distributions to annual peak flow data are stationary, trends in the moments over three different periods, will be examined at currently operated or recently discontinued (through water year 2006) USGS stream-gaging stations. All stations included in this trend analysis will be stream sites that the USGS database indicates have no upstream regulation or diversion effects, nor any significant effects due to urbanization. Thus these sites will represent unregulated systems for which trends, if any, would most likely be the result of changing climate forcings. To ensure that the sampled moments are reasonably stable and reliable representatives of the population moments, a 30-year record period will be used to calculate the mean and standard deviation. The three different periods selected for testing trends will cover the past 30 years (1977-2006), the past 40 years (1967-2006), and the past 50 years (1957-2006). For each successive year in the trend test period, the mean and standard deviation will be calculated using a sliding-window basis using only the previous 30 years. For the 30-year trend analysis, 36 stations with at least 60 years of continuous peak data records were used, the 40-year trend test used 23 stations with at least 70 years of record and the 50-year trend test used 10 stations with at least 80 years of data. No temporal or spatial biases were evident in the dataset of selected stations, even in the limited number used in the 50-year trend test. The Kendall Tau test for monotonic trend was used for testing the mean and standard deviation for each site and each of the three trend-test periods.
H21A-0816
Atmospheric Variability And Its Influence On Winter Hydroclimate In Upstate New York
Hydroclimatic trends are frequently used to assess regional impacts of global climate change, requiring that the combined effects of linear trends and non-linear quasi-periodic fluctuations associated with hemispheric- scale teleconnections be identified. Separating these effects is especially problematic outside the statistical nodes of influence, such as in the Great Lakes drainage area of Upstate New York (UNY). Daily mean temperature, precipitation and snowfall data, and streamflow were analyzed from 1950 to 2007, to determine relationships between regional hydroclimate and atmospheric teleconnections. Non-parameteric testing was used: the Mann-Kendall trend test to assess linear changes; and the Kruskal-Wallis and multiple comparison tests to observe differences due to the NAO, ENSO, PNA, and PDO teleconnections. The results highlight five key findings: the importance of the NAO on regional temperatures in December; the significance of the PNA and PDO on January hydroclimate; the role of ENSO in March snowfall; a strong linear trend in October of increasing precipitation, decreasing temperature and increasing streamflow; and a strong linear increase in snowfall in the lake snow-affected regions. Teleconnection linkages generally parallel relationships observed in neighboring regions; however, effects in UNY follow an intra-seasonal variation in Atlantic and Pacific influences. For winter forecasting these results suggest an affinity with New England in December, Ohio River Valley in January, and the Great Lakes in March. Aside from the NAO, the lack of regional coherence suggests, that although subhemispheric-scale linkages exist, they are not the dominant signal; the increasing January snowfall is a long-term linear change.