Agricultural Drought Assessment in Latin America Based on a Standardized Soil Moisture Index
Carrao, Hugo; Russo, Simone; Sepulcre, Guadalupe; Barbosa, Paulo; Barbosa, Paulo
Drought is the most damaging environmental phenomenon. Because of its slow-onset characteristics and lack of structural impacts, drought is often disregarded unless serious problems appear. This lack of recognition compared to other natural hazards, such as floods, earthquakes or tsunamis, has been an impediment for obtaining adequate research support and, in many cases, an obstacle for building awareness among decision makers from local to regional levels. Drought is part of the natural climate variability and therefore can be observed in all climate regimes and at different spatial scales. There is no unique or universally accepted definition of drought, but three types are commonly distinguished: Meteorological drought: deficit in precipitation over a defined period and region as compared to climatological average values; Agricultural drought: impact of reduced water supply on agricultural crops, leading to a reduction of annual yields in the affected regions; Hydrological drought: impact of reduced water supply on stream-flows, lakes or reservoir levels over a defined period and region as compared to average values.
The different types of drought occur at different time scales, but are intimately interrelated with each other. The longer the rainfall deficiency is, the more likely other types of droughts (namely agricultural or hydrological) will occur as a result. In natural ecosystems, long-term dry conditions cause vegetation to be more prone to forest fires, while in human-induced ecosystems they reduce the fodder available for animals and the agricultural yield, thus leading to a reduction in income. Although poor vegetation conditions (i.e. agricultural drought) often occur during dry, hot periods of low precipitation, they can also occur during periods of average precipitation when soil conditions require extra water. Indeed, deficient topsoil moisture at planting may hinder germination, leading to low plant populations per hectare and a reduction of final yield. Traditionally, soil moisture content has been estimated through the use of hydrological models, i.e. complex representations of the hydrological cycle that take into account a few tens of input variables, such as: precipitation; infiltration; evapotranspiration; land cover; topography; etc. The output results of these models can only be really accurate if the input variables are of good quality and the parameters adequately set, but in many worldwide regions the data is not available a priori and the calibration process of the parameters is extremely complex. As a consequence, traditional methods and indices of drought monitoring are based solely on observed weather station data and used only to follow deficiencies on average meteorological conditions.
The availability of remote sensing data, covering wide regions over long periods of time, has progressively strengthened the role of indices derived from satellite images in environmental studies related to drought episodes. With the history of operational Earth Observation (EO) sensors reaching back over three decades now, it allows retrospective analysis of the state and development of ecosystems at different spatial and temporal resolutions and with different geographical coverage. Remote sensing data provide spatially continuous measurements of different variables related to drought that are periodically updated. For example, soil moisture content can now be measured reliably from space at appropriate temporal and spatial frequencies. In this sense, the Climate Change Initiative (CCI) of the European Space Agency (ESA) recently launched the global Essential Climate Variable (ECV) soil moisture data set. This set has been generated using active and passive microwave spaceborne instruments and covers the 32 year period from 1978 to 2010. The active data set was generated based on observations from the C-band scatterometers on board of ERS-1, ERS-2 and METOP-A. The passive data set was generated based on passive microwave observations from Nimbus 7 SMMR, DMSP SSM/I, TRMM TMI and Aqua AMSR-E.
In this paper we propose a Standardized Soil Moisture Index (SSMI) that is designed to be a relatively simple, spatially invariant and probabilistic year-round index applicable to soil moisture content conditions derived from satellite imagery data. The SSMI is based on soil moisture content alone and is defined as the number of standard deviations that the observed accumulated moisture at a given location and timescale deviates from the long-term normal conditions. Specifically, the SSMI is computed by fitting some probability distribution to a historical soil moisture record and then transforming it into a normal distribution with a mean of zero and standard deviation of one. Positive SSMI values indicate greater than the median precipitation, and negative values indicate less than median precipitation. The fundamental strength of the SSMI is that it can be calculated for a variety of timescales, enabling soil water supply anomalies relevant to vegetation to be readily identified and monitored.
To evaluate the applicability of the SSMI for drought assessment we compute the bias of different parametric and non-parametric probability distributions to accumulated soil moisture observations derived from the ESA CCI ECV soil moisture data set. In detail, we compute and analyzed the goodness-of-fit between the modeled cumulative distribution functions (CDF) and the empirical CDFs, which gives a measure of the discrepancy between the estimation model and the soil moisture content, as well as the uncertainties associated to drought intensities. Drought climatology is calculated on a 0.25dd grid over the Latin American region at timescales of 1, 3, 6, 12, and 24 months for the long-term periods of 1978-2010. We focus our experiences in Latin America because it has been extremely affected by drought events in the past and the climate change scenarios foresee an increased frequency of these events in the region. In addition, the extensive plains located in middle and subtropical latitudes of Latin America are within the main areas that produce cereals worldwide (e.g. Brazil and Argentina), thus constituting an excellent use case study for testing our index.