Contiguous U.S. Daily PM_2.5 Measurements (2021) – Benchmark Dataset for Location‑Encoder Evaluation

Deep learning models have demonstrated success in geospatial applications, yet quantifying the role of geolocation information in enhancing model performance and geographic generalizability remains underexplored. A new generation of location encoders have emerged with the goal of capturing attributes present at any given location for downstream use in predictive modeling. Being a nascent area of research, their evaluation has remained largely limited to static tasks such as species distributions or average temperature mapping. In this paper, we discuss and quantify the impact of incorporating geolocation into deep learning for a real-world application domain that is characteristically dynamic (with fast temporal change) and spatially heterogeneous at high resolutions: estimating surface-level daily PM_2.5 levels using remotely sensed and ground-level data. We build on a recently published deep learning-based PM_2.5 estimation model that achieves state-of-the-art performance on data observed in the contiguous United States. We examine three approaches for incorporating geolocation: excluding geolocation as a baseline, using raw geographic coordinates, and leveraging pretrained location encoders. We evaluate each approach under within-region (WR) and out-of-region (OoR) evaluation scenarios. Aggregate performance metrics indicate that while naïve incorporation of raw geographic coordinates improves within-region performance by retaining the interpolative value of geographic location, it can hinder generalizability across regions. In contrast, pretrained location encoders like GeoCLIP enhance predictive performance and geographic generalizability for both WR and OoR scenarios. However, our qualitative analysis reveals artifact patterns caused by high-degree basis functions and sparse upstream samples in certain areas, and our ablation results indicate varying performance among location encoders such as SatCLIP vs. GeoCLIP. To the best of our knowledge, this is a first integration and systematic evaluation of location encoders in a complex, temporally dynamic estimation scenario. In addition to guiding better model development for air pollution estimation and location encoders, this study provides insights for effective incorporation of location into deep learning for geospatial predictive tasks.

深度学习模型已在地理空间应用中展现出优异性能，但量化地理位置信息对提升模型性能与地理泛化能力的作用，仍有待深入探索。新一代位置编码器（location encoders）应运而生，旨在捕捉任意给定位置的属性特征，以供预测建模的下游任务使用。作为新兴研究方向，此类编码器的评估目前大多局限于物种分布、平均气温制图等静态任务。本文针对一类兼具快速时间变化特性与高分辨率空间异质性的动态真实应用场景——即利用遥感与地面观测数据估算地表每日PM₂.₅浓度——探讨并量化了将地理位置信息融入深度学习模型的影响。本研究基于近期发表的深度学习PM₂.₅估算模型，该模型在美国本土观测数据集上达到了当前最优（state-of-the-art）性能。我们测试了三种融入地理位置信息的方案：以不纳入地理位置信息作为基准对照、使用原始地理坐标，以及利用预训练位置编码器（pretrained location encoders）。我们在区域内（within-region, WR）与跨区域（out-of-region, OoR）两种评估场景下对各方案进行测试。综合性能指标显示，尽管朴素融入原始地理坐标可通过保留地理位置的插值价值提升区域内性能，但却会削弱跨区域泛化能力。与之相反，诸如GeoCLIP这类预训练位置编码器，可在区域内与跨区域场景下同时提升预测性能与地理泛化能力。不过，我们的定性分析发现，部分区域存在由高阶基函数与上游样本稀疏所引发的伪影模式；同时消融实验结果显示，不同位置编码器（如SatCLIP与GeoCLIP）的性能存在差异。据我们所知，本研究首次在复杂的时序动态估算场景中对位置编码器进行集成与系统性评估。本研究不仅为空气污染估算与位置编码器的模型优化开发提供了指导，还为如何在地理空间预测任务中高效融入地理位置信息的深度学习建模提供了参考思路。