{"id":301,"date":"2026-10-17T16:45:39","date_gmt":"2026-10-17T23:45:39","guid":{"rendered":"https:\/\/natkale.com\/?p=301"},"modified":"2026-10-17T16:45:39","modified_gmt":"2026-10-17T23:45:39","slug":"environmental-planning-and-ai","status":"publish","type":"post","link":"https:\/\/natkale.com\/?p=301","title":{"rendered":"Environmental Planning and AI"},"content":{"rendered":"\n

For the environmental planning profession to make positive use of AI, we need a lot more data. I think we should be breaking our ways of doing business to collect orders of magnitude more eDNA<\/strong>, LiDAR<\/strong>, and groundwater<\/strong> data.<\/p>\n\n\n\n

Most people know AI from chatbots, but the technology behind them (trained neural nets) can do a lot more. Two examples are hurricane prediction, where AI is already a lot faster and more accurate than traditional physical models, and protein folding, where a famously difficult class of problems were solved basically overnight.<\/p>\n\n\n\n

What those examples, and chatbots themselves, have in common is lots and lots of data.<\/p>\n\n\n\n

The AI revolution started when scientists took this existing tool of neural nets, and started throwing orders of magnitude more computing power (\u201ccompute\u201d) and training data at it. AI\u2019s power seems to scale with data and compute, way more than anyone expected. I\u2019m sure there\u2019s a limit, but it doesn\u2019t seem like we\u2019ve reached it yet.<\/p>\n\n\n\n

In environmental planning we work at the interface of biological and social systems. I\u2019m pretty skeptical about AI\u2019s value in work on social systems for a whole bunch of reasons, but it\u2019s already proven to be pretty effective at biology.<\/p>\n\n\n\n

The compute is already there (or being built). What we\u2019re missing is data.<\/p>\n\n\n\n

It seems like there is a lot of data already out there about the systems we work with. In my world, we have data about rainfall, stream discharge, juvenile salmon, adult salmon returns, stream bugs, land cover \u2013 all kinds of data.<\/p>\n\n\n\n

The problem is scale. Most of our data sources are orders of magnitude too small.<\/p>\n\n\n\n

Take streamflow data. If you make some very generous estimates about the USGS stream gaging network (about 12,000 sites across the US, generating streamflow data every 15 minutes for the past 50 years), you get about 21 billion measurements. That\u2019s the low end for training really capable neural networks, which often rely on trillions of \u201ctokens\u201d for training. And streamflow is just one of the datasets we rely on. There is much, much less data available about things like salmon or other species of interest.<\/p>\n\n\n\n

What we as environmental planners want to do is estimate the impact of human actions on the natural environment. To do that, I think we need to start collecting a lot more data, starting with three areas.<\/p>\n\n\n\n

Environmental DNA<\/strong> (eDNA) sampling, specifically metabarcoding, should be dramatically more widespread than it is. A geographically and temporally rich eDNA dataset would directly measure what we want to manage \u2013 biological diversity. The main barrier is cost, with samples in the hundreds or tens of dollars, maybe dropping below the ten dollar mark per sample for a really big sample set. Driving that down to cents per sample, probably by scaling up and standardizing materials and methods, would be crucial.<\/p>\n\n\n\n

LiDAR<\/strong> datasets tend to be geographically rich, but temporally poor \u2013 they\u2019re not collected very frequently. But the cost of LiDAR has dropped dramatically, and could drop even further with drone usage. LiDAR is invaluable for characterizing the physical structure of the environment, an essential compliment to tracking biology directly. LiDAR collected every day, instead of every few years, would be much more sensitive to seasonal changes.<\/p>\n\n\n\n

Groundwater monitoring<\/strong> via indirect methods, such as ground penetrating radar, would fill in an essential gap in hydrology. Complemented by existing data collection, especially radar that estimates precipitation and the USGS gaging data I already mentioned, continuous or nearly-continuous groundwater monitoring would complete the monitoring of the hydrologic cycle.<\/p>\n\n\n\n

None of this would be easy. In some cases, especially with eDNA, it would mean inventing new technologies. It would definitely mean new business models and new ways of doing business.<\/p>\n\n\n\n

The truth is, we\u2019re already behind. AI is proving to be an incredibly powerful tool. But the direction<\/em> of AI development is entirely dictated by available data. AI researchers are building new models only where they have the data to do so.<\/p>\n\n\n\n

If environmental planning is going to harness AI as a real, powerful, and beneficial tool, not just a confident and unreliable intern, we need to steer where it goes. And the way to do that is by creating the data it needs.<\/p>\n","protected":false},"excerpt":{"rendered":"

For the environmental planning profession to make positive use of AI, we need a lot more data. I think we should be breaking our ways of doing business to collect orders of magnitude more eDNA, LiDAR, and groundwater data. Most people know AI from chatbots, but the technology behind them (trained neural nets) can do […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-301","post","type-post","status-publish","format-standard","hentry","category-uncategorized"],"_links":{"self":[{"href":"https:\/\/natkale.com\/index.php?rest_route=\/wp\/v2\/posts\/301","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/natkale.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/natkale.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/natkale.com\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/natkale.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=301"}],"version-history":[{"count":1,"href":"https:\/\/natkale.com\/index.php?rest_route=\/wp\/v2\/posts\/301\/revisions"}],"predecessor-version":[{"id":302,"href":"https:\/\/natkale.com\/index.php?rest_route=\/wp\/v2\/posts\/301\/revisions\/302"}],"wp:attachment":[{"href":"https:\/\/natkale.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=301"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/natkale.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=301"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/natkale.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=301"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}