Earth Observation-Based Operational Estimation of Soil Moisture and Evapotranspiration for Agricultural Crops in Support of Sustainable Water Management

George Petropoulos, Prashant Srivastava, Maria Piles, Simon Pearson
2018 Sustainability  
Global information on the spatio-temporal variation of parameters driving the Earth's terrestrial water and energy cycles, such as evapotranspiration (ET) rates and surface soil moisture (SSM), is of key significance. The water and energy cycles underpin global food and water security and need to be fully understood as the climate changes. In the last few decades, Earth Observation (EO) technology has played an increasingly important role in determining both ET and SSM. This paper reviews the
more » ... ate of the art in the use specifically of operational EO of both ET and SSM estimates. We discuss the key technical and operational considerations to derive accurate estimates of those parameters from space. The review suggests significant progress has been made in the recent years in retrieving ET and SSM operationally; yet, further work is required to optimize parameter accuracy and to improve the operational capability of services developed using EO data. Emerging applications on which ET/SSM operational products may be included in the context specifically in relation to agriculture are also highlighted; the operational use of those operational products in such applications remains to be seen. control on the Earth's water, carbon cycles and ecosystem functioning [8, 9] . Tools to quantify both ET and SSM have a crucial role in understanding the Earth's climatic system on the longer term but by enabling increased water use efficiency they can also have significant economic, environmental and social impact on the shorter term. For the latter, data is required at high spatial and temporal resolutions to inform decision makers on sustainable utilisation and management of water [10] . There are numerous techniques available to derive SSM and ET using ground instrumentation. Ground measurements have certain advantages, such as providing a relatively direct measurement, instrument portability, simple installation, operation and maintenance, the ability to provide measurement at the desired depth and also the relative maturity and stability of the methods. Nevertheless, such techniques have proven difficult to implement, especially over large areas. This reflects the complex, expensive and labour-intensive challenges in deploying a network of in-situ sensors at a study site. In addition, current field equipment typically determines only localized estimates of SSM which must be aggregated for measuring the parameter over large spatial scales. The placement of the in-situ sensors is then key to obtain representative estimates [11] . Various ground-based observational networks have been established in basins of different characteristics around the globe, aiming to systematically collect, archive and distribute a wide variety of relevant data for use in research activities [12, 13] . A review of the available ground-based methods including their relative strengths and limitations and available operational networks can be found in [14, 15] . Conversely, the advent of Earth Observation (EO) technology has provided novel and economically feasible means to derive temporally consistent coverage of ET and SSM at different spatial scales [16] [17] [18] . EO provides regional to global scale parameter estimation and enables non-invasive synoptic views in a spatially contiguous fashion, providing estimates from otherwise inaccessible regions. Another advantage is that observations can be obtained over remote areas which are otherwise inaccessible for in-situ measurements. In addition, the development of EO has enabled the scientific community to investigate previously unsolvable problems, such as spatial variability and scale of observation [19] . These are the main characteristics that make EO one of the most efficient and cost-effective techniques for obtaining spatio-temporal estimates of ET as well as of SSM [20, 21] . Remote sensing instruments do not directly measure either soil water content and/or the turbulent heat fluxes. The spectral measures they provide have to be combined with a physical or semi-empirical model and an inversion algorithm in order to estimate the parameters. A plethora of retrieval methods have been proposed for estimating spatially distributed ET and SSM utilising spectral information acquired in selected regions of the electromagnetic radiation spectrum; often combined with ancillary surface and atmospheric observations. Those techniques are well understood but each has atypical strengths and weaknesses, see recent reviews [20] [21] [22] . Some of those approaches have shown promising potential for providing accurate estimates of spatial-temporal estimates of surface heat fluxes and SSM, with accuracies reported in the ranges of 50 Wm −2 and of 0.04 m 3 ·m −3 respectively; accuracy levels required by many applications [23] . At present, there are several operational products available providing ET and SSM estimates, each with their own strengths and limitations. In light of the above, this paper aims at providing an overview of the state of the art of EO techniques to derive operational estimates of both ET and SSM. In this context, we critically discuss the challenges and caveats associated with the derivation of accurate estimates of ET and SSM parameters from space. We make suggestions on how future work should be directed to enhance the accuracy and operational capability of services developed using EO data. The paper closes with a reflective discussion on how existing or future EO operational products could lead to new applications related specifically to attain more sustainable agricultural practices, which is of the utmost importance since agriculture is the world's largest consumer of fresh water.
doi:10.3390/su10010181 fatcat:kott5kqlkfepnbovaualndwyce