Overview
Monitoring drought impacts in forest ecosystems is complex because forests are composed of different species with heterogeneous structural compositions. Even though forest drought status is a key control on the carbon cycle, very few indices exist to monitor forest drought stress. The Forest Drought Indicator (ForDRI) is a new monitoring tool to identify and characterize forest drought stress1. It was developed by the National Drought Mitigation Center (NDMC) at the University of Nebraska-Lincoln (UNL) in collaboration with the U.S. Department of Agriculture (USDA), U.S. Forest Service (USFS), and Center for Advanced Land Management Information Technologies (CALMIT) at UNL.
ForDRI maps are produced weekly at a 1-km spatial resolution and provide information about drought's effects on forest regions of the continental U.S. (CONUS). ForDRI integrates data for 12 different environmental variables, including vegetation health, climate, evaporative demand, ground water and soil moisture, into a single hybrid index to estimate forest-related drought stress. ForDRI uses a principal component analysis (PCA) method to determine the relative contribution of each input variable based on its covariance for an extended historical record (2003-2020). Historical data of the ForDRI is limited based on the availability of the satellite time series data record.
General Questions
What datasets go into the ForDRI?
The ForDRI models integrate several water cycle-related variables (e.g., precipitation, evaporation, soil moisture and vapor pressure deficit) that influence short- and long-term drought conditions, which are combined with satellite-derived vegetation condition indicators to detect and assess forest-related drought. Specifically, the following 12 input variables are used in the weekly production of the ForDRI maps:
Satellite-derived variables
- GRACE Groundwater Storage (GWS) at 1-meter depth is used to help assess the available water used by deeper-rooted tree species. Because this dataset has a coarse resolution of 12.5 km, it was resampled to produce maps at 1-km spatial resolution.
- Moderate Resolution Imaging Spectroradiometer (MODIS) based weekly Normalized Difference Vegetation Index (NDVI) at 1-km resolution is used to characterize forest conditions. Because the MODIS dataset was discontinued in mid-February 2023, the long-term data records from MODIS observations have been extended by the Visible Infrared Imaging Radiometer Suite (VIIRS) instruments. Thus, based on our recent study [3], the MODIS-VIIRS cross-walked dataset is used for the weekly map production of ForDRI.
Climate variables
- The 9-, 12-, 24- and 60-month Standardized Precipitation Index (SPI) are used to quantify the precipitation anomaly. We have used these four SPIs to represent different time scales of the rainfall conditions that would affect forest health.
- The 12-, 24- and 60-month Standardized Precipitation Evapotranspiration Index (SPEI) are used to consider both precipitation and temperature. The historical records of these three SPEIs (i.e., the previous 12, 24 and 60 months) represent the temperature impact on water demand (rainfall).
Biophysical variables
- The 12-month Evaporative Demand Drought Index (EDDI) indicates the anomalous condition of the atmospheric evaporative demand (also known as “the thirst of the atmosphere”).
- Noah soil moisture top layer data (at 10cm depth) derived from the North American Land Data Assimilation System datasets (NLDAS-2) is used to represent shallow soil depth conditions that can be accessed by shallow-rooted species.
- The vapor pressure deficit (VPD) from PRISM represents the difference between the actual water vapor pressure in the air and the vapor pressure when the air at that temperature is saturated.
How are the forest cover areas and forest type groups determined in the ForDRI?
The forest cover areas and forest type groups used in developing the ForDRI are based on the national forest type dataset produced by the U.S. Forest Service (USFS) Forest Inventory and Analysis (FIA) program and the Remote Sensing Applications Center (RSAC). This USFS national forest type dataset was created by modeling several biophysical layers, including digital elevation models (DEM), Moderate Resolution Spectroradiometer (MODIS) multi-date composites, vegetation indices and vegetation continuous fields, class summaries from the National Land Cover Dataset (NLCD), various ecologic zones, and summarized PRISM climate data. The ForDRI maps show only these forest areas with other land cover type areas masked from the map2.
In developing the ForDRI model, how is the contribution (or weight) of each input variable determined?
The current ForDRI model uses Principal Component Analysis (PCA) to determine the contribution of each climate and satellite input variable based on its covariance in the historical records from 2003 to 2020. The weights are determined for every 1-km grid cell separately across the U.S. with unique weights calculated for each of the 52 weekly periods across the year. Weekly defined weights for each grid cell are done to account for the seasonality of the trees’ growth during the year.
Why is the length of the historical record limited to 2003 to 2020?
The length of the historical record is limited to this period because of the satellite record of MODIS data. Ideally, a long historical record would be preferred. However, in this 18-year period, it is assumed that many locations have experienced representative events for most ForDRI drought intensity classifications. Since the MODIS data is already discontinued, we are using the Visible Infrared Imaging Radiometer Suite (VIIRS) data for recent years’ ForDRI maps as well as for operational purposes3.
How is the accuracy of ForDRI assessed? Are future updates or improvements of ForDRI models planned?
The ForDRI values were compared with normalized weekly Bowen ratio data (a biophysically based indicator of stress) from 22 AmeriFlux sites and 66 tree ring sites in the U.S. There were relatively strong and significant correlations between ForDRI and Bowen ratio data as well as ForDRI and tree-ring increments at the sites that had experienced intense drought. However, we are continuing to evaluate the ForDRI model and associated products by collecting feedback from experts (and users) to improve the model. Changes will be made to the existing ForDRI models if we find that new inputs improve the ForDRI maps. Thus, we plan to investigate new model inputs and other changes to the ForDRI model over the upcoming year for the continental U.S.
For technical questions, contact:
Dr. Tsegaye Tadesse, email: ttadesse2@unl.edu
References
- Tadesse, T., Hollinger, D.Y., Bayissa, Y.A., Svoboda, M., Fuchs, B., Zhang, B., Demissie, G., Wardlow, B.D., Bohrer, G., Clark, K.L. and Desai, A.R., 2020. Forest Drought Response Index (ForDRI): a new combined model to monitor forest drought in the eastern United States. Remote Sensing, 12(21), p.3605.
- USDA Forest Service. National Forest Type Dataset. 2020. Available online: https://data.fs.usda.gov/geodata/rastergateway/forest_type/
- Benedict, T.D., Brown, J.F., Boyte, S.P., Howard, D.M., Fuchs, B.A., Wardlow, B.D., Tadesse, T. and Evenson, K.A., 2021. Exploring VIIRS continuity with MODIS in an expedited capability for monitoring drought-related vegetation conditions. Remote Sensing, 13(6), p.1210.