Search for: All records

The US Drought Monitor (USDM) is a hallmark in real time drought monitoring and assessment as it was developed by multiple agencies to provide an accurate and timely assessment of drought conditions in the US on a weekly basis. The map is built based on multiple physical indicators as well as reported observations from local contributors before human analysts combine the information and produce the drought map using their best judgement. Since human subjectivity is included in the production of the USDM maps, it is not an entirely clear quantitative procedure for other entities to reproduce the maps. In this study, we developed a framework to automatically generate the maps through a machine learning approach by predicting the drought categories across the domain of study. A persistence model served as the baseline model for comparison in the framework. Three machine learning algorithms, logistic regression, random forests, and support vector machines, with four different groups of input data, which formed an overall of 12 different configurations, were used for the prediction of drought categories. Finally, all the configurations were evaluated against the baseline model to select the best performing option. The results showed that our proposed framework could reproduce the drought maps to a near-perfect level with the support vector machines algorithm and the group 4 data. The rest of the findings of this study can be highlighted as: 1) employing the past week drought data as a predictor in the models played an important role in achieving high prediction scores, 2) the nonlinear models, random forest, and support vector machines had a better overall performance compared to the logistic regression models, and 3) with borrowing the neighboring grid cells information, we could compensate the lack of training data in the grid cells with insufficient historical USDM data particularly for extreme and exceptional drought conditions. 

The increase in wildfire risk in the United States in recent decades has been linked to rapid growth of the wildland‐urban interface and to changing climate. While there have been numerous studies on wildfires and climate change, few have separately assessed the impact of climate response to land‐use‐land‐cover change (LULCC) on wildfires. In this study, we analyse two 10‐year regional climate simulations driven by the current (2011) and future (2100) land‐use‐land‐cover patterns to assess modifications by the projected LULCC to the frequency and severity of fire‐prone atmospheric conditions described by two fire weather indices, the Canadian Forest Fire Weather Index and the Hot‐Dry‐Windy Index. The simulation corresponding to future land‐use‐land‐cover pattern yields higher surface temperature and vapour pressure deficit and lower precipitation compared to the simulation with the current pattern in areas where urbanized landscapes replace forests and grasslands, such as along the Piedmont and outside the Chicagoland region, while in areas where croplands replace forests, such as the southeast Coastal Plains, the results are reversed. These changes to local and regional atmospheric conditions lead to longer fire seasons and more extreme fire‐weather conditions in much of the eastern United States, specifically in the Southeast and Ohio River Valley where significant urban expansion is projected by the end of the century. Whereas in Southern California where some highly flammable shrublands will be replaced by urban or crop lands, fire‐prone atmospheric conditions are likely to be less frequent and less extreme in the future. However, much of California moves towards a year‐round fire season under the projected LULCC. The results suggest that by altering atmospheric conditions, LULCC may play an important role in determining fire regime, but the effects are highly heterogeneous and regionalized.