FROM SPACE TO THE SUBSURFACE. Examining Relations Between Vegetation Indices and Local Groundwater Storage.

Abstract: This study aimed to examine the relationship between vegetation indices, NDVI and NDWI, and groundwater levels in the county of Kalmar, utilizing correlation and regression analysis. Further, by examining related geospatial features the study aimed to interpret the statistical outcomes to identify significant temporal and spatial patterns. Data used involved NDVI, NDWI, derived from Sentinel 2 level-1C imagery, as well as groundwater measurement. The data was presented in time series showing bi-weekly maximum values, extending over different ranges between 2015 and 2022. Additionally, data related to land cover, soil type, topographic location, distance between groundwater and ground surface have been observed and compared between measurement stations to create a framework for interpretation. While few definite patterns have emerged, results of the study provided notable observations from performed analysis. Results showed varying strengths of correlations over the measurement stations studied, for both indices in relation to groundwater levels, with strongest correlation generally found after three-month time-lag of vegetation indices. NDVI showed positive correlation, indicating high NDVI values correlating with low groundwater levels and vice versa. NDWI over most stations showed negative correlation, indicating high NDWI values indicating high groundwater levels. Also, while soil type and median groundwater depth were features that provided notable findings when analyzed separately in relation to found correlations, combining several features make patterns less certain. Results of this study show variation in correlation being due to variations in local geospatial features. Future studies should more extensively and separately examine the effect of geospatial features related to vegetation and groundwater correlation. The use of different remote sensing data sources such as Sentinel 2 level-2A and Radar could present a more informative result. Also, addressing more careful data collection, seasonal focus and use of visualization. The addition of precipitation data could further be used to provide more detail in explaining the relationship between vegetation and groundwater.