The Brazilian Cerrado is a vital agricultural region, yet its expansion often overlooks the high climate risks posed by the prolonged dry season, particularly from June to September when water demand surges, causing low-flow conditions. Recent studies highlight significant water deficits between May and October, making irrigation water use (IWU) crucial for policymakers and managers. This study estimates IWU for dry beans grown under center pivots during the dry season in the state of Mato Grosso, Brazil, a leading agricultural producer with 309,372 km² of agricultural land. Mato Grosso relies heavily on irrigation from April to October for bean cultivation. We used NDVI time series from Sentinel-2A data from 2019 to 2023 to classify dry beans areas based on t-SNE and k-means cluster classification. The individual NDVI time series for the dry season for each pivot was divided based on the peak NDVI values to analyze phenological parameters- such as duration, start and end of the season- to assess water needs from April to September. ERA-5 Land climate data provided daily reference evapotranspiration (ETo) and precipitation (P), which were used to compute the crop’s water requirement. Irrigation depth (D) was estimated using a water balance equation incorporating crop coefficients (Kc) and daily irrigation needs adjusted for efficiency. The analysis shows that dry-season irrigation in the Cerrado primarily replenishes soil moisture, often leading to inefficient water use. From 2019–2023, IWU increased significantly, with the model showing a strong correlation (R² = 0.92) to reported accumulated irrigation depth for center pivot during the dry season. However, the model underestimated needs from May to July and overestimated in August, with a bias of −21.88 mm. The North subregion, benefiting from favorable conditions, accounted for 43 % of the state's IWU. The study provides valuable insights into IWU trends, supporting strategic decisions and resource allocation, while offering a cost-effective method for real-time IWU estimation.

中文翻译：

---

在巴西塞拉多使用 ERA5 陆地气候变量和 Sentinel 2 NDVI 时间序列评估中心枢轴上干豆的灌溉用水


巴西塞拉多是一个重要的农业区，但其扩张往往忽视了漫长的旱季带来的高气候风险，尤其是在 6 月至 9 月期间，用水需求激增，导致低流量条件。最近的研究强调，5 月至 10 月期间存在严重的缺水，这使得灌溉用水 （IWU） 对政策制定者和管理者至关重要。本研究估计了巴西马托格罗索州旱季在中心支轴下种植的干豆的 IWU，该州是拥有 309,372 平方公里农业用地的领先农业生产国。马托格罗索州在 4 月至 10 月期间严重依赖灌溉来种植豆类。我们使用 2019 年至 2023 年 Sentinel-2A 数据的 NDVI 时间序列，根据 t-SNE 和 k-means 聚类分类对干豆区域进行分类。根据峰值 NDVI 值划分每个枢轴的旱季的单个 NDVI 时间序列，以分析物候参数（例如持续时间、季节的开始和结束），以评估 4 月至 9 月的用水需求。ERA-5 土地气候数据提供了每日参考蒸散量 （ETo） 和降水 （P），用于计算作物的需水量。灌溉深度 （D） 是使用水平衡方程估算的，该方程结合了作物系数 （Kc） 和为效率调整的每日灌溉需求。分析表明，塞拉多的旱季灌溉主要补充土壤水分，这通常会导致用水效率低下。从 2019 年到 2023 年，IWU 显着增加，模型显示与旱季中心支轴式灌溉报告的累积灌溉深度具有很强的相关性 （R² = 0.92）。然而，该模型低估了 5 月至 7 月的需求，高估了 8 月的需求，偏差为 −21.88 毫米。 北部次区域受益于有利条件，占该州 IWU 的 43%。该研究提供了对 IWU 趋势的宝贵见解，支持战略决策和资源分配，同时提供了一种经济高效的实时 IWU 估计方法。