Classifying rivers into homogeneous categories based on hydrological and/or environmental attributes supports the implementation of environmental flows to sustain aquatic ecosystems and support the resource needs of society. Hydrological classifications provide decision-makers with a pragmatic number of water management units by grouping individual rivers or river segments expected to exhibit similar biophysical responses to flow alteration. Such classifications are particularly useful across broad geographies and in data-limited contexts, such as in Tanzania, where the legal requirement to implement environmental flows for all major waterbodies remains constrained by scant data. We present a two-level hydrological classification of all Tanzanian basins and the Rufiji River Basin. For the Rufiji River Basin, the largest river basin in the country, we performed an inductive classification based on the availability of long-term time series of daily average discharge. We clustered 28 gauging stations into seven classes according to ecologically relevant hydrological metrics and used boosted classification trees to predict the hydrological class of all 95,909 river segments in the basin based on environmental attributes that influence flow regimes. In the absence of consistent, readily-available gauged flow data, we conducted a deductive classification of all Tanzanian rivers whereby segments were directly grouped by multivariate similarity using the same environmental attributes. This analysis revealed 10 river classes reflecting the diversity of ecohydrological conditions characterizing the 486,681 river segments draining in and out of Tanzania. The new hydrological classifications presented here provide the foundation to guide implementation of management practices within the water policy framework of Tanzania.

中文翻译：

---

坦桑尼亚河流的水文分类以支持国家水资源政策

根据水文和/或环境属性将河流分类为同质类别，支持环境流量的实施，以维持水生生态系统并支持社会的资源需求。水文分类通过对预计对流量变化表现出类似生物物理响应的单个河流或河段进行分组，为决策者提供了实用数量的水管理单元。这种分类在广泛的地域和数据有限的情况下特别有用，例如在坦桑尼亚，在那里实施所有主要水体环境流量的法律要求仍然受到数据不足的限制。我们提出了所有坦桑尼亚盆地和鲁菲吉河流域的两级水文分类。对于该国最大的流域鲁菲吉河流域，我们根据日平均流量的长期时间序列的可用性进行了归纳分类。我们根据生态相关的水文指标将 28 个测量站分为 7 个类别，并使用增强分类树根据影响流态的环境属性预测流域内所有 95,909 条河段的水文类别。在缺乏一致的、易于获得的测量流量数据的情况下，我们对所有坦桑尼亚河流进行了演绎分类，其中使用相同的环境属性按多元相似性直接分组。该分析揭示了 10 个河流类别，反映了流入和流出坦桑尼亚的 486,681 条河段的生态水文条件的多样性。