Radar Science and Engineering Section, NASA Jet Propulsion Laboratory, California Institute of Technology, Pasadena, USA; Biosystem and Agricultural Engineering, Michigan State University, East Lansing, USA.
Biosystem and Agricultural Engineering, Michigan State University, East Lansing, USA; Civil and Environmental Engineering, Michigan State University, East Lansing, USA.
Sci Total Environ. 2022 Sep 1;837:155893. doi: 10.1016/j.scitotenv.2022.155893. Epub 2022 May 12.
Effective agricultural water management requires accurate and timely identification of crop water stress at the farm-scale for irrigation advisories or to allocate the optimal amount of water for irrigation. Various drought indices are being utilized to map the water-stressed locations/farms in agricultural regions. Most of these existing drought indices provide some degree of characterization of water stress but do not adequately provide spatially resolved high-resolution (farm-scale) information for decision-making about irrigation advisories or water allocation. These existing drought indices need modeling and climatology information, hence making them data-intensive and complex to compute. Therefore, a reliable, simple, and computationally easy method without modeling to characterize the water stress at high-resolution is essential for the operational mapping of water-stressed farms in agricultural regions. The proposed new approach facilitates improved and quick decision-making without compromising much of the skills imparted by the established drought indices. This study aims to formulate a water-demand index (WDI) based on a parameter-independent data-driven approach using readily available remote sensing observations and weather data. We hypothesize that the WDI for an agricultural domain can be characterized by soil moisture, vegetative growth (NDVI), and heat unit (growing degree day, GDD). To this end, we used remote sensing-based soil moisture and NDVI and modeled ambient temperature datasets to generate weekly WDI maps at 1 km. The proposed methodology is verified over a few intensively irrigated agricultural-dominated areas with different climatic conditions. Our results suggest that the proposed approach characterizes water-stressed fields through WDI maps with good spatial representativeness. Overall, this study provides a framework to generate weekly WDI maps quickly with readily available measurements. These water-demand maps will help water resource managers to reduce dependence on established drought indices and prioritize the specific regions/fields with high water demand for optimum water allocations to improve crop health and ultimately maximize water-use efficiency.
有效的农业水资源管理需要在农场尺度上准确、及时地识别作物水分胁迫，以便提供灌溉建议或分配最佳灌溉水量。各种干旱指数正被用于绘制农业区域水分胁迫的位置/农场图。这些现有的干旱指数中的大多数都提供了一定程度的水分胁迫特征描述，但不能充分提供用于灌溉建议或水资源分配的空间分辨率高（农场尺度）信息。这些现有的干旱指数需要建模和气候学信息，因此它们计算起来数据密集且复杂。因此，需要一种可靠、简单、计算上容易的方法，无需建模即可在高分辨率下对水分胁迫进行特征描述，这对于农业区域水分胁迫农场的业务化制图至关重要。所提出的新方法有助于在不影响既定干旱指数所赋予的技能的情况下，进行改进和快速决策。本研究旨在提出一种基于无模型、参数独立的数据驱动方法的水分需求指数（WDI），该方法利用易于获得的遥感观测和气象数据。我们假设，农业领域的 WDI 可以通过土壤湿度、植被生长（NDVI）和热量单位（生长度日，GDD）来描述。为此，我们使用基于遥感的土壤湿度和 NDVI 以及建模的环境温度数据集，以 1 公里的分辨率生成每周的 WDI 图。该方法在几个具有不同气候条件的密集灌溉农业主导地区进行了验证。我们的结果表明，该方法通过具有良好空间代表性的 WDI 图来描述水分胁迫的农田。总体而言，本研究提供了一个快速生成每周 WDI 图的框架，该框架可利用现成的测量值。这些水分需求图将帮助水资源管理者减少对既定干旱指数的依赖，并优先考虑高水分需求的特定地区/农田，以实现最佳水资源分配，从而改善作物健康，最终最大限度地提高水的利用效率。
