Combining spatial downscaling technique and diurnal temperature cycle model to acquire diurnal patterns of land surface temperature at field scale

Authors: SARA, K - Indian Institute Of Technology; RAJASEKARAN, E - Indian Institute Of Technology; Kustas, William; Alfieri, Joseph; Prueger, John; ALSINA, M - E & J Gallo Winery; HIPPS, L - Utah State University; McKee, Lynn; McElrone, Andrew; CASTRO, S - University Of California, Davis; BAMBACH, N - University Of California, Davis

Submitted to: Journal of Photogrammetry and Remote Sensing
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 4/10/2024
Publication Date: 5/21/2024
Citation: Sara, K., Rajasekaran, E., Kustas, W.P., Alfieri, J.G., Prueger, J.H., Alsina, M., Hipps, L.E., Mckee, L.G., Mcelrone, A.J., Castro, S., Bambach, N. 2024. Combining spatial downscaling technique and diurnal temperature cycle model to acquire diurnal patterns of land surface temperature at field scale. Journal of Photogrammetry and Remote Sensing. 92:723-740. https://doi.org/10.1007/s41064-024-00291-1.
DOI: https://doi.org/10.1007/s41064-024-00291-1

Interpretive Summary: Land surface temperature (LST) is a crucial variable controlling the exchange of water and energy fluxes between earth's surface and the atmosphere and is an essential boundary condition for remote sensing models for monitoring evapotranspiration and plant stress. The diurnal cycle in LST is affected not only by incoming solar radiation but by land cover type, vegetation characteristics, agricultural practices and surface moisture availability The diurnal cycle can be monitored using thermal sensors placed in geostationary orbits with a coarser spatial resolution on the order of 2 to 5 km. This coarse-resolution data misses the finer variation in LST over most agricultural fields that are much smaller in size. This study applies Principal Component Regression (PCR) based disaggregation methods using coarser resolution geostationary and polar orbiting satellite-based LST to create diurnal LST observations over a variety of agricultural fields (wheat, rice and vineyard) containing continuous tower-based LST data and also high-resolution satellite LST observations. The PCR technique using multiple indices captured satisfactorily the spatial and diurnal pattern of LST over these croplands suggesting this method has potential to be used with remote sensing-based evapotranspiration models for monitoring diurnal crop water use and stress.

Technical Abstract: The high spatial resolution diurnal land surface temperature is a crucial variable required for monitoring evaporative stress in plants, heatwave events, and drought. In this paper, we have proposed to derive the field scale Diurnal Temperature Cycle (DTC) using a combination of downscaling techniques and a Diurnal Temperature Model, GOT01-ts. For downscaling the Land surface temperature (LST) acquired from medium-resolution sensors like Moderate Resolution Imaging Spectroradiometer (MODIS) and Visible Infrared Imaging Radiometer Suite (VIIRS), we have used two disaggregation techniques: Principal Component Regression based disaggregation, Distrad technique and a Spatio Temporal Integrated Temperature Fusion Model (STITFM). The PCR-based disaggregation technique used multiple fine-resolution auxiliary datasets such as vegetation indices, buildup indices, backscattering coefficient, and topographic datasets. The downscaled LSTs were then fitted using the DTC model to derive the corresponding diurnal temperature cycle at fine resolution. The models were tested in four different sites. The downscaled LSTs were compared with the original fine-resolution ECOSTRESS and Landsat LSTs. The DTC cycle fitted using the downscaled, and original LSTs were compared with the insitu diurnal temperature cycle obtained from net radiometers. The PCR technique using multiple indices captured the spatial and diurnal pattern of LST fairly well (RMSE~ 2.481 K, R2 ~0.95).