@misc{schmitt_modeling_atmosphereland_2023, 
author={Schmitt, A.U.,Ament, F.,de Araújo, A.C.,Sá, M.,Teixeira, P.},
title={Modeling atmosphere–land interactions at a rainforest site – a case study using Amazon Tall Tower Observatory (ATTO) measurements and reanalysis data},
year={2023},
howpublished = {journal article},
doi = {https://doi.org/10.5194/acp-23-9323-2023},
abstract = {Modeling the interactions between atmosphere and soil at a forest site remains a challenging task. Using tower measurements from the Amazon Tall Tower Observatory (ATTO) in the rainforest, we evaluated the performance of the land surface model JSBACH, focusing especially on processes influenced by the forest canopy. As a first step, we analyzed whether high-resolution global reanalysis data sets are suitable to be used as land surface model forcing. Namely, we used data from the fifth-generation ECMWF atmospheric reanalysis of the global climate (ERA5) and the Modern-Era Retrospective analysis for Research and Applications version 2 (MERRA-2). Comparing 5 years of ATTO measurements to near-surface reanalysis data, we found a substantial underestimation of wind speeds by about 1 m s−1. ERA5 captures monthly mean temperatures quite well but overestimates annual mean precipitation by 30 %. Contrarily, MERRA-2 overestimates monthly mean temperatures in the dry season (August–October) by more than 1 K, while mean precipitation biases are small. To test how much the choice of reanalysis data set and the reanalysis biases affect the results of the land surface model, we performed spin-up and model runs using either ERA5 or MERRA-2 and with and without a bias correction for precipitation and wind speed and compared the results. The choice of reanalysis data set results in large differences of up to 1.3 K for soil temperatures and 20 % for soil water content, which are non-negligible, especially in the first weeks after spin-up. Correcting wind speed and precipitation biases also notably changes the land surface model results – especially in the dry season. Based on these results, we constructed an optimized forcing data set using bias-corrected ERA5 data for the spin-up period and ATTO measurements for a model run of 2 years and compared the results to observations to identify model shortcomings. Generally, the shape of the soil water profile is not reproduced correctly, which might be related to a lack of vertical variability of soil properties or of the root density. The model also shows a positive soil temperature bias and overestimates the penetration depth of the diurnal cycle. To tackle this issue, potential improvements can be made by improving the processes related to the storage and vertical transport of energy. For instance, incorporating a distinct canopy layer into the model could be a viable solution.},
note = {Online available at: \url{https://doi.org/10.5194/acp-23-9323-2023} (DOI). Schmitt, A.; Ament, F.; de Araújo, A.; Sá, M.; Teixeira, P.: Modeling atmosphere–land interactions at a rainforest site – a case study using Amazon Tall Tower Observatory (ATTO) measurements and reanalysis data. Atmospheric Chemistry and Physics. 2023. vol. 23, no. 16, 9323-9346. DOI: 10.5194/acp-23-9323-2023}}