Search

The fate of the carbon stored in frozen peatlands in Canada's Hudson Bay Lowlands (HBL) depends strongly on water, with wetter conditions favoring long-term storage. Dynamic surface water products generated from historical Landsat data exist to inform surface water trends in the HBL based on binary classifications of land vs water at 30 m spatial resolution. However, the HBL contains many water features smaller than 30 m, including streams and patterned fens that require a sub-pixel mapping approach to fully capture their dynamics. In this paper, we leverage an existing binary dynamic surface water product to generate a spatially and temporally comprehensive Landsat sub-pixel surface water time-series over the HBL. We first generate Landsat Red, NIR and SWIR composites from 1985 to 2021 and evaluate data-driven machine learning and physical models to derive sub-30 m water fractions across the entire mapping domain. Data-driven models were calibrated on 30 m water fractions obtained from 16 of 17 WorldView scene water fraction maps and assessed on the 17th held-out scene in a jackknife validation. Data-driven models were also calibrated by relating scaled Landsat reflectance to water fractions obtained from binary water masks at corresponding scales from 90 to 270 m. Physical models involved simple linear unmixing using a new approach that samples land endmembers in the vicinity of each pixel to represent land reflectance, and a global water endmember sampled beneath permanent water in each image. The assumption of linear scaling was verified at sub-30 m to 270 m resolution for NIR, SWIR and NDVI, which were all highly correlated with water fraction and had no significant differences among water fraction – spectral feature relationships across the full range of scales. Each of 11 different methods and input spectral features were evaluated against WorldView water fractions and ranked based on nine criteria. We found that machine learning using all spectral features including local NIR land endmembers performed best among data-driven approaches, and linear unmixing of the NIR band using local land endmembers performed best among physical. Both best-performing methods were applied to the entire time-series and surface water area trends were calculated and compared with trends obtained from binary water masks. All three methods agree on the direction of surface water change towards wetting, with sub-pixel inter-annual variation agreeing with stream discharge noted in other studies. • Physical and data-driven Landsat water fraction methods tested in the Hudson Bay Lowlands. • Physical methods perform linear unmixing between local land and global water endmembers. • Data-driven methods consist of machine learning calibrated using high-resolution water. • Best methods applied to 1985–2021 Landsat time-series reveal overall wetting trends. • Spatial trends generally suggest drying in bogs and wetting in fens. [ABSTRACT FROM AUTHOR]

A variety of evidence suggests that the boreal forests of Canada are responding to climate change. Specifically, several studies have inferred that widespread browning trends detected in time series of the Normalized Difference Vegetation Index (NDVI) from the Advanced Very High Resolution Radiometer (AVHRR) reflect the response of boreal forests to longer growing seasons, increased summer drought stress, and higher frequency of fires. Data from the Thematic Mapper (TM5) and Enhanced Thematic Mapper Plus (ETM+) sensors onboard Landsat 5 and 7, respectively, span essentially the same time period as the AVHRR record, but provide data with substantially higher radiometric and spatial fidelity, and by extension, a much improved basis for evaluating decadal-scale trends in spectral vegetation indices such as the NDVI. However, detection of trends, which are often subtle, requires careful attention to ensure that artifacts associated with the quality and stability of inter- and intra-sensor calibration do not lead to spurious conclusions in results from time series analyses. In this paper, we use time series of TM5 and ETM+ images for fifteen sites distributed across the Canadian boreal forest zone to explore if and how sensor geometry and inter- and intra-sensor calibration affect trends in spectral vegetation indices derived from multi-decadal Landsat time series. To do this, we created annual cloud-free composites for each Landsat spectral band based on peak summer NDVI at each site from 1984 to 2011 using all available TM5 and ETM+ data. To distinguish trends arising from long term climate change from those related to disturbance, we isolated areas within each site that were undisturbed during the Landsat record, and used these locations to analyze sources of variance in time series of red reflectance, near-infrared (NIR) reflectance, the NDVI, and the Enhanced Vegetation Index (EVI). Our results highlight the challenges involved in distinguishing trends in surface properties from data artifacts caused by undetected atmospheric effects, changes in sensor view angles, and subtle radiometric differences between the TM5 and ETM+ sensors. In particular, differences in sensor view geometry across adjacent overlapping Landsat scenes cause vegetated pixels in the eastern portion of Landsat scenes to have higher reflectances in the red and NIR bands (by 5 and 6 percent, respectively) than pixels in the western portion of scenes. While this effect does not significantly change NDVI values, it does affect EVI values. We also found modest, but potentially significant, differences between the red band reflectance of each sensor, with TM5 data having 14 percent higher red reflectance on average for vegetated pixels, which can introduce spurious trends in time series that combine TM5 and ETM+ data. More generally, the results from this work demonstrate that while the 30 + year Landsat archive provides unprecedented opportunities for studying changes to the Earth's terrestrial biosphere over the last three decades, care must be taken when inferring trends in these data without considering how sources of variance unrelated to surface processes affect the integrity of Landsat time series. [ABSTRACT FROM AUTHOR]