Spatio-Temporal%20Variability%20of%20the%20North%20American%20Monsoon

1
Spatio-Temporal Variability of the North American
Monsoon

• Balaji Rajagopalan
• Department of Civil, Env. And Arch. Engineering
  (CEAE) and CIRES
• Katrina Grantz, Edie Zagona
• CEAE/CADSWES
• Martyn Clark (CIRES)

2
Introduction

• North American Monsoon
• Dramatic increase in rainfall from an extremely
  dry June to a rainy July, August and September
• SW United States and NW Mexico
• Shift in winds (from westerlies to southerlies)
  brings moisture over the land
• Warm land moisture monsoonal precipitation
• Afternoon thunderstorms

3
Introduction

• Significant portion (30-50) of annual
  precipitation falls in the summer months in
  Arizona and New Mexico
• Water management perspective important to
  predict the strength of the monsoon
• for water supply and environmental planning
• Few studies on timing, streamflow, and
  implications for water management
• important to local communities in the monsoon
  region

4
North American Monsoon Experiment Schematic
Schematic Illustrating the multi-tiered
approach of the North American Monsoon Experiment
(NAME). The schematic also shows mean
(July-September 1979-1995) 925-hPa vector wind
and merged satellite estimates and raingauge
observations of precipitation (shading) in
millimeters.
5
North American Monsoon Schematic
6
Past Studies

• Spatial and temporal variability depends on
• Location of subtropical jet, topography, SSTs
• Moisture comes from Gulf of California and Gulf
  of Mexico
• Wet winter ? Early monsoon onset and vice-versa
  (Higgins Shi, 2000)
• General negative correlation with previous
  winters precipitation (Gutzler, 2002 Higgins
  Shi, 2000) related to North Pacific SSTs
• El Nino ? weaker/southward monsoon ridge (Castro,
  2000)
• Increased monsoonal precipitation with increased
  soil moisture
• (Toshi, et al. 2003 Small,2000)

7
Past StudiesHiggins et al. 1998
Composite evolution of the 30-day running mean
area average precipitation (units mm/day) over
Arizona and New Mexico for wet monsoons (dotted
line), dry monsoons (dot-dashed line) and all
(1963-94) monsoons (solid line). The average
date of monsoon onset is July 1 for wet monsoons,
July 11 for dry monsoons and July 7 for all
monsoons (defined as day 0 in each case).
8
MotivationSeasonal cycle shifts in Western US
hydroclimatology (Regonda et al. 2005)
9
Contd
10
Motivation

• What are the large-scale Ocean/Atmospheric/Land
  drivers of space-time (timing and amount)
  variability of N. American Monsoon (interannual /
  decadal)?
• Potential Long-lead Precitability?
• Implications for water management.
• Need a systematic investigation of the monsoon
  rainfall attributes and also the streamflows.

11
Outline

• Study area, data
• The research project
• Precipitation diagnostics
• Streamflow diagnostics (preliminary results)
• Water management issues
• Incorporating forecasts into water management

12
Study Area

• Monsoon Region (NAME II) Arizona and New Mexico
• Water Management Gila River
• Flows from New Mexico through Arizona, joins
  Colorado River at Yuma, Arizona

AZ
NM

13
Data

• Precipitation
• Monthly climate division data (1948-2004)
• NM (8 divisions), AZ (7 divisions)
• Daily NWS co-op data (1948-1999)
• 219 stations across AZ and NM
• Temperature, PDSI
• Monthly climate division data (1948-2004)
• Large-scale climate variables
• Monthly NOAA NCEP-NCAR reanalysis data
  (1948-2004)
• SST, precipitable water, geopotential heights,
  vector winds
• Streamflow
• Daily HCDN streamflow (1948-1999)
• USGS daily/monthly values (1948-2005)

Co-op Stations
14
The Research Project

• Precipitation diagnostics
• Streamflow diagnostics
• Water management issues
• Incorporating forecasts into water management

15
Precipitation Diagnostics

• Monsoon cycle
• Monsoon rainfall
• Plausible hypothesis
• Antecedent land conditions
• Antecedent ocean Atmospheric conditions

16
Precipitation DiagnosticsMonsoon Cycle

• Monsoon timing at each co-op station
• Calculate Julian day when 10, 25, 50, 75, and
  90th percentile of the monsoon season (July-Sept)
  precipitation has fallen (each station, each
  year)
• -- Julian day time series for each threshold
  for each station
• Perform trend analysis on the Julian day time
  series
• 10th, 25th, 50th, 75th, and 90th percentiles
  capture the entire monsoon cycle onset, peak
  and recedence

                          
17
Precipitation DiagnosticsMonsoon Cycle

• Use Spearman rank correlation to detect trends
• A nonparametric (distribution-free) correlation
  statistic
• doesnt require that data be normally distributed
• similar to Pearsons R, except that the values
  are converted to ranks before computing the
  correlation coefficient.
• where D is the difference of the rank numbers.
  (Spearman, 1904)
• gives p-value and slope
• Correlatate value with time to get trend
• Results similar for Pearsons R

                          
18
Figures
Climatological Julian day
Trends
5th Percentile

• Largest circle gt 21 days
• Second largest 15-21 days
• Second smallest 10-15 days
• Smallest circle lt 10 days
• Filled circles significant at 90

25th Percentile
50th Percentile

• lt July 19th
• Jul 20th 29th
• Jul 20th Aug 8th
• Aug 9th Aug 18th
• Aug 19th 28th
• gt Aug 29th

75th Percentile

• Entire Monsoon Cycle Shifted Later
• Approx. 1015 days shift

95th Percentile
19
(No Transcript)
20
Precipitation DiagnosticsMonsoon Rainfall Amount

• Monsoon rainfall amount at each station and
  climate division
• July, August, September, July - September
• Spearman rank correlation to compute the trends

                          
21
Precipitation DiagnosticsMonsoon Rainfall

• Decrease in July
• Increase in August and September (esp in NM)
• July- September NM increase, AZ mixed/decrease
• Consistent with the monsoon timing results

                          
22
Precipitation DiagnosticsRainfall Amount
Co-op station trends similar to climate division
trends
Relative circle size indicates the magnitude of
the trend (slope) 0.4 (largest circle),
0.3-0.4 (second largest), 0.2-0.3 (second
smallest), less than 0.2 (smallest circle).
23
Precipitation DiagnosticsMonsoon Rainfall

• Is the trend a steady increase, or jump, or ?
• August precipitation over AZ and NM 5 year
  moving window
• Eastern region (NM) gets wetter in the later
  period
• Western region (Arizona) trend not as distinct
• Shift most apparent after a dry spell in the late
  1970s

24
Precipitation DiagnosticsMonsoon Moisture

• Trends in Palmer drought severity index (PDSI)
  and 850mb precipitable water
• Corroborate results seen in precipitation (more
  so with precipitable water)

                          
25
Precipitation DiagnosticsPlausible Hypothesis

• What is driving the delay in the monsoon cycle?
• Hypothesis
• Increased pre-monsoon (antecedent winter/spring)
  soil moisture ? longer summer heating to set up
  the land-ocean gradient ? delaying the monsoon
  cycle
• Wetter antecedent winter / spring conditions in
  southwest driven by increased El Nino Southern
  Oscillation (ENSO) activity in recent decades

                          
26
Precipitation DiagnosticsAntecedent Land
Conditions

• December- May precipitation and PDSI trends
• Increasing trend in southwest, decreasing trend
  in northwest classic ENSO teleconnection pattern

                          
27
Precipitation DiagnosticsAntecedent Land
Conditions

• Relate winter/spring hydroclimate to summer
  monsoon attributes
• Principal Component Analysis (PCA) to find the
  dominant modes of variability in summer timing
  and rainfall
• Leading modes can be thought of as spatial
  average
• Timing PC1 28 of variance
• July precip PC1 45 of variance
• July-Sep precip PC1 45 of variance
• Correlate leading modes with antecedent land
  conditions

                          
28
Precipitation DiagnosticsAntecedent Land
Conditions

• Correlate 50th and 10th percentile timing PC1
  with antecedent precipitation/PDSI across western
  US
• Significant positive (negative) correlations in
  southwest (northwest)
• Correlations are stronger for the 10th
  percentile timing PC ? onset of monsoon more
  strongly affected by antecedent soil conditions

                          
29
Precipitation DiagnosticsAntecedent Land
Conditions

• Correlate July and Jul-Sep precipitation PC1 with
  antecedent precipitation/PDSI across western US
• Significant negative (positive) correlations in
  southwest (northwest)
• This negative correlation between winter/spring
  precipitation and monsoon precipitation noted by
  Gutzler (2000)
• Correlations are stronger for July PC ? Early
  monsoon rainfall more strongly affected by
  antecedent soil conditions

                          
30
Precipitation DiagnosticsAntecedent
Ocean-Atmospheric Conditions

• Large-scale Drivers of summer variability
• Correlate leading modes (PCs) of monsoon rainfall
  and timing with antecedent Ocean Atmospheric
  variables.

                          
31
TimingAntecedent Ocean Conditions

• Correlate summer timing PCs with winter/spring
  SST and Z500
• Positive correlations in equatorial Pacific- ENSO
  pattern
• increased SSTs in winter/spring go with increased
  Julian day (i.e., delayed monsoon)
• Correlations slightly stronger for 10th
  percentile (onset of monsoon)

                          
50th percentile Timing PC
10th percentile Timing PC
32
TimingAntecedent Atmospheric Conditions

• Correlate summer timing PCs with winter/spring
  Z500
• PNA type pattern consistent with SST correlations

50th percentile Timing PC1
10th percentile Timing PC1
                          
50 Timing PC
10 Timing PC
33
Rainfall amountAntecedent Ocean Conditions

• Correlate summer rainfall PCs with winter/spring
  SSTs
• July negative correlations in equatorial Pacific
  La Nina pattern
• Decreased SSTs in winter/spring go with increased
  July precipitation (La Nina typically goes with
  decreased winter/spring precip -gt increased July
  precipitation
• Correlations flipped for
• Aug, almost no pattern
• for Sep and Jul-Sep
• Early monsoon precip
• amount affected by SSTs
• but later monsoon may
• have different drivers

                          
34
Rainfall amountAntecedent Ocean Conditions

• Correlate monsoon rainfall PCs with winter/spring
  Z500
• Results consistent with SST correlations
• July PNA type pattern
• Correlations weaker and
• reversed sign for Aug,
• Sep and Jul-Sep.
• Early monsoon precip
• amount affected by pre-
• monsoon Pacific Ocean
• and Atmospheric features,
• but later monsoon may
• have different drivers

July
Aug
                          
Sep
Jul-Sep
35
Rainfall AmountHigh-Low Composites
Winds
Z500
SST
Jul
Aug
Sep

• Aug and Sep rainfall extremes impacted by the
  surrounding Ocean/Atmospheric status

36
Precipitation DiagnosticsConclusions

• Entire Monsoon cycle shifted approx 1015 days
  later in recent decades
• Consequently, decreased rainfall in July and
  increase in Aug and Sept
• Increased pre-monsoon precip/soil moisture
• (driven largely by large-scale Pacific
  Ocean/Atmospheric features)
• Leading modes of Monsoon timing and early (July)
  rainfall strongly related to pre-monsoon
  Ocean/Atmospheric/Land features
• Aug-Sep rainfall driven by local
  Ocean-Atmospheric conditions
• Significant implications for long-lead Monsoon
  forecast

                          
37
Proposed Hypothesis

• Increased winter/spring wetness ? requires longer
  summer heating to set up adequate land-ocean
  gradient ? delayed monsoon cycle? reduced early
  Monsoon rainfall.
• Large scale Ocean-Atmosphere conditions in Winter
  as main drivers.

38
The Research Project

• Precipitation diagnostics
• Streamflow diagnostics
• Water management issues
• Incorporating forecasts into water management

39
Water Management IssuesBasin Selection

• Significant summer streamflow component
• Affected by large-scale and/or local-scale
  climate drivers (this is important for
  forecasting)
• Water management issues impacted by summertime
  streamflow (e.g., irrigation, MI, hydropower,
  environmental needs)
• Policies or operations that rely on or could
  benefit from knowledge of the summer hydroclimate
• Natural flow data available, either from HCDN
  data set or computed
• Ideally, decision support tool already built and
  in use

                          
40
Water Management IssuesGila River
Gila River Basin Arizona and New Mexico

                          
41
Water Management IssuesGila Basin

• Inadequate surface water supplies to meet
  irrigation, grazing, and mining demands
• Conjunctive use between surface water and
  groundwater resources
• Water quality problems due to excessive
  turbidity, bacteria, total dissolved solids,
  ammonia and acid mine drainage
• some stretches of the river not useful for
  irrigation
• Planning reservoir releases and diversions (4
  major dams) to meet demands
• Increase/ decrease in demands depending on
  summertime precipitation
• E.g., more precip ? decreased demand ? lower
  priority water user getting water (this can
  affect planting)

                          
42
Water Management IssuesExpected Outcomes

• Detailed investigation of the Gila River basin
• Key management issues (both supply and demand)
• Operations and policies
• Decision calendar timeframe of when decisions
  are made about reservoir releases and diversions
• Identify attributes of the hydroclimate that need
  to be predicted for improved water resources
  management
• Forecast variable (precipitation or streamflow)
• timing of the forecasted variable (spring values,
  summer values, or both),
• the forecast issue date
• amount to be forecasted (seasonal, monthly, etc.)

                          
43

• Preliminary Streamflow Analysis

44
Water Management IssuesGila River

• Significant summer streamflow component
• 25 of annual flow comes in July-October

                          
45
Streamflow DiagnosticsVolume Analysis (summer)

• Gila River near Red Rock, NM
• July decreasing trend
• Aug, Sep, Oct
• increasing trend

                          
46
Streamflow DiagnosticsVolume Analysis (summer)

• San Francisco River at Clifton, AZ
• July, Aug decreasing trend
• Sep, Oct increasing trend

                          
47
Streamflow DiagnosticsWinter/Spring Flow

• Winter/Spring Flows strongly related to
  winter/spring
• Pacific SST and Z500
• (ENSO/PNA patterns)

                          
48
Gila River -- Antecedent Flow Relationship

• High Spring flows ? Low Summer flows
• Consistent with Precipitation results

49
Streamflow DiagnosticsPrecipitation
Streamflow Relationship

• Precipitation streamflow relationship is
  non-linear
• High rainfall ? very little infiltration ? high
  streamflow
• Streamflow can be forecast from precipitation.

50
Streamflow DiagnosticsMethodology

• Streamflow stations with significant summer
  component (due to monsoon rains) approx. 40 in
  the region
• Timing analysis
• Trends in initiation, peak, and recedence of
  summer streamflow
• Volume analysis
• Trends in the summer and spring streamflow volume
• Determine the dominant modes of timing and volume
  variability
• Using PCA
• Identify the land/ocean/atmospheric forcings that
  drive the streamflow variability
• correlate antecedent conditions with the leading
  modes
• Determine the relationship between spring and
  summer streamflow precipitation and streamflow
  in the monsoon season and the role of subsurface
  flow
• See how well the Precipitation Hypothesis
  holds with streamflow
• Implications to Water Resources Management

                          
51
Summary and Conclusions

• Antecedent (winter/spring) Pacific
  Ocean-Atmospheric conditions and the continental
  (Western US) land conditions have a substantial
  influence on the following summer monsoon cycle
  and rainfall amount.
• Streamflows in the region too exhibit similar
  connection
• Enhanced prospects of long lead forecast of the
  monsoon hydroclimatology (i.e., timing, rainfall
  amount and streamflow)

                          
52
Future Work

• Further understanding the physical mechanisms of
  the proposed hypothesis via modeling experiments.
• Develop hydroclimate forecasting framework
  incorporating the large-scale climate
  information.
• Evaluate the utility in a water management
  context.

                          
53
Acknowledgements

• Funding provided by NOAA/GAPP (GEWEX Americas
  Prediction Project)
• Grantz, K., B. Rajagopalan, M. Clark, and E.
  Zagona, Spatio-Temporal Variability of the North
  American Monsoon (submitted), Journal of Climate,
  Special issue on the North American Monsoon,
  2005.
• http//civil.colorado.edu/balajir/ ?
  publications

54
Questions / Comments ?
55
Precipitation DiagnosticsRainfall Amount PCA

• Percent of total variance captured by each
  leading PC of monsoonal precipitation in varying
  months and regions