用于湿地测绘的人工智能深度学习模型准确率达94%

2023年1月10日

安纳波利斯, MD – 十大赌博正规老平台’s data science team developed an artificial intelligence deep learning model for mapping 湿地, 这导致了94%的准确率. EPRI支持, 一个独立的, non-profit energy research and development institute; Lincoln Electric System; and the Grayce B. 克尔基金公司., this method for wetland mapping could deliver important outcomes for protecting and conserving 湿地. 研究结果发表在同行评议期刊上整体环境科学.

卡尔弗特悬崖州立公园.
威尔·帕森/切萨皮克湾项目

The team trained a machine learning (convolutional neural network) model for high-resolution (1m) wetland mapping with freely available data from three areas: Mille Lacs County, Minnesota; Kent County, Delaware; and St. 劳伦斯县，纽约. 完整模型, which requires local training data provided by state 湿地 data and the National Wetlands Inventory (NWI), 以94%的准确度绘制湿地地图.

“We’re happy to support this exciting project as it explores new methods for 湿地 delineation using satellite imagery,EPRI首席技术主管Dr. Nalini饶. “It has the potential to save natural resource managers time in the field by using a GIS tool right from their desks. +, it can help companies and the public manage impacts to 湿地 as infrastructure builds are planned to meet decarbonization targets.”

“The Infrastructure Investment and Jobs Act is pouring hundreds of billions of dollars into projects that will have an impact on the landscape. 然而, 我们用来减少对湿地影响的数据已经过时了,” said Environmental Policy Innovation Center’s Restoration Economy Center Director Becca Madsen, 前EPRI研究员. “There has never been a better time to invest in updating our nation’s wetland data and establishing a sustainable and cost-effective process for keeping them updated.”

“When this highly accurate model is scaled up to predict 湿地 in much bigger geography such as the Chesapeake Bay or the contiguous United States, 这将改变游戏规则. It obviates the need for manual mapping of 湿地 as well as mapping 湿地 with traditional machine learning which require a lot of data processing, 策展和手动特征工程, 这两种方法都很耗时, 劳动密集型且非常昂贵,十大赌博正规老平台协会的数据科学主管/高级数据科学家Dr. Kumar Mainali.

这对保护意味着什么 & 保护湿地

新模型将帮助基础设施规划者在规划过程中避免使用湿地, 从而节约成本和保护湿地. Potential beneficial situations include ongoing efforts to expand and develop renewable energy, 这需要扩大电力基础设施.

模型的产物是湿地概率图. 该概率数据可用于绘制最可能的湿地范围, 但是如果用户喜欢的话, 它们可以以较低的概率阈值绘制湿地范围. The resulting map limits the likelihood of wetland omission even though it maps more 湿地 than are present in reality.

There could also be potential to use this model to map locations where 湿地 have already been lost since they were mapped with NWI. 此外，还可以确定湿地恢复的潜在地点. 例如, persistently wet agricultural fields are picked up by the model even though for the purposes of field wetland delineation, 这些地区在被积极耕种时不被认为是湿地.

下一个步骤

The team will expand the model to states or larger regions and continue to train the model on varied geographies.

内布拉斯加州试点模式克服过时数据

遵循最初的模型开发, 该模型被扩展到兰开斯特县, 内布拉斯加州. Modeling 湿地 in this region proved challenging because the NWI data for the area was decades out of date, 其中包括几个因开发而失去湿地的地区. The team was interested to learn whether the model could succeed in mapping 湿地 where no recent high-quality 湿地 datasets were available to train the model.

The wetland model was trained with the decades-old NWI dataset and recent satellite and aerial imagery data. The team found that the NWI data improved the local accuracy of wetland mapping by 10% compared to predictions before training, 显示在新地区使用当地培训数据的重要性. 除了, 该模型正确地忽略了因开发而消失的湿地, 尽管这些湿地的训练数据仍然是过时的, as shown in the image below (outdated training data shown in green; model prediction in purple, 覆盖最近的卫星图像). The performance of the model in determining the dominant pattern in the data to both improve local mapping accuracy but still accurately reflect wetland presence and absence is promising for the usefulness of this approach.

Despite the important role of 湿地 data for planning infrastructure projects and managing wildlife, NWI湿地数据多年未全面更新. 如下图所示, 全国范围内的许多NWI数据可以追溯到20世纪70年代和80年代, 但仍然是现有的最佳数据. A modeling approach to wetland mapping that can utilize training data of varying vintages will be incredibly useful in modernizing wetland mapping where it is most needed. (欲了解更多信息，请参见“是时候投资制作我们国家湿地的现代地图了”: http://www.policyinnovation.org/blog/investing-in-a-modern-map-of-our-nations -湿地)

关于模型

The “predictor” layers used in wetland training from which the model learns the patterns found in 湿地 were: USDA National Agriculture Imagery Program (NAIP) aerial imagery (1m), Sentinel-2光学卫星图像(10-20米), LiDAR-derived geomorphons, an approach to mapping landforms that 十大赌博正规老平台 has been applying to advance high-resolution stream mapping; and LiDAR intensity, 一种常用于识别水和持续潮湿土壤的指数.

另外, the team trained a simpler model using only USDA NAIP and Sentinel-2 data as the input layers, 确保精度为91.6%.

The coauthors of the paper are 十大赌博正规老平台’s Data Science Lead/Senior Data Scientist Kumar Mainali, Ph.D.，高级数据科学家Michael Evans，博士.D., 地理空间技术经理艾米丽·米尔斯(原十大赌博正规老平台协会), 高级地理空间技术主管David Saavedra, 气候战略副总裁苏珊·明尼迈尔, 以及前EPRI项目经理贝卡·马德森, 现在就职于环境政策创新中心.

阅读出版物"Convolutional Neural Network for High-Resolution Wetland Mapping with Open Data: Variable Selection and the Challenges of a Generalizable Model在线整体环境科学.

###