Archives: Articles

Exploring Relationships Between Drought Indices and Ecological Drought Impacts Using Machine Learning and Explainable AI

Authors:	Annie Britton, Garrett Graham, Molly Woloszyn
Volume:	Volume 2024, No. 005
DOI:	https://doi.org/10.46275/JOASC.2024.09.001
Abstract:	Rangeland ecosystems in the United States hold great ecological, economic, and cultural value. However, the increasing frequency and severity of droughts pose potential threats to these ecosystems. This study used remotely sensed data, machine learning, and explainable artificial intelligence (XAI) to explore relationships between drought indices and vegetation health in the Cheyenne River Basin, USA. The study employed XGBoost Regressor and Extra Trees Regressor models in conjunction with SHapley Additive exPlanations (SHAP) to identify the most influential drought indices and environmental variables for predicting Normalized Difference Vegetation Index (NDVI), thereby uncovering indicators of vegetation stress in the basin. The XGBoost regressor was moderately successful at predicting NDVI, making the model suitable for subsequent XAI analysis using SHAP. SHAP results revealed that the Palmer Drought Severity Index (PDSI), the 90-day Standardized Precipitation Index (SPI), and snow water equivalent (SWE) were the most influential predictors of NDVI, indicating their strong association with changes in vegetation health in the Cheyenne River Basin. This study demonstrates the feasibility and value of applying XAI methods to investigate both the strength and directionality of ecological drought indicators—an approach that has been underutilized in drought research. These insights can inform future drought research, improve monitoring efforts, and help anticipate ecological drought impacts in the region.
Link:	https://stateclimate.org/pdfs/journal-articles/2024_5-Britton.pdf

ChatGPT in Climatology: Transforming Climate Research with Conversational AI

Authors:	Jacob L. Fields, Geddy R. Davis, Zachary T. Leasor, Jason Cervenec
Volume:	Volume 2024, No. 004
DOI:	https://doi.org/10.46275/JOASC.2024.07.001
Abstract:	In recent years, advancements in the field of artificial intelligence have increased exponentially, culminating in widely available user-based tools such as ChatGPT. Although fairly new, these tools have as of yet been underutilized by the scientific community, including climatology. As a large language model, ChatGPT’s strong ability to accurately respond to prompts allows it to be used as a comprehensive tool with a variety of applications in climatology, which this article groups into practical and conceptual applications. Practically, ChatGPT excels in the assistance of code creation and troubleshooting, allowing for efficient automation of data collection, as well as the process of basic data sorting. Conceptually, the tool gives a foundation for researchers to “fill the knowledge gap” by gaining a basic understanding of supplementary information presented in the literature review portion of a research project. While ChatGPT is powerful, it contains significant limitations that hinder its status as a standalone tool, such as occasional inaccurate responses, lack of transparency, and absence of data protection. Despite these setbacks, use of ChatGPT in a responsible and ethical manner with awareness of its limitations can be efficient, dynamic, and adherent to the principles of scientific integrity.
Link:	https://stateclimate.org/pdfs/journal-articles/2024_4-Fields.pdf

Development of Alternate Climate Divisions for Colorado Based on Gridded Data

Authors:	Russ S. Schumacher, Rebecca A. Bolinger, Jeffrey J. Lukas
Volume:	Volume 2024, No. 003
DOI:	https://doi.org/10.46275/JOASC.2024.06.002
Abstract:	The official climate divisions for the contiguous United States are used for a wide range of purposes, including ongoing climate monitoring, and through NOAA’s long-standing nClimDiv dataset. In Colorado, the climate divisions are based around the basins of the large rivers that flow out of the state. However, considering the complex topography and climate of the state, these divisions do not always represent key climate variations and changes. This study builds upon an approach first developed by Wolter and Allured to establish alternate climate divisions that more closely reflect observed climate variability across Colorado. Hierarchical cluster analysis is applied to gridded temperature and precipitation data (NOAA’s nClimGrid) from 1950-2021 to identify areas with similar climate variability, then manual inspection is used to establish 11 divisions. These resulting divisions are being used in an updated state-level climate change assessment. The method is flexible and uses open-source tools that could be extended to other regions or datasets.
Link:	https://stateclimate.org/pdfs/journal-articles/2024_3-Schumacher.pdf

Estimating Agricultural Irrigation Water Usage in Delaware, USA

Authors:	Kevin R. Brinson, Tracy L. DeLiberty, Daniel J. Leathers
Volume:	Volume 2024, No. 002
DOI:	https://doi.org/10.46275/JOASC.2024.06.001
Abstract:	Irrigation is an important agricultural management practice and the second largest consumer of fresh-water resources in Delaware. As more farmland is converted to irrigated agriculture, it is crucial that water resource managers be able to determine reasonable estimates of irrigation water usage in order to protect the resource. This study used a soil water balance approach to simulate irrigation for corn and soy-beans in Delaware (USA). The simulations were divided into four scenarios to determine which irrigation management method best represents agricultural irrigation water usage in Delaware. Two scenarios utilized an evapotranspiration-based (ET-based) approach using meteorological data from the Delaware Environmental Observing System (DEOS) with a soil water availability threshold to determine when and how much to irrigate. A second set of scenarios used a calendar-based approach and a rain gauge trigger to simulate irrigation. Analyses were performed to examine the seasonal and spatial variability of irrigation in Delaware and to compare simulated irrigation data to reported water use data provided by the Delaware Department of Natural Resources and Environmental Control’s Water Allocation Permit program. Seasonal irrigation varied due to environmental as well as irrigation decision making factors. Spatially, irrigation varied primarily as a result of the soil water holding properties. The ET-based scenario with a fixed amount of irrigation had the best agreement with the reported irrigated water use data. This study demonstrated the utility of high-resolution, environmental data with a soil water balance approach to improve estimates of irrigated water usage at the state and regional scale.
Link:	https://stateclimate.org/pdfs/journal-articles/2024_2-Brinson.pdf

Drought Assessment in a Changing Climate: A Review of Climate Normals for Drought Indices

Authors:	Joel Lisonbee, John Nielsen-Gammon, Blair Trewin, Gretel Follingstad, and Britt Parker
Volume:	Volume 2024, No. 001
DOI:	https://doi.org/10.46275/JOASC.2024.05.001
Abstract:	Should drought be considered an extreme dry period based on the entire record of available data? Or, should drought be considered a low in precipitation variability within the context of a present, contemporary climate? The two most common reference periods are the full period of record (all observed data or as much as possible) and a 30-year reference climatology. However, climate non-stationarity may render the "all-data" approach an inaccurate or obsolete comparison unless a trend is factored in. The aim of this review is to explore the literature for approaches to addressing these issues. The World Meteorological Organization (WMO) has recommended a 30-year reference period for most climatological applications since 1935, but for drought assessments and drought indices the modus operandi has been to use as much data as possible. However, in the literature, the “all data” approach has been challenged by evident impacts from climate change-induced non-stationarity. Over the past several years, as potential errors in drought assessments became more apparent due to a stationarity assumption when applying drought indices, several studies have adopted shorter reference periods, with 30-years being the most common. Furthermore, several recent papers have recommended using short reference periods with more frequent data updates for drought assessments to be representative of a contemporary climate. Additionally, at least 18 non-stationary drought indices have been proposed in efforts to retain long datasets and account for non-stationarity in the climate system.
Link:	https://stateclimate.org/pdfs/journal-articles/2024_1-Lisonbee.pdf

Using Climatological Data to Identify Locations with Viticultural Potential in Colorado

Authors:	Peter E. Goble, Horst W. Caspari, and Russ S. Schumacher
Volume:	Volume 2023, No. 1
DOI:	https://doi.org/10.46275/JOASC.2023.04.001
Abstract:	Western Colorado’s warm, dry summers and access to mountain river water for irrigation create ideal conditions for the growth of wine grapes, specifically cultivars of the European grape species Vitis vinifera. The largest limiting factor to Vitis vinifera production is nocturnal temperatures cold enough to damage crops, or Low Temperature Injury Events (LTIEs). LTIEs require producers to undergo the time-and-cost prohibitive venture of retraining vines. Eastern Mesa County Colorado has sustained large-scale grape production due to the area’s relatively mild cold season weather. Areas with similarly hospitable conditions may exist elsewhere within western Colorado. Parameter-elevation Regressions on Independent Slopes Model (PRISM) temperatures (1981-2020) were used to estimate the frequency of LTIEs across Colorado, and identify trends associated with a warming climate. In the interest of comparing PRISM temperatures to observations over actual vineyards, thermometers were placed on current vineyards in Montezuma County from 2016 to 2020. Findings suggest additional areas of opportunity for Vitis vinifera production exist in Colorado, particularly western Montezuma County, and western Mesa and Montrose Counties. Like eastern Mesa County, these areas experience a LTIE in fewer than 20 % of years. PRISM data also suggest southeast Colorado is becoming more hospitable for Vitis vinifera growth over time. Temperature measurements in Montezuma County during potentially lethal weather events compared closely with PRISM data, with a mean absolute difference of 1.8 ˚C. This comparison suggests PRISM is a reliable tool for identifying areas of opportunity in spite of western Colorado’s complex terrain.
Link:	https://stateclimate.org/pdfs/journal-articles/2023_1-Goble.pdf

To Plant or Not to Plant? A Soil Temperature Climatology for the Northern and Central Plains

Authors:	Olivia G. Campbell, Natalie A. Umphlett, and Crystal J. Stiles
Volume:	Volume 2022, No. 1
DOI:	http://doi.org/10.46275/JOASC.2022.01.001
Abstract:	Sufficient soil temperatures at the time of planting are essential for a well-established stand in both large-scale agriculture and recreational home gardening. Planting too early in the season increases the risk for frost damage and slow seedling growth while planting too late risks not reaching the required growing degree days (GDD) for plant maturity. In this study, a climatology of the date in which soils reach critical temperature thresholds for crops was developed for the Northern and Central Plains. At least 15 years of soil temperature data from 155 automated stations from six different networks were utilized in this study. Results showed that Minnesota consistently reached each soil temperature threshold last, while south-central Colorado reached each threshold first, with differences in air temperature and soil moisture likely playing a role. These results were incorporated into an online tool that both professional and recreational agriculturists can use to determine when soil temperatures are best for planting. It will also help put soil temperatures into context based on a climatological average
Link:	https://stateclimate.org/pdfs/journal-articles/2022_1-Campbell.pdf

Status and Climate Applications of the 19th Century Forts and Volunteer Observer Database

Authors:	Nancy E. Westcott, Jason Cooper, Karen Andsager, Leslie A. Stoecker, and Karsten Shein
Volume:	Volume 2021, No. 2
DOI:	http://doi.org/10.46275/JOASC.2021.09.001
Abstract:	The Climate Data Modernization Program Forts and Volunteer Observer Database (CDMP-Forts) currently consists of 450 keyed and 355 quality-controlled stations for the period 1788–1892, reaching across the United States. In conjunction with the Global Historical Climate Network (GHCN) daily data, this resource is invaluable for examining 19th century weather and climate in the United States. CDMP-Forts is incomplete, however, with a considerable amount of data remaining to be digitally transcribed and quality controlled. It is the intent of this paper to provide an overview of the processes involved in rescuing these data and to show important ways these data can be used and the considerations that may have to be taken to create meaningful analyses. Finally, the dataset is placed in the context of other global datasets and efforts to rescue historical weather data.
Link:	https://stateclimate.org/pdfs/journal-articles/2021_2-Westcott.pdf

Making sense of flash drought: definitions, indicators, and where we go from here

Authors:	Joel Lisonbee, Molly Woloszyn, Marina Skumanich
Volume:	Volume 2021, No. 1
DOI:	http://doi.org/10.46275/JOASC.2021.02.001
Abstract:	The topic of “Flash Drought” is rapidly gaining attention within both the research and drought management communities. This literature review aims to synthesize the research to-date and provide a basis for future research on the topic. Specifically, our review is focused on documenting the range of definitions of “flash drought” being proposed in the research community. We found that the term first appeared in the peer-reviewed literature in 2002, and by 2020 has become an area of active research. Within that 18-year span, “flash drought” has been given 29 general descriptions, and 20 papers have provided measurable, defining criteria used to distinguish a flash drought from other drought. Of these papers, 11 distinguish flash drought as a rapid-onset drought event while eight distinguish flash drought as a short-term or short-lived, yet severe, drought event and one paper considers flash drought as both a short-lived and rapid onset event. Of the papers that define a flash drought by its rate of onset, the rate proposed ranges from 5 days to 8 weeks. Currently, there is not a universally accepted definition or criteria for “flash drought,” despite recent research that has called for the research community to adopt the principle of rapid-intensification of drought conditions.
Link:	https://stateclimate.org/pdfs/journal-articles/2021_1-Lisonbee.pdf

Glen Conner Memorial

Authors:	Stuart A. Foster
Volume:	Volume 2020, No. 3
DOI:	http://www.doi.org/10.46275/JoASC.2020.12.001
Abstract:	Doral Glen Conner, founder of the Kentucky Climate Center and State Climatologist Emeritus, died peacefully in his home on January 6, 2020, at the age of 89. He was born, raised, and died in rural Allen County, Kentucky, a place that he treasured, but his life was an adventure that took him around the world. The son of Payton Ham Conner and Nellie Helen Weaver Conner, Glen was born on March 30, 1930.
Link:	https://stateclimate.org/pdfs/journal-articles/2020_5-Foster.pdf