A New Method for Drought Risk Assessment by Integrating the TRMM Monthly Rainfall Data and the Terra/MODIS NDVI Data in Fars Province, Iran

Introduction
Drought monitoring and assessment is usually done through either ground observation or
remote sensing. Due to having some limitations, gathering and analyzing ground observations
are a time-consuming and expensive way to approach a precise drought monitoring and
assessment. In contrast, remote sensing represents a fast and economic way of monitoring, but
an applicable approach needs to be developed. To this end, using satellite sensor data which are
continuously available provides cost-effective data for a better understanding of the region.
They can be used to detect the drought commencement, duration and magnitude. Tropical
Rainfall Measuring Mission monthly data (TRMM-3B43) and Monthly Normalized Difference
Vegetation Index (NDVI) data of the MODIS on Terra satellite are freely available for this
objective. The main objectives of the present study, which was carried out in the Fars Province,
Iran, were: 1. integrating the satellite data for mapping drought severity classes using the
Standardized Precipitation Index (SPI) and the NDVI anomaly maps, 2. creating drought risk
maps, 3. calculating the percentage of drought affected area by drought risk level, 4. showing
the effectiveness of satellite derived drought indices as an indicator for drought assessment, and
5. identifying the most drought vulnerable areas of the surveyed region.
􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯􀀯
∗E-mail: Erfanian.ma@gmail.com Tel: +98 9123328494
Physical Geography Research Quarterly, 46 (1), Spring 2014 17
Methodology
This research was carried out in Fars Province, Iran. It is located between 50􀃛30’ and 55􀃛36 E
longitude and from 27􀃛03’ to 31􀃛42 N latitude and cover an approximate area of 122661 km2.
This study aimed to map drought risk area in the Fars Province, by integrating the Standard
Precipitation Index (SPI) and the Normalized Difference Vegetation Index (NDVI) Anomaly
methods. As the first step, the growing season-based SPI (April- September) at 44 stations were
calculated for 2000-2008 period using the standard normal distribution. The SPI raster layer (for
each year), was created using the ordinary Kriging method. Then, all SPI maps were reclassified
into five drought severity classes. As the second step, NDVI anomaly maps were created for the
growing season based-NDVI anomaly of MODIS during the same period (9-year period). The
NDVI anomaly map in each year was reclassified into five classes in a similar way. At the next
part, for both methods, Boolean drought frequency map (presence or absence of drought)
derived for each year. The derivation of final drought risk map was done by a simple weighted
linear combination of the drought frequency maps. In this research, another drought risk map
was created by integrating the NDVI anomaly and the TRMM-based SPI maps to introduce a
new remote sensing method.
Results and Discussion
The ground-based SPI method applied for the growing seasons showed that in 2000, 2001, 2005
and 2008, some severe droughts occurred whereas the NDVI anomaly resulted in 2000, 2001
and 2008. The drought severity maps of TRMM based on SPI method indicated some noticeable
drought occurrences in the Fars Province in 2000, 2005, and 2008 as well. The comparison of
drought risk maps created by the TRMM-based SPI and the ground-based SPI methods showed
that the majority of the surveyed regions are highly prone to drought occurrence. The TRMM
could predict the monthly rainfall at most of 44 rain-gauge stations. Comparing drought risk
maps, the high and moderate risk classes in the first method contain % 59.58 and % 39.84,
while in the TRMM based method, they cover %61.1 and %37.12 of the area, respectively.
Before drought risk assessment, it is highly recommended to evaluate the TRMM data for future
events. The risk maps can be compared with the actual decrease in agricultural products for a
better understanding of the events and their verifications.
Conclusion
The method applied in this study showed that almost whole the province is prone to drought
occurrences. The northern and southern areas of the province were more susceptible to drought
with different severities during the growing seasons in 2000-2008. It is notable to express that
there are still some limitations to apply the satellite data for a long period. These might be data
availability problem with moderate spatial resolution. The TRMM and the MODIS data have
been available since 2000 and 1998, respectively. Furthermore, the TRMM data calibration and
validation is required before creating the TRMM-based SPI maps. Despite their shortages, the
application of remote sensing data for drought risk assessment can still be done as an acceptable
method in ungauged regions.