Trabalho de Revisão - Exemplo
Trabalho de Revisão - Exemplo
Trabalho de Revisão - Exemplo
Abstract: In this paper, we compared the concept of agricultural drought and its relationship
with other types of droughts and reviewed the progress of research on agricultural drought
monitoring indices on the basis of station data and remote sensing. Applicability and limita-
tions of different drought monitoring indices were also compared. Meanwhile, development
history and the latest progress in agricultural drought monitoring were evaluated through
statistics and document comparison, suggesting a transformation in agricultural drought
monitoring from traditional single meteorological monitoring indices to meteorology and re-
mote sensing-integrated monitoring indices. Finally, an analysis of current challenges in ag-
ricultural drought monitoring revealed future research prospects for agricultural drought
monitoring, such as investigating the mechanism underlying agricultural drought, identifying
factors that influence agricultural drought, developing multi-spatiotemporal scales models for
agricultural drought monitoring, coupling qualitative and quantitative agricultural drought
evaluation models, and improving the application levels of remote sensing data in agricultural
drought monitoring.
Keywords: agricultural drought; drought monitoring; research progress; research prospect
1 Introduction
Global climate change, which is both the biggest problem and the most complicated chal-
lenge faced by human beings, has attracted the attention of the public and governments
worldwide (Ye, 1992). The frequency and intensity of extreme climate events like drought
have increased significantly since the 1970s. Since extreme climate events tend to be more
www.geogsci.com www.springerlink.com/content/1009-637x
LIU Xianfeng et al.: Agricultural drought monitoring: Progress, challenges, and prospects 751
abnormal, unexpected, unpredictable, and sensitive to climate change, they are considered
the main source of terrestrial ecosystem instability and have a substantial impact on sus-
tainable development of both ecosystems and human economy (Liu et al., 2015; Stocker et
al., 2013). Moreover, according to the earth system model, the risk of global drought will
further increase in the 21st century (Dai, 2011). Therefore, how to react to and reduce
drought and its impact has become an urgent scientific issue. Among the various adverse
effects of drought, its influence on agriculture is most significant and direct. According to
statistics, economic loss due to global meteorological disasters accounts for 85% of that due
to all natural disasters, and economic loss due to drought accounts for more than 50% of that
due to all meteorological disasters. Agriculture, which is related to national food security
and social stability, is severely constrained by the climate and weather (Dai, 2012). There-
fore, research on agricultural drought has become the focus of governments and scholars
worldwide.
Past experiences in dealing with major natural disasters have shown that risk aversion is
more rewarding than disaster relief with respect to reducing hazard risks (Wu et al., 2015),
and the 3rd UN World Conference on Disaster Reduction also emphasized the importance
and urgency of disaster monitoring and loss preventing when establishing the goals and pri-
orities of disaster reduction. Although agricultural drought monitoring is important for re-
ducing disaster loss and impact, it is still poorly understood. Organizations and scholars
worldwide have performed a series of fruitful research projects on the cause of drought and
monitoring methods and influence of agricultural drought. For example, the Group on Earth
Observations (GEO) has developed the Global Agricultural Monitoring (GLAM) initiative to
evaluate agricultural drought monitoring (Fan et al., 2014). Some scholars have also sum-
marized the concept of drought and its monitoring methods and developing trends system-
atically and comprehensively (AghaKouchak et al., 2015; Chen et al., 2009; Heim, 2002;
Mishra et al., 2010; Wang et al., 2005; Zhang et al., 2011; Zhang et al., 2014). Since agri-
cultural drought is associated with various subjects like agriculture, meteorology, hydrology,
and plant physiology and it is an interaction field for natural systems and artificial systems,
research on agricultural drought monitoring is faced with many difficulties both theoretically
and technically (Li et al., 2012). Previous reports have mostly summarized agricultural
drought studies with respect to drought indices, but a comprehensive understanding of agri-
cultural drought monitoring field remained unclear. Therefore, it is necessary to sort out the
monitoring methods and development history of agricultural drought.
Since agriculture provides the foundation of China’s economy and climate change will
directly influence our food security and sustainable development (Zhao et al., 2010), break-
throughs are needed in the study of agricultural drought, both theoretically and technically,
to help us deal with the negative influence of complicated climate change on agricultural
production. In particular, summarizing the current status of agricultural drought monitoring
is an important step for further theoretical studies and evaluation of novel methods. There-
fore, the aim of this study was to summarize the concept, monitoring indices for agricultural
drought, and review the development history and recent progress of agricultural drought
monitoring through document statistical analysis and citation of important papers. Further-
more, challenges and weaknesses of previous studies and the future prospects have both
been discussed to meet the practical need of the government to monitor agricultural drought.
752 Journal of Geographical Sciences
drought on the socio-economic system (Christopher et al., 2012; Zhang et al., 2014).
Drought events are normally characterized by drought indices because the phenomenon is
very complicated and the time, development process, and scope of influence are difficult to
observe directly (Dai, 2011; Heim et al., 2002; Solomon et al., 2007). Currently, there are
hundreds of indices to characterize drought, and they can be divided into site-based and re-
mote sensing-based indices (Heim et al., 2002). Site-based indices include the standardized
precipitation index (SPI), Palmer drought severity index (PDSI), and crop moisture index
(CMI). Remote sensing-based indices are divided into indices based on bare surface (in-
cluding thermal inertia and microwave moisture inversion) and indices based on vegetation
cover. Vegetation-based drought monitoring indices can be further classified into crop mor-
phological indices (like condition vegetation index and normalized vegetation index), crop
physiological indices (like canopy temperature and canopy moisture content), and crop
comprehensive indices (like vegetation supply water index and temperature-vegetation-
drought index) (Figure 2 and Table 1).
Proposed
Indices Main author Indices description
time
Drought occurs when precipitation during 21 days or a longer
PA 1906 Henry (1906)
period is equal to or less than 30% of the normal precipitation.
Water deficit of actual water supply continues to be less than the
PDSI 1965 Palmer (1965)
local climate water supply in a period.
Mainly used for agricultural drought monitoring and analyzing
CMI 1968 Palmer (1968)
conditions of crop drought on the basis of a water balance model.
Determines crop water deficit by considering the relationship
CWSI 1988 Jackson (1988) between soil moisture and farmland evapotranspiration on the
basis of the water and energy balance principle.
Assumes that rainfall conforms to Person III distribution, and
Z 1990 Me (1990)
through precipitation normalization to determine drought index.
Reflects the probability of precipitation occurring during a certain
SPI 1993 McKee (1993) period, which is suitable for monthly or even longer-scale
drought monitoring.
This index is established by a combination of the differences
between air and land surface temperature and vegetation index,
WDI 1994 Moran (1994) considering the nearly linear relationship between vegetation
cover and the most theoretical parameter in the crop water stress
index.
Overcomes the shortage of anomaly vegetation and normalized
VCI 1995 Kogan (1996) vegetation index, which can effectively monitor the spatiotempo-
ral variation in drought.
This index can effectively detect water content in vegetation
NDWI 1996 Gao (1996) canopy and respond promptly when vegetation undergoes water
stress by introducing shortwave infrared bands.
Integrates the standardized precipitation index and relative hu-
CI 1998 Zhang (1998) midity index, which is suitable for near real-time and historical
meteorological drought.
Characterizes crop water stress through the dry and wet equation
TVDI 2002 Sandholt (2002)
determined by vegetation cover and surface temperature.
This index, combined with the land surface temperature index
VSWI 2004 Haboudane
and vegetation index, is used for agricultural drought monitoring.
This index is self-calibrated PDSI, which can determine model
SC-PDSI 2004 Wells (2004)
calibration parameters according to local climate characteristics.
This index, used for meteorological and agricultural droughts, is
K 2007 Wang(2007) defined as the ratio of the relative variability in seasonal rainfall
and relative variability in evaporation.
This is a synthesized drought index that includes information on
VegDRI 2008 Brown (2008) vegetation, meteorology, and soil water capacity by using data
mining.
regions. However, PDSI still had limitations because of the fixed time scale. In terms of ag-
ricultural drought monitoring, Shafer (1982) and Jackson (1988) proposed the surface water
supply index (SWSI) and crop water stress index (CWSI), respectively, after comprehensive
consideration of surface water supply and crop water demand, and they both worked well. In
1993, McKee (1993) found that the observed precipitation has a skewed distribution rather
than a normal distribution and proposed SPI; SPI has become one of the most popular indi-
ces because the calculation is simple and can achieve multi-scale monitoring for different
LIU Xianfeng et al.: Agricultural drought monitoring: Progress, challenges, and prospects 755
types of drought. However, since it only considers precipitation and neglects the influence of
evaporation on drought, the method was also incomplete. To integrate the effects of precipi-
tation and evaporation on drought, Vicente-Serrano (2010) established the standardized pre-
cipitation and evaporation index (SPEI), which had the multi-scale advantage of SPI and the
advantage of considering evaporation of PDSI. SPEI has become one of the ideal drought
monitoring tools. Then, Vicente-Serrano (2012) compared the performances of SPI, SPEI,
and PDSI with respect to global drought monitoring and found that SPI and SPEI were bet-
ter than PDSI for hydrological and agricultural drought monitoring and that SPEI was ex-
cellent for monitoring summer drought. A recent study established the standardized relative
humidity index (SRHI) by applying relative humidity data, and this index can detect the on-
set start of a drought earlier than SPI and is considered an ideal index for drought warning
(Farahmand et al., 2015).
In China, scholars made considerable efforts to build drought monitoring indices con-
tinuously and have tried to integrate various meteorological indices to improve monitoring
ability. Zhang (1998) proposed the comprehensive index (CI), which made weighted sum-
mation of the standardized precipitation index and relative humidity index, and this index is
widely used for drought monitoring in meteorological departments of China. Wang (2007)
proposed the K index, which is defined as the ratio of the relative variability in seasonal
precipitation and the relative variability in evaporation; it is suitable for monitoring mete-
orological and agricultural droughts. In summary, the above review indicated that site-based
drought monitoring indices have been developed over a long period, and it has become the
main method for drought monitoring. In terms of data source, site-based indices mainly rely
on the data records of meteorological stations. However, uncertainties still exist with respect
to observed datasets, including uneven distribution in space and inconformity in time-series
induced by site migration. Although a series of methods have been implemented to enhance
observation station network density, such as addition of automatic weather stations, and de-
velop data homogenization methods to correct abnormal sequences caused by non-climatic
factors, there are few data records of new time series data and a shortage of stations in key
ecological regions, especially in agricultural ecosystems.
2.2.2 Remote sensing-based drought monitoring
Since agricultural drought is closely connected to soil moisture and crop water deficit, re-
mote sensing of water inversion in soil and vegetation is an effective way for large-scale
agricultural monitoring. Data assimilation methods are generally used to estimate soil mois-
ture (Kumar et al., 2014). Among these methods, thermal inertia models of different soil
textures, established by Chen (1999), improved the accuracy of water inversion by intro-
ducing parameters of topography and wind field; however, the parameters are difficult to
determine in practice. Then, Zhang (2001) integrated the thermal inertia model, heat balance
model, and temperature difference model by using temperature differences of the soil and
leaves facing the sun and away from the sun, developing a new method of soil moisture in-
version on the basis of multi-angled remote sensing data. In addition, by applying the im-
proved IEM model, Rajat Bindlish (2006) obtained an inversion result whose correlation
with actual soil moisture was 0.95. Although microwave remote sensing can work continu-
ously without being influenced by clouds, it is only capable of inverting the soil moisture of
the surface (2–5 cm), while crop roots are usually 10–20 cm under the surface. Hence, water
756 Journal of Geographical Sciences
stress in crops cannot be examined accurately, and the result is highly uncertain (Chen et al.,
2012). However, accurate estimation of soil moisture at different depths is very important,
since it is a key parameter for agricultural drought monitoring. Therefore, despite the limita-
tions of the applying remote sensing method to agricultural monitoring, further studies are
still required; the microwave inversion results should be coupled with the terrestrial surface
model, and field survey data should be collected to increase inversion accuracy and depth
(AghaKouchak et al., 2015).
With respect to crop water deficit, CWSI was developed by analyzing the empirical rela-
tionship between air vapor and temperature differences of the canopy and air (Idso et al.,
1981). Later, Moran (1994) developed the water deficit index (WDI) on the basis of the
two-layer model of the energy balance model, while Gao (1996) proposed the normalized
difference of water index (NDWI). To eliminate the influence of both Normalized Difference
Vegetation Index (NDVI) spatial variation and other parameters of geography and ecosystem,
Kogan (1995) proposed the vegetation condition index (VCI) for drought monitoring, and
then Wang (2003) developed the vegetation temperature condition index (VTCI) in 2003. On
basis of this, Kogan proposed the vegetation health index (VHI) by linear integration of TCI
and VCI (Boken et al., 2004; Kogan et al., 2013), which was proven to be effective in re-
flecting the drought situation of crops (Mu et al., 2007). In 2004, Haboudane proposed the
vegetation supply water index (VSWI), which is a relatively simple synthesis index for ve-
getation and temperature (Haboudane et al., 2004). Previous studies have shown that VSWI
is appropriate for regions with high vegetation coverage, and it is widely applied in practice
(Liu et al., 1998). Sandholt (2002) proposed the temperature vegetation drought index
(TVDI) to estimate soil moisture on the basis of the relationship between land surface
temperature (LST) and vegetation index (VI), which is an important method to reflect agri-
cultural drought conditions through soil moisture monitoring. A hypothesis of TVDI is that
NDVI is negatively correlated with LST (Karnieli et al., 2010). However, NDVI is nega-
tively correlated with LST when water is the limitation factor of vegetation growth, whereas
NDVI is positively correlated with LST when energy is the limiting factor of vegetation
growth. Moreover, TVDI can well explain regions with drought episodes, but failure in per-
formance of agricultural monitoring and earlier warning.
2.2.3 Comprehensive drought monitoring on the basis of both meteorological and remote
sensing data
With the introduction of multi-source data, considerable efforts for drought monitoring were
made to integrate meteorological and remote sensing data. By a review of recent studies on
comprehensive drought monitoring indices, Hao (2015) pointed out that the US drought
monitoring (USDM) model is a relatively successful model. However, application of USDM
at a regional scale has many uncertainties because of its limited spatial resolution. Based on
the classification and regression model, Brown (2008) proposed the vegetation drought re-
sponse index (VegDRI) by combining meteorological drought indices (SPI and PDSI), the
vegetation index, and DEM. The index can provide near real country-scale drought informa-
tion and has become a model for comprehensive drought monitoring indices. Wu established
an integrated drought monitoring model for different growth stage of crops in China (Wu et
al., 2013; Wu et al., 2015). Du (2013) developed a synthesis drought index by using TRMM
precipitation data, LST data, and VI data in Shandong Province, and it has achieved good
LIU Xianfeng et al.: Agricultural drought monitoring: Progress, challenges, and prospects 757
results. By linear combination, Rhee (2010) proposed a drought monitoring index suitable
for arid and humid regions by using a linear combination of LST, NDVI, and TRMM data-
sets, whereas Zhang (2013) established a meteorological drought index based on satel-
lite-derived precipitation, AMSER-E soil moisture, and NDVI data. In addition, Mu (2013)
developed a satellite-based near real drought severity index (DSI) by using land surface
evapotranspiration and VI data, which could successfully detect the drought episode in Eu-
rope in 2003 and the Amazon drought in 2005 and 2010. Moreover, estimation by DSI is
highly correlated with the results of PDSI at a site scale. By comparing different drought
indices, Hao and AghaKouchak (2014) proposed the multi-variable drought index (MSDI),
which combined precipitation and soil moisture data and is considered a valid drought
monitoring index. In addition, AghaKouchak (2015) modified MSDI through introduced
ensemble runoff prediction and carried out drought monitoring and early warning in East
Africa. Deepthi (2015) proposed the multi-variable drought index (MDI), which compre-
hensively considers precipitation, runoff, evapotranspiration, and soil moisture and can si-
multaneously monitor meteorological, agricultural, and hydrological droughts. In addition,
scholars recently attempted to assimilate historical data and real-time data to establish a data
drive for near real drought monitoring indices (Shah and Mishra, 2015). On the basis of the
aforementioned analysis, scholars have made considerable efforts to establish comprehen-
sive drought monitoring indices and obtained valuable results. It should be noted that al-
though several comprehensive drought indices have been proposed, the research is still in
the infant stage. Moreover, multi-spatiotemporal scale drought monitoring indices are still
required, and comprehensive drought monitoring indices have limitations. For example, al-
though the linear combination method integrated multiple factors, it is difficult to explain
the physical implication, whereas the copula function can obtain probability, facilitating risk
analysis; however, it only considers the statistical characteristic of data and lacks a descrip-
tion of the physical process.
ecological environment are most heavily and directly impacted by drought, and they have
attracted worldwide attention (Figure 4).
Figure 3 Statistics of issued and citation literatures in agricultural drought monitoring during 1990–2014
assessment. Rhee (2010) and Zhang (2013) discussed the performance of drought monitor-
ing from different linear combinations of precipitation, vegetation, and LST in meteorologi-
cal and agricultural droughts. Anderson (2012) proposed a drought monitoring method by
combining multiple soil moisture datasets through the triple collocation analysis method,
while Hao (2013) established multi-variable drought monitoring indices by using precipita-
tion and soil moisture data. Third, with respect to the data collection method, traditional
ground-based observation cannot reflect spatial characteristics because of its uneven distri-
bution. A combination of gauged data and remote sensing data has become popular in recent
studies, especially for remote sensing-based drought monitoring, because of the increase in
different types of sensors and enhancement in spatial and temporal resolution. However,
relatively short data records have limitations. Thus, we should highlight the construction of
long-term surface parameter products by assimilation of multi-source datasets. In addition,
development of satellite-based drought monitoring models are needed to enhance our ability
to forecast drought, for example, the microwave remote sensed drought model (Kongoli et
al., 2012; Rott et al., 2010). Fourth, in terms of drought monitoring products, several
global-scale products were widely used, including the PDSI product developed by Dai
(2004), SPEI products with a 0.5°×0.5° resolution developed by Vicente-Serrano (2010), and
remote sensed global terrestrial drought severity index products developed by Mu (2013).
Recently, Hao et al. (2014) released the global integrated drought monitoring and prediction
system (GIDMaPS), including the near real-time drought monitoring model and seasonal
prediction model, and it can provide global-scale meteorological drought and agricultural
drought products. Together, the aforementioned drought products provide data support for
global drought monitoring and assessment.
the main development directions for the future (Figure 5). In theory, we need to further un-
derstand the mechanism and process of agricultural drought and identify the influencing
factors and feedback mechanism to integrate multiple influencing factors and establish a
synthesized drought monitoring model. We also need to build drought monitoring and
warning models at multi-spatiotemporal scales, couple the drought monitoring model and
crop growth model to bridge the qualitative description to semi-quantitative and even quan-
titative description for agricultural drought monitoring, and establish a new drought index
that can monitor drought as well as predict crop failure. In practice, we need to use
multi-source remote sensing data to a greater extent to facilitate agricultural drought moni-
toring.
Soil moisture, a key parameter that integrates the responses of climate, soil, and vegetation
to water balance and the influence of water balance on vegetation dynamics, plays a very
important role in the terrestrial water cycle. Agricultural drought reflects the extent to which
soil moisture is lower than crop demand water, resulting in crop wilting and even failure.
Since the tolerance to soil water deficit is different among crops and regions, time lags to
precipitation shortage and soil moisture deficit also differ, and the difference varies with the
growing stage as well. With an increase in water shortage, evaporation decreases and the
surface temperature increases, forming a positive feedback process. The precipitation model
that neglects the abovementioned process cannot explain agricultural drought monitoring
with respect to crop water balance. Currently, the comprehensive response process of crops
to precipitation and temperature is still unclear, and we should further understand the crop
drought mechanism. We should also make better use of the time lag between water defi-
ciency and crop drought stress for early warning of drought.
LIU Xianfeng et al.: Agricultural drought monitoring: Progress, challenges, and prospects 761
In addition, agronomic parameters are the basic parameters for describing crop growth,
and they are, therefore, an important aspect of research on the agricultural drought mecha-
nism. The crop reflectance spectrum is significantly influenced by agronomic parameters,
and the red edge, which is closely related to chlorophyll content, is one of the most impor-
tant parameters. When crops are under water stress, the parameters values will change and
further result in the displacement of the red edge. Therefore, analyzing variations in the crop
reflectance spectrum by hyperspectral remote sensing provides a new angle for research on
the agricultural drought mechanism. Since monitoring models can hardly detect the start and
end time of agricultural drought, studies on the mechanism may be a key to solving the
problem. Upon understanding the agricultural drought mechanism, we can predict the onset
of drought in time and establish an agricultural drought remote sensing model on the basis of
crop physiological and ecological characteristics, facilitating prompt measures for reducing
drought-related losses and influences.
3.2 Identify factors that influence agricultural drought
Agricultural drought is caused by an abnormal decrease in precipitation, and it is influenced
comprehensively by land surface temperature, evapotranspiration, soil properties, and
physiological or ecological characteristics of the crop. We should pay more attention to fac-
tors closely related to agricultural drought in the future, understand the interaction mecha-
nism, screen the key influencing factors to cause variations in drought, and create agricul-
tural drought monitoring models based on multiple geological factors. In practice, since sat-
ellite remote sensing can acquire precise spatial and temporal information on the land sur-
face and at a large scale, it is widely used in agricultural drought monitoring. It is considered
an ideal data source because it provides not only information on the land surface environ-
ment but also information on crop growth, like vegetation index, land surface temperature,
and precipitation. By using remote sensing as a data source, we can integrate soil character-
istics and topography features and add both environment and crop information in the agri-
cultural monitoring model to achieve the integration of multiple geological factors for
drought monitoring.
Currently, some countries have already developed agricultural drought monitoring sys-
tems at country or regional scales, and satellite data-driven comprehensive models are being
established (Brown et al., 2008). However, integration of remote sensing data with mete-
orological data obtained by stations has not been achieved. It should be noted that even
though remote sensing data can be an effective supplement for meteorological data, because
they cover large regions at a high frequency, they cannot fully replace the latter. The reason
is that although land surface observation stations are few and unevenly distributed, the data
are precise with a long time series and the most important data source for remote sensing
data validation. Therefore, it is important to build a bridge between remote sensing moni-
toring models and meteorological observation models, integrating the advantages of both
model types. To achieve this goal, technologies and methods still need improvement. For
example, we should build a more complete ground observation network, develop a perfect
coupling method, and determine a uniform influencing factor framework.
3.3 Expansion of the spatiotemporal scale of agricultural drought monitoring models
The concept of scale, both temporal and spatial scales, is frequently mentioned. For agricul-
762 Journal of Geographical Sciences
tural drought monitoring, there are intrinsic differences among the results at different spatial
and temporal scales. For example, the monitoring result of months may be different from the
result of years, and at a different spatial scale, the result may also vary with the input data.
Previous studies have tried to create agricultural drought monitoring models at different
temporal and spatial scales and achieved some models with good validation results at a re-
gional scale (Du et al., 2013; Rhee et al., 2010). However, most studies were limited to spe-
cific regions and time scales, and the scale of the models should be further expanded.
Therefore, how to integrate the existing agricultural drought monitoring models and develop
new multi-scale agricultural drought monitoring models will be a key question for estab-
lishing a drought monitoring platform with multi-spatiotemporal scales in the future.
Agricultural drought monitoring provides information on crop growth and production
worldwide, and it facilitates crop production prediction as well. However, since there are
differences in the developing stage, technology level, and disaster reduction ability among
different countries, the request for agricultural drought monitoring information also varies.
Therefore, building an agricultural drought monitoring platform at multi-spatiotemporal
scales not only meets the demand of different countries but is also significant for reducing
the influence of drought and improving the ability to cope it worldwide. More specifically,
(1) at temporal scales, monitoring models should be able to manage early warning at 10 days,
a month, a season, a year, and a decade in regular situations, whereas during emergencies,
the models should be able to monitor near real-time early warning. (2) At spatial scales, the
model should satisfy the needs of the world, different continents, different countries, and
different regions. How to achieve a monitoring platform at multi-spatiotemporal scales is a
key issue for future research on agricultural drought monitoring.
3.4 Coupling qualitative and quantitative models for agricultural drought monitoring
Qualitative description accounts for the most in the results of existing agricultural drought
monitoring model, and the lack of quantitative monitoring and warning assessment limits the
association between monitoring results and practical loss assessment. Although technologi-
cal methods for drought monitoring are developing fast, attention to methods for qualitative
assessment of the results is still lacking. With the rapid development of numerical simula-
tions and further understanding of the drought mechanism of crops, the crop growth model
has made great progress with respect to quantitative assessment of agro-meteorological haz-
ards, and it is considered the core step to push agricultural drought monitoring into the stage
for quantitative assessment. The advantage of the crop growth model in agro-meteorological
hazard assessment is that the mechanism is clear and it can reflect the active relationship
between the growing process; crop production; and temperature, precipitation, and soil
moisture at every growth stage. Therefore, we should use functions of the crop growth
model when creating agricultural drought monitoring models in the future.
In terms of technology, the spatial crop vulnerable model can be drawn from station-scale
vulnerable models through the scale upscaling algorithm, which is a promising method for
coupling an agricultural drought monitoring model at a large scale and the crop growth
model; it will become the core and critical point of agricultural drought monitoring. On the
basis of spatial information, especially the advantage of remote sensing technology, and by
improving the integration of indices for crop physiology, morphology, and soil moisture, we
LIU Xianfeng et al.: Agricultural drought monitoring: Progress, challenges, and prospects 763
can achieve the coupling of an agricultural drought monitoring model at a large scale and the
crop growth model, and create a remote sensing-based agricultural drought monitoring plat-
form with a clear physical mechanism, multiple factors, and progress and at multiple scales.
This is an ideal way for realizing the goal of reducing agricultural hazard loss to the largest
extent.
Remote sensing is a very important data source for agricultural drought monitoring, and with
launching of satellites for different uses, abundant remote sensing data are acquired by
scholars to analyze land surface processes. Compared with ground observation data, the
biggest challenge for remote sensing data is the short time sequence, which can barely pre-
sent the variation in drought at a long time scale. The diversified temporal and spatial scales
of different data sources also limit the comprehensive use of the data. Although we have
developed transforming technology for multi-scale remote sensing data, comprehensive ap-
plications have not been completely achieved, especially with respect to applications of mi-
crowave remote sensing data for the influence of drought on vegetation (Andela et al., 2013).
In future, we should enhance the use of remote sensing in drought monitoring by exploring
more land surface parameters and increasing the applications of remote sensing data (Rodell,
2012).Uncertainties of remote sensing data are a key problem with respect to their applica-
tions, like the difference in uniformity caused by the change in sensors. Therefore, to de-
velop data assimilation technology, upgrading the comprehensive applications and quantita-
tive estimation of the uncertainties of remote sensing data are important. Finally, successful
launches of new types of satellites such as the Soil Moisture Active Passive is offering new
opportunities for agricultural drought monitoring, hence increasing the importance of remote
sensing data in agricultural drought monitoring (Figure 6).
References
AghaKouchak A, 2015a. A multivariate approach for persistence-based drought prediction: Application to the
764 Journal of Geographical Sciences
Hao Z, AghaKouchak A, 2013. Multivariate Standardized Drought Index: A parametric multi-index model. Ad-
vances in Water Resources, 57: 12–18.
Hao Z, AghaKouchak A, 2014. A Nonparametric Multivariate Multi-Index Drought Monitoring Framework.
Journal of Hydrometeorology, 15(1): 89–101.
Hao Z, AghaKouchak A, Nakhjiri N et al., 2014. Global integrated drought monitoring and prediction system.
Scientific Data, 1.
Hao Z, Singh V P, 2015. Drought characterization from a multivariate perspective: A review. Journal of Hydrol-
ogy, 527: 668–678.
Heddinghaus T R, Sabol P, 1991. A review of the Palmer Drought Severity Index and where do we go from here.
In: Proceedings of the seventh conference on applied climatology. American Meteorological Society Boston,
MA.
Heim R R, 2002. A review of twentieth-century drought indices used in the United States. Bulletin of the Ameri-
can Meteorological Society, 83(8): 1149.
Henry A J, 1906. The Climatology of the United States. Weather Bureau Bulletin Q, Washington D C, 51–58.
Idso S B, Jackson R D, Pinter P J et al., 1981. Normalizing the stress-degree-day parameter for environmental
variability. Agricultural Meteorology, 24: 45–55.
IPCC, 2007. Climate Change 2007: The Physical Science Basis: Summary for Policymakers. Intergovernmental
Panel on Climate Change Secretariat.
Jackson R D, Kustas W P, Choudhury B J, 1988. A reexamination of the crop water stress index. Irrigation Sci-
ence, 9(4): 309–317.
Karnieli A, Agam N, Pinker R T et al., 2010. Use of NDVI and land surface temperature for drought assessment:
Merits and limitations. Journal of Climate, 23(3): 618–633.
Kincer J B, 1919. The seasonal distribution of precipitation and its frequency and intensity in the United States.
Monthly Weather Review, 47(9): 624–631.
Kogan F, Adamenko T, Guo W, 2013. Global and regional drought dynamics in the climate warming era. Remote
Sensing Letters, 4(4): 364–372.
Kogan F N, 1995. Droughts of the late 1980s in the United Statesas derived from NOAA polar-orbiting satellite
data. Bulletin of the American Meteorological Society, 76(5): 655–668.
Kongoli C, Romanov P, Ferraro P J, 2012. Snow cover monitoring from remote sensing satellites. In: Remote
Sensing of Drought: Innovative Monitoring Approaches. CRC Press, 359–386.
Kumar S V, Peters-Lidard C D, Mocko D et al., 2014. Assimilation of remotely sensed soil moisture and snow
depth retrievals for drought estimation. Journal of Hydrometeorology, 15(6): 2446–2469.
Li Hailiang, Dai Shengpei, Luo Hongxia, 2012. Status and prospects of agricultural drought monitoring. China
Rural Technology, 5: 46–48. (in Chinese)
Liu Li, Zhou Ying, 1998. Drought monitoring based on vegetation supply water index in Guizhou Province.
Guizhou Meteorology, 22(6): 17–21. (in Chinese)
Liu Xianfeng, Zhu Xiufang, Pan Yaozhong et al., 2015. Spatiotemporal changes of cold surges in Inner Mongolia
between 1960 and 2012. Journal of Geographical Sciences, 25(3): 259–273.
Liu W T, Kogan F N, 1996. Monitoring regional drought using the vegetation condition index. International
Journal of Remote Sensing, 17(14): 2761–2782.
Marcovitch S, 1930. The measure of droughtiness. Monthly Weather Review, 58(3): 113–113.
McGuire J K, Palmer W C, 1957. The 1957 drought in the eastern United States. Monthly Weather Review, 85(9):
305–314.
McKee T B, Doesken N J, Kleist J, 1993. The relationship of drought frequency and duration to time scales. In:
Proceedings of the 8th Conference on Applied Climatology. American Meteorological Society Boston, MA,
179–183.
766 Journal of Geographical Sciences