An extended multivariate framework for drought monitoring in MexicoRoberto A. Real-Rangel(1,2), Adrián Pedrozo-Acuña(1), J. Agustín Breña-Naranjo(1), and Víctor H. Alcocer-Yamanaka(3)
(1) Institute of Engineering, National Autonomous University of Mexico, Mexico City, Mexico ([email protected])(2) Program of Master and Doctorate in Civil Engineering, National Autonomous University of Mexico, Mexico City, Mexico(3) National Water Commission
Introduction/Background
Conclusions
References
• A reliable drought monitoring system allows to
identify regions affected by these phenomena so that
early response measures can be implemented
(Wilhite, 2000).
• Drought monitoring systems around the world have
taken advantage of global hydroclimatological
datasets derived from remote sense tools (Hao et al.,
2014; Sepulcre-Canto et al., 2012; Sheffield et al.,
2014; Svoboda et al., 2002). However, its use in
Mexico for drought assessment is still incipient (e. g.,
de Jesús, 2016).
• Mexico has the Mexico’s Drought Monitor (MDM),
derived from the North American Drought Monitor
(NADM), since 2014. Although it has inherited several
strengths from the NADM, its main limitation is the
scarcity of ground-based data, as well as the
subjective criteria to represent the spatial extent of
droughts.
• MDM has failed to detect some past events.
AghaKouchak (2015). DOI: 10.1016/j.advwatres.2014.11.012
AghaKouchak and Nakhjiri (2012). DOI: 10.1088/1748-9326/7/4/044037
de Jesús et al. (2016). DOI: 10.3390/w8080325
Dracup et al. (1980). DOI: 10.1029/WR016i002p00297
Hao and AghaKouchak (2014). DOI: 10.1175/JHM-D-12-0160.1
Hao et al. (2014). DOI: 10.1038/sdata.2014.1
Kao and Govindaraju (2010). DOI: 10.1016/j.jhydrol.2009.10.029
McKee et al. (1993). URL: http://ccc.atmos.colostate.edu/relationshipofdroughtfrequency.pdf
Rienecker et al. (2011). DOI: 10.1175/JCLI-D-11-00015.1
Sepulcre-Canto et al. (2012). DOI: 10.5194/nhess-12-3519-2012
Sheffield et al. (2014). DOI: 10.1175/BAMS-D-12-00124.1
Svoboda et al. (2002). DOI: 10.1175/1520-0477(2002)083<1181:TDM>2.3.CO;2
Wilhite (2000). ISBN: 0415168333
Tim
e (
layers
, T
)
Objectives
Provide an operational framework for drought
monitoring in Mexico, based on univariate and
multivariate nonparametric standardized
indexes.
Atmospheric reanalysis Modern-Era Retrospective analysis
for Research and Applications version 2 (MERRA-2;
Rienecker et al., 2011).
Data
I = ilat: 14° ≤ lat ≤ 34°
J = jlon: −119° ≤ lon ≤ −86°
T = tmonth: jan/1980 ≤ month < presentMERRA-2
temp. coverage
Study area
Mexico
Input/Output data array used
D4 D3 D2 D1 D0 W0 W1 W2 W3 W4
Drought intensity map
Drier Wetter
M < 1 1 ≤ M < 3 3 ≤ M < 6 6 ≤ M < 9 9 ≤ M < 12 M ≥ 12
Drought magnitude map
More persistent
• The applied framework succeed in detecting
observed drought events.
• Maps of magnitude are helpful for identifying areas
with drought persistence, which can help to define
priority regions for aid relief.
• While MSDI offer insight on the whole dry spell
across the hydrological continuum, the analysis of
univariate SI allows to track its propagation.
• Further work shall aim at correcting MERRA-2 output
with ground-based data in order to improve the
results.
Methodology
MSDI has been extended to three variables
associated to droughts (precipitation, runoff and
soil moisture) in order to detect the whole dry
spell across the hydrological continuum.
Drought maps and
time series
Intensity Magnitude
MSDISSISRISPI
MERRA-2
PRECTOTLAND BASEFLOW + RUNOFF RZMC
PRECTOTLAND: Total precipitation in land
BASEFLOW: Baseflow flux
RUNOFF: Overland runoff
RZMC: Water in the root zone
SPI, SRI, SSI and MSDI: nonparametric
standardized drought indices (SI) for
precipitation, runoff and soil moisture, and
multiple variables (Hao and AghaKouchak,
2013; Farahmand and AghaKouchak, 2015;
McKee et al., 1993; Shukla and Wood,
2008).
X, Y, Z i,j = x, y, z ti,j: t ∈ T
𝑝 𝑥𝑖 =𝑖 − 0.44
𝑛 + 0.12
𝑆𝐼 = 𝜙−1 𝑝
where 𝑝 𝑥𝑖 denotes the empirical
probability; i is the rank of non-zero data in
ascending order; n is the sample size; 𝜙 is
the standard normal distribution function.
Data validation
Clim. station
Mexicali (02033)Stream gauge
La Flor (36039)
Stream gauge
Tepehuaje (24301)
Clim. station
Cañada Honda
(01004)
Clim. station
Tonalá (07168)
Clim. station
Los Ídolos (30068)
Clim. station
Callejones (06003)
Stream gauge
Bolaños (12484)
Stream gauge
San Bernardo
(09067)
Stream gauge
Jesús Carranza II (29006)
Cañada Honda Mexicali Callejones Tonalá Los Ídolos
ME
RR
A-2
Precipitation (×10-3 mm/month)RMSE=0.02
0.0
0.2
0.4
0.6
0.0
0.2
0.4
0.6
0.0
0.2
0.4
0.6
0.0
0.2
0.4
0.6
0.0
0.2
0.4
0.6
0.0
0.2
0.4
0.6
San Bernardo Bolaños Tepehuaje Jesús Carranza II La Flor
ME
RR
A-2
Runoff (×10-3 millions of m3/month)
0.00.20.40.60.81.0
0.0
0.2
0.4
0.6
0.8
1.0
Mill
are
s
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0
0.2
0.4
0.6
0.8
1.0
Results
Sample event: Michoacán 2015-present
-2.0
-1.0
0.0
1.0
2.0
01/1
0
07/1
0
01/1
1
07/1
1
01/1
2
07/1
2
01/1
3
07/1
3
01/1
4
SPI01
-2.0
-1.0
0.0
1.0
2.0
01/1
0
07/1
0
01/1
1
07/1
1
01/1
2
07/1
2
01/1
3
07/1
3
01/1
4
SRI01
SPI01
SRI01
-2.0
-1.0
0.0
1.0
2.0
01/1
0
07/1
0
01/1
1
07/1
1
01/1
2
07/1
2
01/1
3
07/1
3
01/1
4
SSI01
SSI01
-2.0
-1.0
0.0
1.0
2.0
01/1
0
07/1
0
01/1
1
07/1
1
01/1
2
07/1
2
01/1
3
07/1
3
01/1
4
MSDI01
MSDI01
February/2017 February/2017
Newspaper (March 15, 2017)
“Drought in Cuitzeo Lake is due to natural causes:
National Water Commission”
http://primeraplananoticias.mx
-2
-1
0
1
2
01/15 07/15 01/16 07/16 01/17
p0
p25
p50
p75
p100
Drought intensity time series (MSDI01)
Michoacán Michoacán
Michoacán
MDM drought map
NRMSE=19.3%RMSE=0.03
NRMSE=6.0%RMSE=0.05
NRMSE=7.6%RMSE=0.07
NRMSE=8.6%RMSE=0.07
NRMSE=15.3%
RMSE=0.08NRMSE=12.0%
RMSE=1.56NRMSE=110.3%
RMSE=0.05NRMSE=7.3%
RMSE=0.37NRMSE=18.6%
RMSE=0.12NRMSE=43.4%