Multivariate drought monitor system: biophysical modelling, remote sensing and hydroclimatic information for drought analyisis and forecasting in agriculture

Folio: 1210526

drought

Abstract

Droughts are a recurrent phenomenon that produce severe impacts on social and ecological systems. Because droughts are complex natural phenomena, and their impacts on agriculture are enormous, early detection of droughts is crucial to mobilize resources to prevent and mitigate socioeconomic consequences. Traditional approaches for quantifying the severity of droughts, usually based on meteorological and hydrological indices, have important limitations to represent the nature of drought impacts on vegetation and agriculture. For instance, they do not account for the available soil moisture, variable which plays a key role in the behavior of vegetation, such as reductions in gross primary productivity, changes in phenology and senescence. On the other hand, model based on meteorological and hydrological indices are limited in spatial resolution compared to the variability of vegetation and agriculture. To overcome this limitation remote sensing methods provide a fairly simple and reliable mean to retrieve environmental information at different spatial scales, these indices can be related to canopy growth and vegetation productivity, becoming an adequate tool to characterize the state of vegetation and monitor drought.

The main objective of this research is to develop and test a multivariate drought monitoring/early warning system based on the combination of surface weather information, remote sensing data, soil and crop simulation models to explain the spatiotemporal evolution of drought vegetation indices derived from remote sensing methods.

Addressing our main research questions require characterizing biological and physical parameters at different scales and timing. Therefore, we will follow a multi-methodological and multi-scale approach in order to achieve the proposed objectives. At regional scale (area of 100 x 100 km2), we will estimate the spatiotemporal variability of vegetation indices using analysis of coarse-resolution satellite dataset from the Moderate Resolution Imaging Spectroradiometer (MODIS). These regional scale proxies will be integrated with precipitation and evapotranspiration time series in order to derive a model relating vegetation performance with meteorological factors controlling drought. This regional-scale model, in turns, will be calibrated with measurements of vegetation productivity and soil properties conducted over areas of approximately 250 x 250 m2 at experimental sites (irrigated crops and native forests). In the experimental sites, we will monitor: (i) available soil moisture with geophysical techniques and soil characterization, and (ii) vegetation productivity using high-resolution satellite images from the Sentinel-2 mission.

To undertake this we have assembled a multidisciplinary team combining disciplines of Agronomy, Meteorology, Geophysics and Remote Sensing. After a 4 year project we expect to generate a calibrated multivariate system for drought monitoring and forecasting that can be used operationally to detect drought changes and generate early warning systems. This method could be transferred to the national water authority that seeks for objective methods to declare emergencies and is currently revising the decrees of water scarcity.